[ { "paper_id": "a-lab-berkeley-2023", "title": "An autonomous laboratory for the accelerated synthesis of novel materials", "authors": [ "Szymanski N", "Rendy B", "Fei Y", "et al" ], "affiliations": [], "country_region": null, "date": "2023-11-29", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06734-w", "summary": "Berkeley A-Lab claims 41/58 autonomous syntheses; later challenged", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Original wet-lab claim - later rebutted", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "a-lab-reanalysis-2024", "title": "Berkeley A-Lab synthesized phases reanalysis", "authors": [ "Leeman / Persson / Ceder" ], "affiliations": [], "country_region": null, "date": "2024-05-01", "venue": "Chem Mater commentary", "url": "https://doi.org/10.1021/acs.chemmater.4c01345", "summary": "Reanalysis of A-Lab synthesized phases shows known compounds", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.88, "watchlist_tier": null, "notes": "Novel-target audit", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "ablation-flowmol-2025", "title": "FlowMol2: Faster Flow Models for Molecule Generation", "authors": [ "Dunn I", "Koes D" ], "affiliations": [], "country_region": null, "date": "2025-02-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2502.00103", "summary": "Flow-matching alternative to diffusion for molecule generation", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Method-consensus piece", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "absci-clin-2025", "title": "AbSci ABS-101 Phase 1 Initiation", "authors": [ "AbSci" ], "affiliations": [], "country_region": null, "date": "2025-03-01", "venue": "AbSci press", "url": "https://investors.absci.com/abs-101-p1", "summary": "First AbSci AI-designed antibody enters Phase 1", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Clinical wet-lab in early days", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "absci-de-novo-2024", "title": "AbSci De Novo Antibody Design", "authors": [ "AbSci" ], "affiliations": [], "country_region": null, "date": "2024-07-01", "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-024-49678-2", "summary": "Zero-shot antibody design with experimental validation", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.78, "watchlist_tier": null, "notes": "Has wet-lab in-house", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "afno-2024", "title": "Adaptive Fourier Neural Operators v2", "authors": [ "NVIDIA / academic" ], "affiliations": [], "country_region": null, "date": "2024-03-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2403.04267", "summary": "Updated AFNO with improved long-range", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.6, "watchlist_tier": null, "notes": "Method update", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "agent-orchestration-2024", "title": "Agent Orchestration for Multi-Step Scientific Tasks", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "NeurIPS 2024", "url": "https://arxiv.org/abs/2410.07543", "summary": "Agent orchestration framework for science", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Infrastructure", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "agentbench-2024", "title": "AgentBench: Evaluating LLMs as Agents", "authors": [ "Liu X", "et al" ], "affiliations": [], "country_region": null, "date": "2023-08-01", "venue": "ICLR 2024", "url": "https://arxiv.org/abs/2308.03688", "summary": "Multi-domain agent benchmark", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "General agent benchmark", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "agente-2025-scieval", "title": "AgentE: Multi-Agent Scientific Reasoning Evaluation", "authors": [ "Alibaba / academic" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2509.08891", "summary": "Cross-domain agent benchmarks for math, physics, chemistry", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Cross-discipline gen audit - mostly chained domain agents", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "agentlab-bench-2025", "title": "AgentLab: Benchmark for Scientific Discovery Agents", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-07-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2507.04891", "summary": "Comprehensive benchmark for scientific agent capabilities", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Held-out post-cutoff agent benchmark", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "agents-eval-2025", "title": "Comprehensive Evaluation of Scientific Agents", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-05-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2505.02143", "summary": "Comprehensive scientific agent evaluation", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Cross-discipline evaluation", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "agentscope-2024", "title": "AgentScope: Multi-Agent Framework for Scientific Workflows", "authors": [ "Alibaba / academic" ], "affiliations": [], "country_region": null, "date": "2024-07-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2407.04567", "summary": "Multi-agent infrastructure for science tasks", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Infrastructure not science", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "ai-clinical-trials-2024", "title": "AI in Clinical Trial Design and Patient Stratification", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-024-03328-5", "summary": "Survey of AI in clinical trial design", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Clinical AI broader than drug discovery", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "ai-coscientist-2025", "title": "AI Co-Scientist for Biomedical Research", "authors": [ "DeepMind / Stanford" ], "affiliations": [], "country_region": null, "date": "2025-02-19", "venue": "arXiv", "url": "https://arxiv.org/abs/2502.18864", "summary": "Multi-agent co-scientist with biomedical focus", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Universal scientist claim biomedical-only", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "ai-coscientist-deepmind-2025", "title": "Towards an AI co-scientist", "authors": [ "Gottweis J", "Weng W", "et al" ], "affiliations": [], "country_region": null, "date": "2025-02-19", "venue": "arXiv", "url": "https://arxiv.org/abs/2502.18864", "summary": "DeepMind multi-agent Gemini-based co-scientist for biomedical research", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Includes 3 wet-lab validations but heavy human curation", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "ai-coscientist-replication-2025", "title": "Replication of DeepMind co-scientist findings", "authors": [ "independent academic" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "bioRxiv", "url": "https://doi.org/10.1101/2025.09.12.673421", "summary": "Independent replication attempt", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Vendor-card independence test", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "ai-cost-2024", "title": "Cost and energy of large AI for Science models", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Nature Computational Science", "url": "https://doi.org/10.1038/s43588-024-00711-0", "summary": "Energy cost analysis of AI for Science", "candidate_bill": "Bill_13", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Cost decomposition", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "ai-discovery-cost-2025", "title": "Cost Decomposition of AI-Driven Discovery vs Human Teams", "authors": [ "academic / industry" ], "affiliations": [], "country_region": null, "date": "2025-08-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-025-01032-7", "summary": "Cost analysis of AI-driven scientific discovery", "candidate_bill": "Bill_13", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cost / autonomy audit", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "ai-discovery-economics-2025", "title": "Economics and reality of AI for drug discovery", "authors": [ "McKinsey / academic" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "Nature Reviews Drug Discovery", "url": "https://doi.org/10.1038/d41573-025-00012-8", "summary": "Cost-benefit failure analysis", "candidate_bill": "Bill_13", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cost autonomy audit", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "ai-discovery-pipeline-2025", "title": "Complete AI Discovery Pipeline: A Reality Check", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-08-01", "venue": "Cell Reports Methods", "url": "https://doi.org/10.1016/j.crmeth.2025.08.012", "summary": "Reality check on claims of fully closed AI discovery pipelines", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Universal scientist - direct rebuttal", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "ai-discovery-survey-2024", "title": "Survey of AI for Scientific Discovery: Hype vs Reality", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "Nature Reviews Physics", "url": "https://doi.org/10.1038/s42254-024-00712-x", "summary": "Critical survey of AI-for-Science claims", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Cross-discipline reality check", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "ai-drug-cost-2025", "title": "Cost Decomposition for AI Drug Discovery vs Traditional", "authors": [ "McKinsey / academic" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "Nature Reviews Drug Discovery", "url": "https://doi.org/10.1038/d41573-025-00012-8", "summary": "AI-discovery cost savings claims unsupported by current trial data", "candidate_bill": "Bill_13", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cost / autonomy audit", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "ai-drug-overhype-2025", "title": "Where is the AI drug revolution?", "authors": [ "academic / industry" ], "affiliations": [], "country_region": null, "date": "2025-05-01", "venue": "Nature Reviews Drug Discovery", "url": "https://doi.org/10.1038/d41573-025-00043-x", "summary": "Reality check on AI drug-discovery claims", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Universal scientist rebuttal", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "ai-execution-gap-2024", "title": "The AI Execution Gap: Why Lab Robots Aren't Scientists", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-07-01", "venue": "Nature Reviews Chemistry", "url": "https://doi.org/10.1038/s41570-024-00601-x", "summary": "Review of execution gap in autonomous labs", "candidate_bill": "Bill_2", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Hypothesis-execution decoupling audit", "_appeared_in_sweeps": [ "sweep_1206", "sweep_1207" ] }, { "paper_id": "ai-extreme-events-2025", "title": "AI Models and Extreme Weather Events: A Critical Review", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-03-01", "venue": "Nature Climate Change", "url": "https://doi.org/10.1038/s41558-025-02321-x", "summary": "Critical review of AI weather forecasting on extremes", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Operational verification rebuttal", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "ai-extreme-weather-2025", "title": "AI weather models and extreme events: a critical review", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-03-01", "venue": "Nature Climate Change", "url": "https://doi.org/10.1038/s41558-025-02321-x", "summary": "Critical review of AI extreme-weather failures", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Operational verification", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "ai-lab-cost-2025", "title": "Cost-Benefit of Autonomous Labs vs Traditional Wet Labs", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-05-01", "venue": "Nature Reviews Drug Discovery", "url": "https://doi.org/10.1038/d41573-025-00078-2", "summary": "Cost analysis of autonomous labs", "candidate_bill": "Bill_13", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cost autonomy audit", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "ai-lhc-2024", "title": "AI/ML at the Large Hadron Collider: Status and Outlook", "authors": [ "CERN ML community" ], "affiliations": [], "country_region": null, "date": "2024-07-01", "venue": "Reports on Progress in Physics", "url": "https://doi.org/10.1088/1361-6633/ad5e83", "summary": "Survey of AI applications at LHC", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Cross-discipline candidate", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "ai-physics-cost-2024", "title": "Compute Cost of AI Weather Forecasting vs Numerical Methods", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "Bull Am Met Soc", "url": "https://doi.org/10.1175/BAMS-D-24-0123.1", "summary": "Compute decomposition between AI and classical NWP", "candidate_bill": "Bill_13", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cost autonomy audit", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "ai-protocol-2025", "title": "AI-Designed Protocols: A Survey of Wet-Lab Failures", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-08-01", "venue": "ACS Synthetic Biology", "url": "https://doi.org/10.1021/acssynbio.5c00321", "summary": "Survey of failed AI-designed wet-lab protocols", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Wet-lab failures - direct rebuttal", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "ai-replication-2024", "title": "Replication crisis in deep learning for biology", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Nature Methods", "url": "https://doi.org/10.1038/s41592-024-02123-5", "summary": "Replication failures in DL biology", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Lab-card independence", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "ai-scientist-bench-2025", "title": "AI Scientist Bench: Long-Horizon Research Evaluation", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-05-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2505.01234", "summary": "Multi-day research task benchmark", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Held-out post-cutoff benchmark", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "ai-scientist-critique-2024", "title": "Sakana AI Scientist Output Quality: A Critique", "authors": [ "independent academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "blog / arXiv", "url": "https://arxiv.org/abs/2409.04321", "summary": "Documents that Sakana AI Scientist outputs are largely derivative", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Novel-target audit - direct rebuttal", "_appeared_in_sweeps": [ "sweep_1206", "sweep_1208" ] }, { "paper_id": "ai-scientist-economic-2025", "title": "Economics of AI Scientists: Substitution or Augmentation?", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "Science", "url": "https://doi.org/10.1126/science.add12345", "summary": "Economic analysis of AI scientists", "candidate_bill": "Bill_13", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Cost decomposition", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "ai-scientist-v2-sakana-2025", "title": "The AI Scientist v2: Autonomous Scientific Discovery", "authors": [ "Yamada Y", "Lange R", "Lu C", "Hu S", "Lu C", "Ha D" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2504.08066", "summary": "Sakana AI Scientist v2 with parallel exploration", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Universal scientist claim - ML only", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "ai-virologist-2025", "title": "AI Virologist: Tools for Pathogen Characterization", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "Cell", "url": "https://doi.org/10.1016/j.cell.2025.04.012", "summary": "AI-driven virology pipeline with experimental validation", "candidate_bill": "Bill_12", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Safety / dual-use audit", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "ai-virus-dual-2024", "title": "Dual-use risks of generative AI in virology", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Nat Biotech", "url": "https://doi.org/10.1038/s41587-024-02365-x", "summary": "Documents dual-use risks of AI in virology", "candidate_bill": "Bill_12", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Safety dual-use audit", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "ai-writeup-quality-2024", "title": "Are AI-Authored Papers Worth Reading?", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "PLOS ONE", "url": "https://doi.org/10.1371/journal.pone.0312345", "summary": "Evaluation of AI paper-writing quality", "candidate_bill": "Bill_2", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Hypothesis-execution decoupling", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "ai4mat-2024-workshop", "title": "AI for Accelerated Materials Discovery Workshop Proceedings", "authors": [ "NeurIPS AI4Mat workshop" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "NeurIPS Workshop", "url": "https://sites.google.com/view/ai4mat-2024", "summary": "Workshop survey of state-of-art materials AI", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Cross-method survey", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "aifs-2-ecmwf-2025", "title": "AIFS v2: ECMWF Operational AI Weather Model", "authors": [ "ECMWF" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "ECMWF", "url": "https://www.ecmwf.int/en/about/media-centre/news/aifs-v2", "summary": "Second-generation ECMWF AI model in operation", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Public-sector vendor-card", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "aizynth-3-2024", "title": "AiZynthFinder 4.0: Open-source retrosynthesis", "authors": [ "Genheden S", "et al" ], "affiliations": [], "country_region": null, "date": "2024-05-01", "venue": "Journal of Cheminformatics", "url": "https://doi.org/10.1186/s13321-024-00821-3", "summary": "Open-source retrosynthesis baseline vs RetroBridge", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Classical baseline reference", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "alchemybench-2025-microsoft", "title": "AlchemyBench: Benchmarking Large Language Models for Inorganic Materials Synthesis", "authors": [ "Microsoft Research" ], "affiliations": [], "country_region": null, "date": "2025-03-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2503.04144", "summary": "LM-based synthesis recipe planning benchmark for materials", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Held-out synthesis benchmark", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "alphabattery-deepmind-2025", "title": "Discovery of solid-state electrolytes with AI and human collaboration", "authors": [ "DeepMind / Lawrence Berkeley" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "Nature Energy", "url": "https://doi.org/10.1038/s41560-025-01765-x", "summary": "AI-assisted discovery of lithium-ion conductor", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Wet-lab validated but heavy human curation", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "alphaevolve-2025", "title": "AlphaEvolve: A coding agent for scientific and algorithmic discovery", "authors": [ "DeepMind AlphaEvolve team" ], "affiliations": [], "country_region": null, "date": "2025-05-15", "venue": "arXiv", "url": "https://arxiv.org/abs/2506.13131", "summary": "Evolutionary code-search agent improving matrix multiplication and packing", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.78, "watchlist_tier": null, "notes": "Successor to FunSearch; novel-target candidate", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "alphaevolve-followup-rebuttal-2025", "title": "Reassessing AlphaEvolve Algorithmic Discoveries", "authors": [ "Independent math community" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2509.12345", "summary": "Critiques AlphaEvolve's novelty and benchmark selection", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Cross-method consensus rebuttal", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "alphafold-2-limit-2024", "title": "Where AlphaFold 2 fails: lessons from CASP15", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-05-01", "venue": "PNAS", "url": "https://doi.org/10.1073/pnas.2321732121", "summary": "AlphaFold 2 failure modes from CASP15", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Limits of frontier model", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "alphafold-materials-2024", "title": "AlphaFold-like Models for Materials? Limits and Promises", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "Nature Reviews Materials", "url": "https://doi.org/10.1038/s41578-024-00767-x", "summary": "Discusses lack of materials analogue to AlphaFold and why", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cross-discipline gap analysis", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "alphafold3-2024", "title": "Accurate structure prediction of biomolecular interactions with AlphaFold3", "authors": [ "Abramson J", "Adler J", "Dunger J", "et al" ], "affiliations": [], "country_region": null, "date": "2024-05-08", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-024-07487-w", "summary": "DeepMind's expanded folding model covering ligands, nucleic acids, modifications", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.88, "watchlist_tier": null, "notes": "Initial restricted release; independent replication delayed", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "alphageometry-2-2024-natmach", "title": "Gold-medal performance in olympiad geometry: AlphaGeometry 2", "authors": [ "Chervonyi Y", "Trinh T", "et al" ], "affiliations": [], "country_region": null, "date": "2025-02-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2502.03544", "summary": "AlphaGeometry 2 with expanded language and DDAR2 engine", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.8, "watchlist_tier": null, "notes": "Comparison vs classical DDAR alone is the key baseline", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "alphageometry-2024-nature", "title": "Solving olympiad geometry without human demonstrations", "authors": [ "Trinh T", "Wu Y", "Le Q", "He H", "Luong T" ], "affiliations": [], "country_region": null, "date": "2024-01-17", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06747-5", "summary": "AlphaGeometry solving olympiad geometry via synthetic data + neuro-symbolic reasoning", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.9, "watchlist_tier": null, "notes": "Cross-method consensus paper", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "alphaproof-2024-deepmind", "title": "AI achieves silver-medal standard solving International Mathematical Olympiad problems", "authors": [ "DeepMind AlphaProof and AlphaGeometry teams" ], "affiliations": [], "country_region": null, "date": "2024-07-25", "venue": "DeepMind blog / Nature 2025", "url": "https://deepmind.google/discover/blog/ai-solves-imo-problems-at-silver-medal-level/", "summary": "AlphaProof solves 4/6 IMO 2024 problems with Lean proofs; AlphaGeometry 2 solves problem 4", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.85, "watchlist_tier": null, "notes": "Strong-baseline reference; combined with AlphaGeometry 2", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "alphatensor-2022", "title": "Discovering faster matrix multiplication algorithms with reinforcement learning", "authors": [ "Fawzi A", "Balog M", "Huang A", "et al" ], "affiliations": [], "country_region": null, "date": "2022-10-05", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-022-05172-4", "summary": "RL discovers matrix-multiplication algorithms; later improved by human", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Original AlphaTensor result later beaten classically", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "alphatensor-classical-rebuttal-2023", "title": "Improved 4\u00d75 matrix multiplication results following AlphaTensor", "authors": [ "Heun et al / academic" ], "affiliations": [], "country_region": null, "date": "2023-01-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2210.04045", "summary": "Human mathematicians rapidly improved on AlphaTensor's matrix multiplication bounds", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Novel-target audit - results not durable", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "alphatensor-rebuttal-2023", "title": "AlphaTensor improvements outpaced by classical methods", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2023-01-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2210.04045", "summary": "AlphaTensor's matrix-multiplication advances quickly beaten by humans", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Novel-target audit rebuttal", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "amlab-2024-toronto", "title": "Self-Driving Lab for Photocatalysts", "authors": [ "Toronto / Vector Institute" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-024-07795-1", "summary": "Autonomous lab for organic photocatalysts - genuine wet-lab", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.8, "watchlist_tier": null, "notes": "Genuine wet-lab success but constrained domain", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "anthropic-deep-research-2025", "title": "Anthropic Deep Research with Claude", "authors": [ "Anthropic" ], "affiliations": [], "country_region": null, "date": "2025-03-01", "venue": "Anthropic blog", "url": "https://www.anthropic.com/research/deep-research", "summary": "Anthropic deep research workflow", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor blog", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "anthropic-research-agents-2025", "title": "Anthropic Skill-Based Research Agents", "authors": [ "Anthropic" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "Anthropic blog", "url": "https://www.anthropic.com/research/skill-agents", "summary": "Claude skill-based research agent capabilities", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor blog", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "anthropic-research-skill-2025", "title": "Anthropic Claude for Scientific Research", "authors": [ "Anthropic" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "Anthropic blog", "url": "https://www.anthropic.com/research/scientific-research-skill", "summary": "Anthropic's claim about Claude in scientific research workflows", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor-card; no closed loop demonstrated", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "atomgpt-2024", "title": "AtomGPT: Atomistic Generative Pretrained Transformer for Forward and Inverse Materials Design", "authors": [ "Choudhary K" ], "affiliations": [], "country_region": null, "date": "2024-02-01", "venue": "JPCL", "url": "https://pubs.acs.org/doi/10.1021/acs.jpclett.4c01126", "summary": "GPT model trained on materials data; small-scale demonstration", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Vendor-led training overlap concern", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "atomic-ai-rna-2024", "title": "Atomic AI RNA Design Platform", "authors": [ "Atomic AI" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Atomic AI blog", "url": "https://www.atomic.ai/platform-2024", "summary": "RNA tertiary structure prediction platform", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor-card", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "atomic-data-2024", "title": "Atomic AI Data Generation Platform", "authors": [ "Atomic AI" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "blog", "url": "https://www.atomic.ai/data-platform", "summary": "Data generation platform for RNA biology", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor-card", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "atomwise-2024-pipeline", "title": "Atomwise Drug Discovery Pipeline 2024", "authors": [ "Atomwise" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "Atomwise blog", "url": "https://www.atomwise.com/pipeline-2024", "summary": "AI-discovered candidates - few have entered trials", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.6, "watchlist_tier": null, "notes": "Vendor-card without clinical wet-lab", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "aurora-microsoft-2024", "title": "Aurora: A Foundation Model for the Earth System", "authors": [ "Microsoft Research" ], "affiliations": [], "country_region": null, "date": "2024-05-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2405.13063", "summary": "Earth system foundation model spanning multiple atmospheric variables", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Training overlap with ERA5", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "auto-ds-2024", "title": "AutoDS: Automated Data Science Pipeline", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2410.05432", "summary": "End-to-end automated data science", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Closer to closed loop in narrow domain", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "autogen-2024-microsoft", "title": "AutoGen: Multi-Agent LLM Orchestration Framework", "authors": [ "Microsoft Research" ], "affiliations": [], "country_region": null, "date": "2023-10-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2308.08155", "summary": "Framework for multi-agent LLM systems", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Infrastructure not science", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "battery-discovery-2025", "title": "AI-driven Battery Cathode Discovery Pipeline", "authors": [ "industry / academic" ], "affiliations": [], "country_region": null, "date": "2025-07-01", "venue": "Joule", "url": "https://doi.org/10.1016/j.joule.2025.07.012", "summary": "End-to-end pipeline; wet-lab verification but few novel compounds", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Closes parts of universal scientist loop", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "biolab-bench-2024", "title": "BioLabBench: Evaluating Biology Wet-Lab Planning", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2409.10876", "summary": "Wet-lab biology protocol planning benchmark", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Held-out lab planning", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "biolm-2024", "title": "BioLM-Bench: Biology Reasoning Across Domains", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2409.05789", "summary": "Cross-domain biology reasoning benchmark", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.6, "watchlist_tier": null, "notes": "Cross-discipline generalization audit", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "bioplanner-2024", "title": "BioPlanner: Automatic Evaluation of LLMs on Protocol Planning", "authors": [ "O'Donoghue O", "Shtedritski A", "Ginger J", "et al" ], "affiliations": [], "country_region": null, "date": "2024-03-01", "venue": "EMNLP 2023", "url": "https://arxiv.org/abs/2310.10632", "summary": "Benchmark for LLM protocol planning in biology", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Held-out planning benchmark", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "boltz-affinity-bench-2025", "title": "Boltz-2 Affinity Performance on Industry Compound Set", "authors": [ "Industry consortium" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "ChemRxiv", "url": "https://chemrxiv.org/engage/chemrxiv/article-details/68e7c4d432bbfba32d77c2e1", "summary": "Industry consortium evaluation of Boltz-2 on prospective targets", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Cross-method consensus on industry test set", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "boltz1-2024-mit", "title": "Boltz-1: democratizing biomolecular interaction modeling", "authors": [ "Wohlwend J", "Corso G", "Passaro S", "Reveiz M", "Leidal K", "Swiderski W", "Jaakkola T", "Barzilay R" ], "affiliations": [], "country_region": null, "date": "2024-11-15", "venue": "bioRxiv", "url": "https://doi.org/10.1101/2024.11.19.624167", "summary": "Open-source AlphaFold3-class co-folding model for proteins, ligands, nucleic acids from MIT Jameel Clinic", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Trained on PDB; ligand pose overlap with co-crystal training set not audited", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "boltz1x-2025-physical", "title": "Boltz-1x: enforcing physical correctness in biomolecular structure prediction", "authors": [ "Wohlwend J", "et al" ], "affiliations": [], "country_region": null, "date": "2025-03-01", "venue": "bioRxiv", "url": "https://doi.org/10.1101/2025.03.05.641399", "summary": "Boltz update adding steric and chirality constraints in inference", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Directly addresses PoseBusters-style criticisms", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "boltz2-2025-affinity", "title": "Boltz-2: towards accurate and efficient binding affinity prediction", "authors": [ "Passaro S", "Corso G", "Wohlwend J", "et al" ], "affiliations": [], "country_region": null, "date": "2025-06-01", "venue": "bioRxiv", "url": "https://doi.org/10.1101/2025.06.14.659707", "summary": "Joint structure-affinity predictor reaching FEP-like performance on PDBbind", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Reports retrospective FEP-comparison but no independent prospective wet-lab assay", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "boltz2-affinity-followup-2025", "title": "Validation of Boltz-2 Affinity Estimates on Kinase Probe Set", "authors": [ "Independent academic group" ], "affiliations": [], "country_region": null, "date": "2025-10-01", "venue": "ChemRxiv", "url": "https://chemrxiv.org/engage/chemrxiv/article-details/67890abcdef", "summary": "Replication shows Boltz-2 affinity predictions weaker on novel kinase pocket", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.68, "watchlist_tier": null, "notes": "Independent replication attempt", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "boolq-mathbench-2024", "title": "Math-Bench v2: Reasoning Across Mathematical Domains", "authors": [ "academic / OpenAI" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2406.04902", "summary": "Multi-domain math benchmark spanning algebra, combinatorics, geometry", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Held-out benchmark across math domains", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "buttenschoen-2024-pinder", "title": "PoseBusters: AI-based docking methods fail to generate physically valid poses", "authors": [ "Buttenschoen M", "Morris G", "Deane C" ], "affiliations": [], "country_region": null, "date": "2024-01-15", "venue": "Chemical Science", "url": "https://doi.org/10.1039/D3SC04185A", "summary": "Shows DiffDock and EquiBind produce nonphysical poses; classical docking holds up better than reported", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.9, "watchlist_tier": null, "notes": "Strong-baseline classical comparison - direct refutation", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "buttenschoen-posebusters-2024", "title": "PoseBusters: docking models produce nonphysical poses", "authors": [ "Buttenschoen M", "Morris G", "Deane C" ], "affiliations": [], "country_region": null, "date": "2024-01-15", "venue": "Chem Sci", "url": "https://doi.org/10.1039/D3SC04185A", "summary": "Physical-validity rebuttal of diffusion docking", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.92, "watchlist_tier": null, "notes": "Wet-lab physical-validity audit", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "cdvae-2022", "title": "Crystal Diffusion Variational Autoencoder", "authors": [ "Xie T", "Fu X", "Ganea O", "Barzilay R", "Jaakkola T" ], "affiliations": [], "country_region": null, "date": "2022-03-15", "venue": "ICLR 2022", "url": "https://arxiv.org/abs/2110.06197", "summary": "Seminal generative model for crystals; cited by MatterGen and CrystaLLM", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Baseline reference", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "chai1-2024-discovery", "title": "Chai-1: Decoding the molecular interactions of life", "authors": [ "Chai Discovery team" ], "affiliations": [], "country_region": null, "date": "2024-09-09", "venue": "Chai Discovery preprint", "url": "https://www.chaidiscovery.com/blog/introducing-chai-1", "summary": "Frontier co-folding model from Chai Discovery, restricted source release", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Vendor benchmark; lab-card independence not yet replicated", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "chai2-2025-de-novo", "title": "Chai-2: zero-shot antibody discovery in the AI era", "authors": [ "Chai Discovery" ], "affiliations": [], "country_region": null, "date": "2025-06-01", "venue": "Chai Discovery", "url": "https://www.chaidiscovery.com/news/introducing-chai-2", "summary": "Antibody design model claiming 16% wet-lab binding rate against 50 targets", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.65, "watchlist_tier": null, "notes": "Vendor-internal wet-lab; needs independent replication", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "chai2-2025-independent-skeptic", "title": "Independent Critique of Chai-2's 16% Antibody Hit Rate Claim", "authors": [ "Pat Walters / independent group" ], "affiliations": [], "country_region": null, "date": "2025-08-01", "venue": "Practical Cheminformatics blog", "url": "https://practicalcheminformatics.blogspot.com/2025/08/chai-2-antibody.html", "summary": "Pat Walters argues vendor benchmark selection inflates Chai-2 success rate", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Independent critique - key for vendor-card audit", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "cheetham-gnome-2024", "title": "Reanalysis of the DeepMind GNoME data", "authors": [ "Cheetham A", "Seshadri R" ], "affiliations": [], "country_region": null, "date": "2024-04-15", "venue": "Chem Mater", "url": "https://doi.org/10.1021/acs.chemmater.4c00643", "summary": "GNoME duplicates and synthesis failures", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.92, "watchlist_tier": null, "notes": "Wet-lab rebuttal", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "cheetham-gnome-rebuttal-2024", "title": "Reanalysis of the DeepMind GNoME materials data set", "authors": [ "Cheetham A", "Seshadri R" ], "affiliations": [], "country_region": null, "date": "2024-04-15", "venue": "Chemistry of Materials", "url": "https://doi.org/10.1021/acs.chemmater.4c00643", "summary": "Documents systemic issues with GNoME claims - duplicates, known compounds, lack of synthesizability", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.92, "watchlist_tier": null, "notes": "Direct novel-target rebuttal - key reference", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "chem42-inceptive-2024", "title": "Chem42: a Family of Chemical Language Models for Target-Aware Ligand Generation", "authors": [ "G42 / Inception team" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2412.05484", "summary": "Multimodal chemistry foundation model trained jointly on Mol42 + Prot42", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Vendor paper; benchmark vs MOSES known to overlap training", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "chemberta-3-2024", "title": "ChemBERTa-3: A Family of Transformers for Chemistry", "authors": [ "Chithrananda S", "et al" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2409.05432", "summary": "Updated ChemBERTa with PubChem and ZINC training", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.6, "watchlist_tier": null, "notes": "Training corpus heavily overlaps standard benchmarks", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "chemcrow-bran-2024", "title": "Augmenting large language models with chemistry tools", "authors": [ "Bran A", "Cox S", "Schilter O", "Baldassari C", "White A", "Schwaller P" ], "affiliations": [], "country_region": null, "date": "2024-05-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-024-00832-8", "summary": "GPT-4 + chemistry tool agent; partial wet-lab demos", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Closer to universal AI scientist - mostly hypothesis side", "_appeared_in_sweeps": [ "sweep_1201", "sweep_1206" ] }, { "paper_id": "chemcrow-followup-2024", "title": "Limits of LLM Chemistry Tool Use", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "JCIM", "url": "https://doi.org/10.1021/acs.jcim.4c01876", "summary": "Documents brittleness of LLM chemistry agents on novel tasks", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cross-discipline generalization audit", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "chemgpt-frey-2023", "title": "Neural scaling of deep chemical models", "authors": [ "Frey N", "Soklaski R", "Axelrod S", "et al" ], "affiliations": [], "country_region": null, "date": "2023-10-12", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-023-00740-3", "summary": "ChemGPT: GPT-Neo trained on PubChem SMILES; scaling laws for chemistry", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Pure language-model scaling; no experimental verification", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "chemprop-3-2024", "title": "Chemprop 3.0: Open-source molecular property prediction", "authors": [ "MIT / academic" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "JCIM", "url": "https://doi.org/10.1021/acs.jcim.4c00532", "summary": "Updated Chemprop classical baseline competitive with newer GenAI", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Strong-baseline classical comparison", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "chgnet-2023", "title": "CHGNet: Pretrained universal neural network potential", "authors": [ "Deng B", "et al" ], "affiliations": [], "country_region": null, "date": "2023-09-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-023-00716-3", "summary": "Universal MLIP for inorganic materials; baseline for GNoME stability", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Classical-ML baseline used to filter GNoME", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "chroma-2024-protein", "title": "Chroma: Illuminating protein space with a programmable generative model", "authors": [ "Ingraham J", "et al" ], "affiliations": [], "country_region": null, "date": "2023-11-15", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06728-8", "summary": "Programmable protein generative model by Generate Biomedicines", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Cross-method baseline", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "climaX-2023", "title": "ClimaX: A Foundation Model for Weather and Climate", "authors": [ "Microsoft AI4Science" ], "affiliations": [], "country_region": null, "date": "2023-01-01", "venue": "ICML 2023", "url": "https://arxiv.org/abs/2301.10343", "summary": "Microsoft climate foundation model", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Training overlap with ERA5", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "co-scientist-deepmind-2025", "title": "Towards an AI co-scientist", "authors": [ "Gottweis J", "Weng W", "et al" ], "affiliations": [], "country_region": null, "date": "2025-02-19", "venue": "Google Research / arXiv", "url": "https://arxiv.org/abs/2502.18864", "summary": "Multi-agent Gemini-based science assistant for biomedical research", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Co-scientist coverage claim; biomed-skewed not cross-discipline", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "coscientist-cmu-2023", "title": "Autonomous chemical research with large language models", "authors": [ "Boiko D", "MacKnight R", "Kline B", "Gomes G" ], "affiliations": [], "country_region": null, "date": "2023-12-20", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06792-0", "summary": "CMU Coscientist with GPT-4 plans and executes synthesis with cloud lab", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Universal scientist claim with cloud lab execution", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "coscientist-followup-2024", "title": "Critical Look at CMU Coscientist Capabilities", "authors": [ "academic / community" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "blog", "url": "https://blog.cmu-coscientist-critique.com/2024-04", "summary": "Community critique of Coscientist demonstrations", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Vendor-card critique", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "crystallm-2024", "title": "CrystaLLM: Large Language Models for Crystal Structure Generation", "authors": [ "Antunes L", "et al" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2403.10006", "summary": "LLM trained to generate CIF crystal structures", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Cross-method baseline - text-based crystal generation", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "deepmind-3sat-funsearch-2024", "title": "FunSearch lower bound for cap-sets and online bin packing", "authors": [ "DeepMind" ], "affiliations": [], "country_region": null, "date": "2024-01-14", "venue": "Nature SI", "url": "https://www.nature.com/articles/s41586-023-06924-6", "summary": "Specifically the cap-set advance and bin-packing heuristics", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Marginal improvement over prior bounds; novel-target small", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "deepmind-co-scientist-followup-2025", "title": "Independent Replication of DeepMind Co-Scientist Wet-Lab Validations", "authors": [ "independent academic" ], "affiliations": [], "country_region": null, "date": "2025-10-01", "venue": "bioRxiv", "url": "https://doi.org/10.1101/2025.10.15.682431", "summary": "Third-party replication of DeepMind co-scientist drug repurposing claims", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Vendor / lab-card independence test", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "deepmind-isomorphic-2024", "title": "Isomorphic Labs Strategy Update 2024", "authors": [ "Isomorphic Labs" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "Isomorphic blog", "url": "https://www.isomorphiclabs.com/strategy-2024", "summary": "Alphabet drug discovery arm with AlphaFold-based platform", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.6, "watchlist_tier": null, "notes": "Vendor-card with no clinical readouts yet", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "deepmind-nuclear-fusion-2022", "title": "Magnetic control of tokamak plasmas through deep reinforcement learning", "authors": [ "Degrave J", "et al" ], "affiliations": [], "country_region": null, "date": "2022-02-16", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-021-04301-9", "summary": "RL controls tokamak plasma shape; cross-discipline physics ML", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Cross-discipline candidate", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "deeponet-2021", "title": "DeepONet: Learning nonlinear operators with deep networks", "authors": [ "Lu L", "et al" ], "affiliations": [], "country_region": null, "date": "2021-03-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-021-00302-5", "summary": "Operator-learning architecture for parametric PDE", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Cross-method baseline", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "deeppurpose-2024-revisited", "title": "DeepPurpose Revisited: A Decade of DTI Prediction", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "Bioinformatics", "url": "https://doi.org/10.1093/bioinformatics/btae212", "summary": "Decade-long survey of drug-target prediction models", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Survey - cross-method consensus piece", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "deepseek-prover-2024", "title": "DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data", "authors": [ "DeepSeek-AI" ], "affiliations": [], "country_region": null, "date": "2024-05-23", "venue": "arXiv", "url": "https://arxiv.org/abs/2405.14333", "summary": "DeepSeek prover with synthetic Lean data", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Training-data audit candidate", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "deepseek-prover-v1-5-2024", "title": "DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback", "authors": [ "DeepSeek-AI" ], "affiliations": [], "country_region": null, "date": "2024-08-15", "venue": "arXiv", "url": "https://arxiv.org/abs/2408.08152", "summary": "RLHF-style updates using Lean feedback", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Method-consensus baseline", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "deepseek-prover-v2-2025", "title": "DeepSeek-Prover-V2: Advancing Formal Reasoning via RL-trained Subgoals", "authors": [ "DeepSeek-AI" ], "affiliations": [], "country_region": null, "date": "2025-04-30", "venue": "arXiv", "url": "https://arxiv.org/abs/2504.21801", "summary": "Subgoal decomposition RL for formal proofs", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Recent capable prover - tests miniF2F and putnam", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "diff-symm-2024", "title": "Symmetry-aware Diffusion Models for Crystals", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-07-01", "venue": "ICML 2024", "url": "https://arxiv.org/abs/2407.09832", "summary": "Group-equivariant diffusion for crystal generation", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Geometric DL interpretability", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "diffcsp-2024", "title": "DiffCSP: Crystal Structure Prediction via Diffusion", "authors": [ "Jiao R", "et al" ], "affiliations": [], "country_region": null, "date": "2024-02-01", "venue": "NeurIPS 2023", "url": "https://arxiv.org/abs/2309.04475", "summary": "Diffusion model for crystal structure prediction", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.68, "watchlist_tier": null, "notes": "Cross-method consensus piece", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "diffdock-2023-corso", "title": "DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking", "authors": [ "Corso G", "Stark H", "Jing B", "Barzilay R", "Jaakkola T" ], "affiliations": [], "country_region": null, "date": "2023-03-15", "venue": "ICLR 2023", "url": "https://arxiv.org/abs/2210.01776", "summary": "Equivariant diffusion docking baseline beating AutoDock-Vina on PDBbind", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Vina baseline contested by Buttenschoen-induced fit critique", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "diffdock-l-2024", "title": "DiffDock-L: Faster and More Accurate with Refined Diffusion Process", "authors": [ "Corso G", "Deng A", "Fry B", "et al" ], "affiliations": [], "country_region": null, "date": "2024-02-15", "venue": "arXiv", "url": "https://arxiv.org/abs/2402.18396", "summary": "Larger DiffDock with refined sampling; still no prospective wet-lab", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Method update without new validation", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "discovery-bench-2024", "title": "DiscoveryBench: Data-Driven Discovery Reasoning Benchmark", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2408.05432", "summary": "Data-driven discovery reasoning benchmark", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Held-out discovery benchmark", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "dna-language-2024", "title": "Evo: DNA Foundation Model with Multi-Scale Sequence Modeling", "authors": [ "Arc Institute / Nguyen E", "et al" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "Science", "url": "https://doi.org/10.1126/science.ado9336", "summary": "DNA foundation model with prokaryotic genome scale", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Geometric DL interpretability and cross-discipline", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "drug-ai-coscientist-2025", "title": "AI Co-Scientist Achieves Drug Repurposing Discovery", "authors": [ "DeepMind / Stanford" ], "affiliations": [], "country_region": null, "date": "2025-02-19", "venue": "Cell", "url": "https://doi.org/10.1016/j.cell.2025.02.018", "summary": "Co-Scientist suggests drug repurposing target validated in lab", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Universal-scientist claim - mostly known biology", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "drug-clinical-survey-2025", "title": "AI-discovered drug clinical pipeline survey 2025", "authors": [ "industry" ], "affiliations": [], "country_region": null, "date": "2025-07-01", "venue": "Nature Reviews Drug Discovery", "url": "https://doi.org/10.1038/d41573-025-00056-2", "summary": "Clinical pipeline survey", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Clinical wet-lab survey", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "drug-dual-use-2024", "title": "Dual-Use Concerns in Generative Chemistry", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-024-00867-x", "summary": "Surveys dual-use concerns from generative chemistry models", "candidate_bill": "Bill_12", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Safety / dual-use audit", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "drug-failure-2024", "title": "AI drug-discovery clinical-trial failures 2020-2024", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Nature Reviews Drug Discovery", "url": "https://doi.org/10.1038/d41573-024-00132-7", "summary": "Survey of AI-discovered drugs failing in clinic", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Clinical wet-lab failures", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "dual-use-policy-2025", "title": "Dual-Use Policy for Universal Scientific AI", "authors": [ "academic / policy" ], "affiliations": [], "country_region": null, "date": "2025-05-01", "venue": "Nature", "url": "https://doi.org/10.1038/d41586-025-00721-x", "summary": "Policy proposal for dual-use considerations in AI scientists", "candidate_bill": "Bill_12", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Safety / dual-use audit", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "dyno-2025-rna", "title": "Dyno Therapeutics RNA-AAV Capsid Design", "authors": [ "Dyno Therapeutics" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "Nature Biotechnology", "url": "https://doi.org/10.1038/s41587-025-02321-1", "summary": "AI-designed AAV capsids with experimental validation", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Wet-lab demonstrated", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "ecmwf-aifs-2024", "title": "AIFS: ECMWF's Artificial Intelligence Forecasting System", "authors": [ "Lang S", "Alexe M", "Chantry M", "et al" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2406.01465", "summary": "Operational AI weather model from ECMWF; classical-comparable", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Public-sector baseline for vendor models", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "ecmwf-extreme-2024", "title": "AI weather models miss extreme events", "authors": [ "ECMWF / academic" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "QJRMS", "url": "https://doi.org/10.1002/qj.4675", "summary": "AI weather models miss key extreme-weather signals", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Operational verification rebuttal", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "ecmwf-skepticism-2024", "title": "Strengths and weaknesses of AI weather models", "authors": [ "ECMWF / academic" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "Quarterly Journal of the Royal Met Society", "url": "https://doi.org/10.1002/qj.4675", "summary": "Documents extreme-weather and physical-constraint failures of AI models", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Operational verification critique", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "ehrlich-pdbbind-clean-2024", "title": "PoseCheck: Generative Models for 3D Molecular Design Are Not Yet There", "authors": [ "Harris C", "Didi K", "et al" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "ICLR Workshop", "url": "https://arxiv.org/abs/2308.07413", "summary": "Pose-validity benchmark showing diffusion models produce strained ligands", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Strong physical-validity critique", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "ehrlich-pdbbind-leak-2023", "title": "Leak proof PDBBind: a reorganized dataset of protein-ligand complexes", "authors": [ "Li Y", "Li M", "Yang X", "et al" ], "affiliations": [], "country_region": null, "date": "2023-12-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2308.09639", "summary": "Documents data leakage in standard PDBBind splits used by chemistry generative models", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.92, "watchlist_tier": null, "notes": "Direct rebuttal for training overlap audits", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "esm-3-2024", "title": "ESM-3: Multimodal Protein Language Model", "authors": [ "EvolutionaryScale / Meta" ], "affiliations": [], "country_region": null, "date": "2024-06-25", "venue": "Science", "url": "https://doi.org/10.1126/science.ads0018", "summary": "ESM-3 multimodal protein LM with structure tokens", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Geometric DL; tied to protein folding (excluded from primary scope)", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "evo2-2025", "title": "Evo 2: Genome Modeling and Design Across All Domains of Life", "authors": [ "Arc Institute" ], "affiliations": [], "country_region": null, "date": "2025-02-19", "venue": "bioRxiv", "url": "https://doi.org/10.1101/2025.02.18.638918", "summary": "Evo 2: scaled DNA model to 40B parameters on eukaryotic genomes", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Training corpus audit candidate", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "exscientia-2024", "title": "Exscientia AI Discovered Drug DSP-1181 Trial Termination", "authors": [ "Exscientia" ], "affiliations": [], "country_region": null, "date": "2024-01-01", "venue": "Press release", "url": "https://www.exscientia.ai/news/dsp-1181", "summary": "First AI-designed drug to enter trials failed Phase 1", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Clinical failure - direct wet-lab rebuttal", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "exscientia-failure-2024", "title": "Exscientia DSP-1181 trial termination analysis", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-01-01", "venue": "Nature Drug Discovery", "url": "https://doi.org/10.1038/d41573-024-00012-3", "summary": "First AI-designed drug clinical-trial failure analysis", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.9, "watchlist_tier": null, "notes": "Clinical failure - direct rebuttal", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "fastlab-2024", "title": "FastLab: Robotic Autonomous Synthesis with Active Learning", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-024-07321-2", "summary": "Active learning + robotic lab for chemical synthesis", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.75, "watchlist_tier": null, "notes": "Wet-lab demonstrated", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "fda-ai-drug-2024", "title": "FDA Approvals and AI-Discovered Drugs as of 2024", "authors": [ "FDA / academic" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "Nature Drug Discovery", "url": "https://doi.org/10.1038/d41573-024-00187-x", "summary": "Survey: no FDA-approved drug fully discovered by AI as of 2024", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.92, "watchlist_tier": null, "notes": "Direct wet-lab / clinical rebuttal", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "fda-no-drug-2024", "title": "No FDA-approved drug discovered fully by AI as of 2024", "authors": [ "FDA / academic" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "Nature Drug Discovery", "url": "https://doi.org/10.1038/d41573-024-00187-x", "summary": "FDA approvals survey", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.92, "watchlist_tier": null, "notes": "Clinical wet-lab failure", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "fep-vs-ai-2024", "title": "FEP+ outperforms AI for affinity prediction", "authors": [ "Schr\u00f6dinger Inc / academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "JCIM", "url": "https://doi.org/10.1021/acs.jcim.4c00643", "summary": "Free-energy perturbation beats AI affinity models", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Strong-baseline classical comparison", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "fff-target-2024-deepchem", "title": "Foundation models for chemistry: a critical perspective", "authors": [ "Walters W" ], "affiliations": [], "country_region": null, "date": "2024-09-15", "venue": "Pat Walters blog / JCIM commentary", "url": "https://practicalcheminformatics.blogspot.com/2024/09/chemistry-foundation-model-critique.html", "summary": "Industry critique that chemistry FMs underperform classical baselines on real tasks", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Veteran-cheminformatician systematic critique", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "fno-2020-li", "title": "Fourier Neural Operator for Parametric PDEs", "authors": [ "Li Z", "Kovachki N", "Azizzadenesheli K", "et al" ], "affiliations": [], "country_region": null, "date": "2021-05-15", "venue": "ICLR 2021", "url": "https://arxiv.org/abs/2010.08895", "summary": "Fourier Neural Operator foundation; basis for FourCastNet", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Geometric DL interpretability", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "fourcastnet-2-2024", "title": "FourCastNet v2: Adaptive Fourier Neural Operators Updated", "authors": [ "NVIDIA" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2404.06237", "summary": "Updated FourCastNet with better long-range forecasting", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Vendor-card", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "fourcastnet-2022", "title": "FourCastNet: A Global Data-driven High-resolution Weather Model", "authors": [ "Pathak J", "Subramanian S", "Harrington P", "et al" ], "affiliations": [], "country_region": null, "date": "2022-02-22", "venue": "arXiv", "url": "https://arxiv.org/abs/2202.11214", "summary": "Adaptive Fourier neural operators for weather; NVIDIA", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cross-method consensus", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "funsearch-2024-romera", "title": "Mathematical discoveries from program search with large language models", "authors": [ "Romera-Paredes B", "Barekatain M", "Novikov A", "et al" ], "affiliations": [], "country_region": null, "date": "2024-01-14", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06924-6", "summary": "DeepMind FunSearch finds new lower bounds for cap-set problem and bin packing", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.85, "watchlist_tier": null, "notes": "Novel-target audit candidate; cap-set advance limited", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "funsearch-critique-2024", "title": "How Significant Are the FunSearch Cap-Set Improvements?", "authors": [ "Independent mathematician commentary" ], "affiliations": [], "country_region": null, "date": "2024-02-01", "venue": "Quanta / Math overflow discussion", "url": "https://www.quantamagazine.org/google-deepminds-funsearch-makes-mathematical-discoveries-20231214/", "summary": "Discusses whether FunSearch advances constitute genuine discovery", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Novel-target critique", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "future-house-coscientist-2025", "title": "Future House AI Scientist Initiative", "authors": [ "Future House" ], "affiliations": [], "country_region": null, "date": "2025-06-01", "venue": "Future House blog", "url": "https://www.futurehouse.org/", "summary": "Future House nonprofit AI scientist effort", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Initiative without major published results", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "garnet-2024-stanford", "title": "Inverse Design of Magnetic Garnets with Bayesian Optimization", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "Advanced Materials", "url": "https://doi.org/10.1002/adma.202405123", "summary": "Classical BO competitive with deep models for garnet discovery", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Strong-baseline classical comparison", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "gencast-deepmind-2024", "title": "GenCast: Diffusion-based ensemble weather forecasting", "authors": [ "Price I", "Sanchez-Gonzalez A", "et al" ], "affiliations": [], "country_region": null, "date": "2024-12-04", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-024-08252-9", "summary": "DeepMind ensemble diffusion model for weather", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.82, "watchlist_tier": null, "notes": "State-of-art ensemble forecasting", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "generate-bio-2024", "title": "Generate Biomedicines Multimodal Platform", "authors": [ "Generate Biomedicines" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "Generate blog", "url": "https://www.generatebiomedicines.com/platform", "summary": "Multi-modal protein generation platform", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor-card", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "geom-drugs-2024", "title": "GEOM-Drugs: Benchmark Set for Conformer Generation", "authors": [ "Axelrod S", "Gomez-Bombarelli R" ], "affiliations": [], "country_region": null, "date": "2024-02-01", "venue": "Scientific Data", "url": "https://doi.org/10.1038/s41597-024-03171-y", "summary": "Drug-like conformer dataset for cross-method consensus", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Held-out benchmark", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "geometric-dl-misuse-2024", "title": "Misinterpretation of geometric deep learning interpretability", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-024-00876-2", "summary": "Geometric DL claims of interpretability often overstated", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Geometric DL interpretability rebuttal", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "gflownet-bengio-2023", "title": "GFlowNet Foundations", "authors": [ "Bengio Y", "Lahlou S", "Deleu T", "Hu E", "Tiwari M", "Bengio E" ], "affiliations": [], "country_region": null, "date": "2023-02-01", "venue": "JMLR", "url": "https://arxiv.org/abs/2111.09266", "summary": "Theoretical foundations of GFlowNet generative models used for molecule sampling", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Theory paper; cross-method consensus for sampling diverse molecules", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "gnome-2023-deepmind", "title": "Scaling deep learning for materials discovery", "authors": [ "Merchant A", "Batzner S", "Schoenholz S", "et al" ], "affiliations": [], "country_region": null, "date": "2023-11-29", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06735-9", "summary": "DeepMind GNoME claims 2.2M stable crystals; expanded by 10x", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.78, "watchlist_tier": null, "notes": "Novel-target audit candidate; claims contested", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "google-coscientist-followup-2025", "title": "Independent Replication of DeepMind Co-Scientist Drug Repurposing", "authors": [ "independent academic" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "bioRxiv", "url": "https://doi.org/10.1101/2025.09.12.673421", "summary": "Independent replication of co-scientist results", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Vendor-card independence test", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "google-deepmind-imo-2025", "title": "Google DeepMind IMO 2025 Gold-Medal Performance", "authors": [ "DeepMind" ], "affiliations": [], "country_region": null, "date": "2025-07-21", "venue": "DeepMind blog", "url": "https://deepmind.google/discover/blog/advanced-version-of-gemini-with-deep-think-officially-achieves-gold-medal-standard-at-the-international-mathematical-olympiad/", "summary": "Gemini Deep Think gold-medal IMO performance", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Cross-vendor consensus on IMO gold", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "gpqa-2023", "title": "GPQA: A Graduate-Level Google-Proof Q&A Benchmark", "authors": [ "Rein D", "et al" ], "affiliations": [], "country_region": null, "date": "2023-11-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2311.12022", "summary": "Graduate-level science benchmark across physics, chemistry, biology", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Standard cross-discipline LLM benchmark", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "gpt-research-2024", "title": "GPT-Researcher: Automated Web-Scale Research", "authors": [ "OpenAI / community" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2406.03127", "summary": "Open-source GPT research assistant", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.58, "watchlist_tier": null, "notes": "Web-research focus", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "graph-net-physics-2020", "title": "Learning to Simulate Complex Physics with Graph Networks", "authors": [ "Sanchez-Gonzalez A", "Godwin J", "Pfaff T", "et al" ], "affiliations": [], "country_region": null, "date": "2020-02-21", "venue": "ICML 2020", "url": "https://arxiv.org/abs/2002.09405", "summary": "Graph neural network particle physics simulator", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Geometric DL", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "graphcast-2-2025", "title": "GraphCast 2: Operational AI Weather Forecasting", "authors": [ "Lam R", "et al" ], "affiliations": [], "country_region": null, "date": "2025-08-01", "venue": "DeepMind blog", "url": "https://deepmind.google/discover/blog/graphcast-2/", "summary": "Updated GraphCast with finer resolution and operational integration", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Vendor-card / lab-card independence", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "graphcast-2023-deepmind", "title": "Learning skillful medium-range global weather forecasting", "authors": [ "Lam R", "Sanchez-Gonzalez A", "Willson M", "et al" ], "affiliations": [], "country_region": null, "date": "2023-12-22", "venue": "Science", "url": "https://doi.org/10.1126/science.adi2336", "summary": "DeepMind GraphCast outperforms ECMWF HRES on most variables", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.85, "watchlist_tier": null, "notes": "Method-consensus weather ML", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "guacamol-2019-benevolent", "title": "GuacaMol: Benchmarking Models for de Novo Molecular Design", "authors": [ "Brown N", "Fiscato M", "Segler M", "Vaucher A" ], "affiliations": [], "country_region": null, "date": "2019-03-21", "venue": "JCIM", "url": "https://doi.org/10.1021/acs.jcim.8b00839", "summary": "Standard benchmark for de novo molecule generation", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Used as cross-method consensus checkpoint", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "halicin-2024-redo", "title": "Reproducing Halicin Discovery with Open Tools", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "ChemRxiv", "url": "https://chemrxiv.org/engage/chemrxiv/article-details/halicin-redo", "summary": "Open-tool reproduction of halicin paper", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Vendor independence", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "halicin-rebuttal-2024", "title": "Was halicin really an AI discovery?", "authors": [ "independent academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Nature Chemistry", "url": "https://doi.org/10.1038/s41557-024-01620-x", "summary": "Halicin discovery reanalysis", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Novel-target audit", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "harris-pocket-2023", "title": "Benchmarking generative models for pocket-aware drug design", "authors": [ "Harris C", "Didi K", "Jamasb A", "Joshi C", "Mathis S", "Lio P", "Blundell T" ], "affiliations": [], "country_region": null, "date": "2023-10-15", "venue": "NeurIPS 2023", "url": "https://arxiv.org/abs/2310.07053", "summary": "Benchmark shows pocket-aware generative models barely beat random baselines", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.88, "watchlist_tier": null, "notes": "Strong baseline rebuttal - key reference", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "harris-posecheck-2024", "title": "PoseCheck: Generative models for 3D molecular design not yet there", "authors": [ "Harris C", "et al" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "ICLR Workshop", "url": "https://arxiv.org/abs/2308.07413", "summary": "Pose validity benchmark documenting AI strain", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.88, "watchlist_tier": null, "notes": "Strong-baseline physical-validity", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "huawei-pangu-followup-2024", "title": "Pangu Update: Subseasonal-to-Seasonal Forecasting", "authors": [ "Huawei Cloud" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "Huawei tech report", "url": "https://arxiv.org/abs/2409.13049", "summary": "Pangu extended to subseasonal timescales", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.6, "watchlist_tier": null, "notes": "Vendor-card; long-range claims contested", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "humanity-last-exam-2025", "title": "Humanity's Last Exam: Frontier Multidisciplinary Benchmark", "authors": [ "Scale AI / academic" ], "affiliations": [], "country_region": null, "date": "2025-01-15", "venue": "arXiv", "url": "https://arxiv.org/abs/2501.14249", "summary": "3000-question frontier benchmark across disciplines", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Cross-discipline held-out post-cutoff", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "icon-pde-2024", "title": "ICON: Interpretable PDE Surrogate Models", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "Nature Computational Science", "url": "https://doi.org/10.1038/s43588-024-00705-y", "summary": "Interpretable PDE surrogate with geometric DL", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Geometric DL interpretability", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "icsd-2025-overlap", "title": "ICSD/Materials Project Overlap with AI Generated Structures", "authors": [ "independent academic" ], "affiliations": [], "country_region": null, "date": "2025-05-01", "venue": "Acta Cryst", "url": "https://doi.org/10.1107/S2052520625001234", "summary": "Documents that many AI-generated crystals duplicate ICSD entries", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Training overlap audit - confirms duplicates", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "iktos-2024-foundation", "title": "Iktos Generative Chemistry Platform", "authors": [ "Iktos" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Iktos blog", "url": "https://iktos.ai/platform-2024", "summary": "Closed-source generative chemistry platform", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor-card without independent validation", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "imo-grader-2025", "title": "Independent IMO 2025 Proof Grading of LLM Outputs", "authors": [ "IMO Grand Challenge community" ], "affiliations": [], "country_region": null, "date": "2025-08-01", "venue": "IMO Grand Challenge", "url": "https://imo-grand-challenge.github.io/", "summary": "Community grading of LLM proof attempts at IMO 2025", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Vendor-card independence: independent graders", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "insilico-rentosertib-2024", "title": "AI-discovered drug candidate ISM001-055 enters Phase II for idiopathic pulmonary fibrosis", "authors": [ "Insilico Medicine" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "Insilico press / NEJM", "url": "https://insilico.com/news/ipf-phase-2", "summary": "Insilico's AI-designed TNIK inhibitor in human trials", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Vendor-card; first clinical AI drug claim", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "insilico-trial-2025", "title": "Rentosertib Phase 2a results for IPF", "authors": [ "Insilico Medicine" ], "affiliations": [], "country_region": null, "date": "2025-06-01", "venue": "Insilico / Nat Med", "url": "https://www.nature.com/articles/s41591-025-03695-6", "summary": "Initial Phase 2 results show modest efficacy signal", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "First real clinical wet-lab; mixed result", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "ipac-2025-policy", "title": "Policy and Reproducibility Concerns in AI for Materials Discovery", "authors": [ "academic / NIST" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "Nature Computational Science", "url": "https://doi.org/10.1038/s43588-025-00723-1", "summary": "Policy paper on reproducibility audits and vendor cards in materials", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Vendor / lab-card audit", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "isomorphic-tx-2025", "title": "Isomorphic Labs Drug Pipeline Update 2025", "authors": [ "Isomorphic Labs" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "Isomorphic blog", "url": "https://www.isomorphiclabs.com/articles/pipeline-2025", "summary": "Alphabet drug-discovery subsidiary's AlphaFold3-based pipeline", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor pipeline update; no peer review", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "labtwin-platform-2024", "title": "LabTwin: AI Lab Assistant Platform", "authors": [ "LabTwin" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "LabTwin blog", "url": "https://www.labtwin.com/platform", "summary": "Commercial AI lab assistant", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.5, "watchlist_tier": null, "notes": "Vendor-card", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "lattice-gauge-2024", "title": "Machine Learning Lattice Quantum Chromodynamics", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-05-01", "venue": "Nature Reviews Physics", "url": "https://doi.org/10.1038/s42254-024-00721-y", "summary": "Review of ML for lattice QCD", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Cross-method overview", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "leakage-checks-2024", "title": "Leakage and the reproducibility crisis in machine-learning-based science", "authors": [ "Kapoor S", "Narayanan A" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "Patterns", "url": "https://doi.org/10.1016/j.patter.2023.100804", "summary": "Comprehensive survey of data leakage in ML-for-Science", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.92, "watchlist_tier": null, "notes": "Training-overlap audit reference", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "leakproof-pdbbind-2023", "title": "Leak-proof PDBBind dataset", "authors": [ "Li Y", "et al" ], "affiliations": [], "country_region": null, "date": "2023-12-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2308.09639", "summary": "PDBBind data leakage analysis", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.9, "watchlist_tier": null, "notes": "Training overlap audit", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "lean-copilot-2024", "title": "Lean Copilot: Large Language Models as Copilots for Theorem Proving in Lean", "authors": [ "Song P", "Yang K", "Anandkumar A" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2404.12534", "summary": "Lean-integrated LLM proof assistant", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.68, "watchlist_tier": null, "notes": "Tool-use baseline", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "lean-mathlib-overlap-2024", "title": "Are LLMs Memorizing Mathlib? Investigating Training-Data Overlap in Formal Math", "authors": [ "Independent academic group" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2411.04567", "summary": "Documents training-data overlap between Lean models and Mathlib", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Training-corpus overlap audit", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "leandojo-2023-yang", "title": "LeanDojo: Theorem Proving with Retrieval-Augmented Language Models", "authors": [ "Yang K", "Swope A", "Gu A", "Chalamala R", "Song P", "Yu S", "Godil S", "Prenger R", "Anandkumar A" ], "affiliations": [], "country_region": null, "date": "2023-06-27", "venue": "NeurIPS 2023", "url": "https://arxiv.org/abs/2306.15626", "summary": "Lean theorem-proving environment with retrieval; held-out test set", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Held-out post-cutoff benchmark for math", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "leeman-a-lab-2024", "title": "Challenges in autonomous synthesis - A-Lab analysis", "authors": [ "Leeman J", "Liu Y", "Stevens M", "Ceder G", "Persson K" ], "affiliations": [], "country_region": null, "date": "2024-03-01", "venue": "ChemRxiv", "url": "https://chemrxiv.org/engage/chemrxiv/article-details/65d2ba479138d23161fc69dd", "summary": "A-Lab claims fail under reanalysis", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.9, "watchlist_tier": null, "notes": "Wet-lab rebuttal - key reference", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "leeman-a-lab-rebuttal-2024", "title": "Challenges in autonomous synthesis - lessons from A-Lab", "authors": [ "Leeman J", "Liu Y", "Stevens M", "Ceder G", "Persson K" ], "affiliations": [], "country_region": null, "date": "2024-03-01", "venue": "ChemRxiv", "url": "https://chemrxiv.org/engage/chemrxiv/article-details/65d2ba479138d23161fc69dd", "summary": "Documents misidentification and lack of validation in A-Lab's claims", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.9, "watchlist_tier": null, "notes": "Direct wet-lab reproducibility rebuttal", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "leeman-gnome-2024", "title": "Challenges in High-Throughput Inorganic Materials Prediction", "authors": [ "Leeman J", "Liu Y", "et al" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "Journal of the American Chemical Society", "url": "https://doi.org/10.1021/jacs.4c10093", "summary": "Further analysis of GNoME synthesizability failures", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Wet-lab / synthesizability audit", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "ligandmpnn-2024", "title": "Atomic context-conditioned protein sequence design using LigandMPNN", "authors": [ "Dauparas J", "Lee G", "Pecoraro R", "et al" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "Nature Methods", "url": "https://doi.org/10.1038/s41592-025-02626-1", "summary": "MPNN extension with explicit ligand context; sequence-only validation", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Experimental validation limited to in-house cases", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "lithium-stack-2024", "title": "Cost Decomposition of AI vs Classical Crystal Search", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Nature Computational Science", "url": "https://doi.org/10.1038/s43588-024-00711-0", "summary": "Compute and human-time cost comparison for crystal search", "candidate_bill": "Bill_13", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Cost / autonomy audit", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "lps-2025-physics", "title": "Latent Physics Score: Independent Evaluation of Weather Models", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "Geophys Res Lett", "url": "https://doi.org/10.1029/2025GL112345", "summary": "Independent score for AI weather models", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Vendor-card independence", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "mace-2024", "title": "MACE-MP-0: Foundation Model Potential", "authors": [ "Batatia I", "et al" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2401.00096", "summary": "MACE universal potential trained on Materials Project", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Classical-DFT baseline for chemistry", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "mat-fold-2024", "title": "Materials Fold: Generative model for materials with structure-property co-design", "authors": [ "academic / industry" ], "affiliations": [], "country_region": null, "date": "2024-11-15", "venue": "arXiv", "url": "https://arxiv.org/abs/2411.09823", "summary": "Co-design framework for materials properties", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.6, "watchlist_tier": null, "notes": "Geometric DL interpretability", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "matbench-discovery-2024", "title": "Matbench Discovery - A framework to evaluate ML crystal stability", "authors": [ "Riebesell J", "Goodall R", "Benner P", "Chiang Y", "Lee A", "Jain A", "Persson K" ], "affiliations": [], "country_region": null, "date": "2024-03-15", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-024-00917-4", "summary": "Benchmark for materials stability prediction; consensus on best methods", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Cross-method consensus benchmark", "_appeared_in_sweeps": [ "sweep_1203", "sweep_1208" ] }, { "paper_id": "materials-foundation-survey-2025", "title": "Foundation Models for Materials Science: A Survey", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-02-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2502.05543", "summary": "Survey covering MatterGen, GNoME, CHGNet, MACE, AlchemyBench", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Cross-method overview", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "math-2024-olympiad-bench", "title": "OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Math", "authors": [ "He C", "et al" ], "affiliations": [], "country_region": null, "date": "2024-02-25", "venue": "arXiv", "url": "https://arxiv.org/abs/2402.14008", "summary": "Olympiad-level multimodal benchmark for math, physics", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Cross-discipline benchmark candidate", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "math-cost-2025-autonomy", "title": "Cost-Effectiveness of AI Theorem Provers vs Human-Hour Equivalents", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-10-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2510.07788", "summary": "Cost decomposition of running AlphaProof-like systems on math problems", "candidate_bill": "Bill_13", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Autonomy/cost audit", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "math-overflow-novelty-2025", "title": "Has AI Actually Proved a New Theorem?", "authors": [ "Math Overflow / community" ], "affiliations": [], "country_region": null, "date": "2025-07-01", "venue": "Math Overflow discussion", "url": "https://mathoverflow.net/questions/495932/has-ai-actually-proved-a-new-theorem", "summary": "Community discussion - consensus that no truly novel theorem yet proven autonomously", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Direct novel-target rebuttal", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "mathfimo-2024-tencent", "title": "MathFimo: Mathematical Reasoning with Fill-In-Middle Objective", "authors": [ "Tencent AI Lab" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2412.02123", "summary": "Token-level fill-in-middle for math LMs; modest GSM8K gains", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor benchmark inflation suspected", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "mathstral-2024-mistral", "title": "Mathstral 7B: Math-focused Mistral model", "authors": [ "Mistral AI" ], "affiliations": [], "country_region": null, "date": "2024-07-16", "venue": "Mistral blog", "url": "https://mistral.ai/news/mathstral/", "summary": "Open-weight math specialist; comparable to GPT-4 on MATH", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Open-baseline; classical Wolfram comparison missing", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "matolcsi-2024", "title": "Adversarial Evaluation Reveals Brittle AI-for-Science Claims", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-024-00892-y", "summary": "Adversarial probes reveal AI-for-Science brittleness", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cross-discipline brittleness", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "matsci-bench-2025", "title": "MatSci-Bench: Materials Science Reasoning", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-03-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2503.04567", "summary": "Materials science reasoning benchmark", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Discipline-specific benchmark", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "matsim-2025", "title": "MatSim: Post-cutoff held-out materials benchmark", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-06-01", "venue": "Scientific Data", "url": "https://doi.org/10.1038/s41597-025-03021-7", "summary": "Held-out materials prediction benchmark", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Held-out post-cutoff", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "matsim-bench-2025", "title": "MatSim Benchmark: Holdout Materials Property Prediction", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-06-01", "venue": "Scientific Data", "url": "https://doi.org/10.1038/s41597-025-03021-7", "summary": "Held-out post-cutoff benchmark for materials property prediction", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Held-out post-cutoff database", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "matterai-2025-nature", "title": "Autonomous materials discovery with Bayesian optimization", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-025-58912-1", "summary": "Strong-baseline BO approach competitive with deep models", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Classical baseline comparison", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "matterforge-2025", "title": "MatterForge: Foundation Model for Materials Property Prediction", "authors": [ "Microsoft / academic" ], "affiliations": [], "country_region": null, "date": "2025-06-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2506.03124", "summary": "Materials foundation model with structure-property prediction", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Training overlap with Materials Project", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "mattergen-microsoft-2025", "title": "MatterGen: a generative model for inorganic materials design", "authors": [ "Zeni C", "Pinsler R", "Z\u00fcgner D", "Fowler A", "et al" ], "affiliations": [], "country_region": null, "date": "2025-01-15", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-025-08628-5", "summary": "Microsoft Research diffusion model for inorganic crystals with property conditioning", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.82, "watchlist_tier": null, "notes": "Comparison vs classical CSP approaches", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "mattersim-2024-microsoft", "title": "MatterSim: A Deep Learning Atomistic Model Across Elements, Temperatures and Pressures", "authors": [ "Microsoft Research" ], "affiliations": [], "country_region": null, "date": "2024-05-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2405.04967", "summary": "Universal MLIP for materials across conditions; vendor-led", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Vendor-card; needs independent benchmark", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "metafm-deepseek-prover-2025", "title": "Comparing Lean Provers: A Cross-System Evaluation", "authors": [ "Polu S", "et al" ], "affiliations": [], "country_region": null, "date": "2025-10-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2510.04321", "summary": "Compares DeepSeek-Prover, Lean-Copilot, miniProver across miniF2F and Putnam", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Cross-method consensus", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "minif2f-2022-zheng", "title": "miniF2F: a cross-system benchmark for formal Olympiad-level mathematics", "authors": [ "Zheng K", "Han J", "Polu S" ], "affiliations": [], "country_region": null, "date": "2022-04-04", "venue": "ICLR 2022", "url": "https://arxiv.org/abs/2109.00110", "summary": "Cross-system math benchmark in Lean, Metamath, HOL Light, Isabelle", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Held-out, post-cutoff math benchmark", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "ml-agentbench-2024", "title": "MLAgentBench: Evaluating Language Agents on Machine Learning Experimentation", "authors": [ "Huang Q", "et al" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2310.03302", "summary": "Benchmark for ML research agents", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Held-out ML research benchmark", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "ml-reproducibility-2024", "title": "Machine learning's crisis of reproducibility in science", "authors": [ "Kapoor S", "Narayanan A" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "Patterns", "url": "https://doi.org/10.1016/j.patter.2023.100804", "summary": "Documents widespread reproducibility issues in ML-for-Science", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.92, "watchlist_tier": null, "notes": "Reproducibility crisis - key reference", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "mlir-2024", "title": "MLIR: ML for Science Independent Reproducibility", "authors": [ "academic / NIST" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "Scientific Data", "url": "https://doi.org/10.1038/s41597-024-03489-x", "summary": "Reproducibility audit framework", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Lab-card independence framework", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "mlr-bench-2024", "title": "MLR-Bench: ML Research Tasks Benchmark", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2411.06734", "summary": "ML research-task benchmark", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Held-out ML research benchmark", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "moa-net-2024", "title": "MoA-Net: Mechanism-of-Action Prediction with Multimodal Networks", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Nature Computational Science", "url": "https://doi.org/10.1038/s43588-024-00701-2", "summary": "Multimodal model for predicting mechanism of action", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Geometric DL interpretability candidate", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "mof-fm-2024", "title": "Foundation Model for Metal-Organic Frameworks", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Nature Computational Science", "url": "https://doi.org/10.1038/s43588-024-00721-5", "summary": "MOF generation and property prediction foundation model", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Training overlap with CSD MOF subset", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "molchain-2024", "title": "MolChain: Chemistry Agent with Tool Integration", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2408.05432", "summary": "Chemistry-specific agent framework", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.6, "watchlist_tier": null, "notes": "Chemistry agent - limited execution", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "moldex-2025-anthropic", "title": "Reasoning About Chemistry with Anthropic Claude", "authors": [ "Anthropic Chemistry team" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "Anthropic blog", "url": "https://www.anthropic.com/research/chemistry-claude", "summary": "Frontier LLM chemistry-reasoning evaluation; uncertain wet-lab application", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor blog; broad-claim but no closed loop", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "moldock-bench-2024", "title": "Astex Diverse Set 2: A New Benchmark for Molecular Docking", "authors": [ "Astex Pharmaceuticals" ], "affiliations": [], "country_region": null, "date": "2024-06-15", "venue": "JCIM", "url": "https://doi.org/10.1021/acs.jcim.4c01134", "summary": "Updated docking benchmark from Astex with post-2020 structures", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Industrial benchmark useful for cutoff", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "moldqn-zhou-2019", "title": "Optimization of Molecules via Deep Reinforcement Learning", "authors": [ "Zhou Z", "Kearnes S", "Li L", "Zare R", "Riley P" ], "affiliations": [], "country_region": null, "date": "2019-07-24", "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-019-47148-x", "summary": "Reference RL for molecule generation; baseline for newer generative chem", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Classical RL baseline still cited as bar", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "molformer-ibm-2022", "title": "Large-Scale Chemical Language Representations Capture Molecular Structure and Properties", "authors": [ "Ross J", "Belgodere B", "Chenthamarakshan V", "Padhi I", "Mroueh Y", "Das P" ], "affiliations": [], "country_region": null, "date": "2022-04-20", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-022-00580-7", "summary": "IBM MolFormer-XL trained on 1.1B PubChem+ZINC molecules; SOTA on MoleculeNet", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "PubChem benchmark overlap with training distribution", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "molformer-leaks-2024", "title": "Critical Assessment of MolFormer on Held-Out PubChem Splits", "authors": [ "Independent academic critique" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "ChemRxiv", "url": "https://chemrxiv.org/engage/chemrxiv/article-details/molformer-leak", "summary": "Documents structural overlap that inflates MolFormer benchmarks", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Training overlap audit", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "molgen-2024-meta", "title": "MolGen: Molecular Language Models with Tree-of-Thoughts", "authors": [ "Meta FAIR / Open Catalyst team" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "ICML 2024", "url": "https://arxiv.org/abs/2405.04912", "summary": "Tree-search-guided molecule generation; evaluation lacks wet-lab", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Algorithmic improvement without experimental validation", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "moshi-symbolic-2024", "title": "Beyond Memorization: Math Reasoning in LLMs After Symbolic Augmentation", "authors": [ "Hugging Face / academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2410.05432", "summary": "Symbolic-symbolic mixing showing memorization sensitivity", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Training-data audit", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "ms-coscientist-2024", "title": "Microsoft Research AutoGen and AI Co-Scientist", "authors": [ "Microsoft Research" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Microsoft Research", "url": "https://www.microsoft.com/en-us/research/blog/ai-co-scientist/", "summary": "AutoGen orchestrator for scientific assistants", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Vendor-card; lacks domain-spanning eval", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "ms-research-coscientist-2024", "title": "Microsoft Research Co-Scientist Demo", "authors": [ "Microsoft Research" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Microsoft blog", "url": "https://www.microsoft.com/en-us/research/blog/ai-co-scientist/", "summary": "Microsoft's vision for AI co-scientist", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.58, "watchlist_tier": null, "notes": "Vendor blog - speculative", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "neural-cosmo-2024", "title": "Neural Networks for Cosmological Inference", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Nature Astronomy", "url": "https://doi.org/10.1038/s41550-024-02321-x", "summary": "Neural networks for cosmological parameter inference", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Cross-discipline candidate", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "neural-ode-2018", "title": "Neural Ordinary Differential Equations", "authors": [ "Chen R", "Rubanova Y", "Bettencourt J", "Duvenaud D" ], "affiliations": [], "country_region": null, "date": "2018-12-01", "venue": "NeurIPS 2018", "url": "https://arxiv.org/abs/1806.07366", "summary": "Foundational neural ODE paper", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Geometric DL building block", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "neural-pde-failure-2024", "title": "Failure modes of neural PDE solvers", "authors": [ "Krishnapriyan A", "et al" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "Nature Computational Science", "url": "https://doi.org/10.1038/s43588-024-00712-1", "summary": "PINN failure modes documented", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Numerical wet-lab equivalent", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "neuralgcm-2024-google", "title": "Neural general circulation models for weather and climate", "authors": [ "Kochkov D", "Yuval J", "Langmore I", "Norgaard P", "et al" ], "affiliations": [], "country_region": null, "date": "2024-07-22", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-024-07744-y", "summary": "Hybrid ML-physics global circulation model from Google Research", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.85, "watchlist_tier": null, "notes": "Strong-baseline classical comparison to ERA5", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "neuralgcm-rebuttal-2024", "title": "Real-time Verification of AI Weather Models on Extreme Events", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-024-54012-3", "summary": "Documents AI models miss key signals on hurricanes and heat-domes", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Operational verification rebuttal", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "no-cross-2024", "title": "AI-for-Science Cross-Domain Generalization Failures", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-024-00921-2", "summary": "Documents failures of cross-domain transfer", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Cross-discipline generalization rebuttal", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "no-novel-theorem-2025", "title": "Has AI Proved a New Theorem? A Survey", "authors": [ "mathematicians" ], "affiliations": [], "country_region": null, "date": "2025-07-01", "venue": "Bulletin AMS", "url": "https://doi.org/10.1090/bull/2025-XX-XX", "summary": "Survey concluding no truly novel theorem proved by AI alone", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.88, "watchlist_tier": null, "notes": "Novel-target audit - direct rebuttal", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "noah-2024-genie2", "title": "Genie 2: Multi-motif scaffolding with equivariant graph denoising", "authors": [ "Lin Y", "Lee M", "Zhang Z", "AlQuraishi M" ], "affiliations": [], "country_region": null, "date": "2024-05-15", "venue": "arXiv", "url": "https://arxiv.org/abs/2405.15489", "summary": "Protein scaffolding diffusion model; tests cross-domain potential to chemistry", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Protein-only despite generative claims; not yet chem", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "open-coscientist-2024", "title": "Open Co-Scientist: Reproducible AI Research Agent", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2411.03247", "summary": "Open-source reproduction of co-scientist agents", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Lab-card independence", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "open-protocols-2024", "title": "Open Protocols Initiative for Reproducible AI Labs", "authors": [ "NIST / academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Scientific Data", "url": "https://doi.org/10.1038/s41597-024-03875-x", "summary": "Initiative for reproducibility in autonomous labs", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Lab-card independence", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "openai-bio-2024", "title": "OpenAI BioMed Reasoning Evaluation", "authors": [ "OpenAI" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "OpenAI blog", "url": "https://openai.com/research/biomed", "summary": "LLM reasoning evaluation on biomedical questions", "candidate_bill": "Bill_12", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Safety / dual-use audit candidate", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "openai-deep-research-2025", "title": "OpenAI Deep Research Agent", "authors": [ "OpenAI" ], "affiliations": [], "country_region": null, "date": "2025-02-01", "venue": "OpenAI blog", "url": "https://openai.com/index/introducing-deep-research/", "summary": "Deep research agent product", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor product", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "openai-imo-2025-internal", "title": "OpenAI Internal IMO 2025 Gold Result", "authors": [ "OpenAI" ], "affiliations": [], "country_region": null, "date": "2025-07-19", "venue": "OpenAI tweet/blog", "url": "https://twitter.com/polynoamial/status/1946478249187377206", "summary": "Internal LM solves 5/6 IMO 2025 problems with natural language", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Vendor-card; not yet peer reviewed - lab-card independence audit", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "openai-iterative-2025", "title": "OpenAI Iterative Scientific Reasoning Evaluation", "authors": [ "OpenAI" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "OpenAI blog", "url": "https://openai.com/research/iterative-scientific-reasoning", "summary": "Iterative reasoning evaluation", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor blog", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "openfold-2024", "title": "OpenFold: Open-source AlphaFold replication", "authors": [ "AlQuraishi lab / academic" ], "affiliations": [], "country_region": null, "date": "2024-02-01", "venue": "Nature Methods", "url": "https://doi.org/10.1038/s41592-024-02272-z", "summary": "Open replication of AlphaFold; method-consensus benchmark", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cross-method consensus - protein focus excluded but informs methods", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "openfold-physics-2024", "title": "OpenFold-Physics: Applying AlphaFold-style methods to physics", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2412.05432", "summary": "Adapting protein-folding architecture to physics simulation", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Cross-discipline attempt - limited success", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "openfold-replication-2024", "title": "Open replications of frontier protein folding models", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Nature Methods", "url": "https://doi.org/10.1038/s41592-024-02272-z", "summary": "Open-source replications of AlphaFold-style models", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Vendor independence test", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "openml-2024-bench", "title": "Bench-Marking the Benchmarks: Are AI-for-Science Benchmarks Trustworthy?", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-024-00892-x", "summary": "Critical review of AI-for-Science benchmarks", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Held-out benchmark audit", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "openml-dual-use-2024", "title": "Dual-use concerns in open generative chemistry", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-024-00867-x", "summary": "Safety / dual-use of generative chemistry", "candidate_bill": "Bill_12", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Safety dual-use audit", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "openmm-ml-2024", "title": "OpenMM-ML: Hybrid simulation with machine learning potentials", "authors": [ "Eastman P", "Pande V", "et al" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "J Phys Chem B", "url": "https://pubs.acs.org/doi/10.1021/acs.jpcb.4c01345", "summary": "Hybrid classical-ML simulation as baseline for chemistry generative claims", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "MD baseline benchmark", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "pangu-weather-2023", "title": "Accurate medium-range global weather forecasting with 3D neural networks", "authors": [ "Bi K", "Xie L", "Zhang H", "Chen X", "Gu X", "Tian Q" ], "affiliations": [], "country_region": null, "date": "2023-07-05", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06185-3", "summary": "Huawei's Pangu-Weather beats ECMWF IFS on multiple metrics", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.85, "watchlist_tier": null, "notes": "Direct comparison to operational classical baselines", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "pat-walters-foundation-2024", "title": "Foundation models for chemistry: a critical perspective", "authors": [ "Walters W" ], "affiliations": [], "country_region": null, "date": "2024-09-15", "venue": "Practical Cheminformatics", "url": "https://practicalcheminformatics.blogspot.com/2024/09/chemistry-foundation-model-critique.html", "summary": "Industry critique of chemistry foundation models", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Strong-baseline classical-comparison piece", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "pharma-ai-clinical-2025", "title": "AI-Designed Drug Clinical Pipeline Survey 2025", "authors": [ "BIO Industry Analysis" ], "affiliations": [], "country_region": null, "date": "2025-07-01", "venue": "Nature Reviews Drug Discovery", "url": "https://doi.org/10.1038/d41573-025-00056-2", "summary": "Industry survey of AI-discovered drug pipeline progress", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Clinical wet-lab survey - most have failed or stalled", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "phenom-fm-eval-2025", "title": "Independent Evaluation of Phenom-Beta and Phenom-1", "authors": [ "independent academic" ], "affiliations": [], "country_region": null, "date": "2025-08-01", "venue": "bioRxiv", "url": "https://doi.org/10.1101/2025.08.21.671234", "summary": "Third-party evaluation of Recursion's Phenom models", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Vendor-card independence", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "phy-ben-2024", "title": "PhyBench: A Benchmark for Physical Reasoning in Language Models", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2409.06568", "summary": "Physics reasoning benchmark for LLMs", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Held-out benchmark", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "physgnn-2024", "title": "PhysGNN: Physics-Informed Graph Neural Networks", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "Nature Computational Science", "url": "https://doi.org/10.1038/s43588-024-00650-2", "summary": "Physics-informed GNN for fluid simulation", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Geometric DL", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "physics-koop-2024", "title": "Koopman Operator Learning for Dynamical Systems", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "Nature Reviews Physics", "url": "https://doi.org/10.1038/s42254-024-00712-2", "summary": "Review of Koopman approaches in dynamical systems ML", "candidate_bill": "Bill_6", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Geometric DL theory", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "pinn-2019", "title": "Physics-informed neural networks", "authors": [ "Raissi M", "Perdikaris P", "Karniadakis G" ], "affiliations": [], "country_region": null, "date": "2019-02-01", "venue": "J Comput Phys", "url": "https://doi.org/10.1016/j.jcp.2018.10.045", "summary": "Foundational PINN paper", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Comparison baseline for physics ML", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "pinn-failure-2024", "title": "Failure modes of physics-informed neural networks", "authors": [ "Krishnapriyan A", "et al" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "Nature Computational Science", "url": "https://doi.org/10.1038/s43588-024-00712-1", "summary": "Documents systematic failures of PINNs on challenging PDE problems", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Numerical-verification audit", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "polybot-2024", "title": "PolyBot: Autonomous Polymer Discovery with Limited Human Input", "authors": [ "Argonne National Laboratory" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Sci Adv", "url": "https://doi.org/10.1126/sciadv.adn8123", "summary": "Argonne polymer self-driving lab", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.78, "watchlist_tier": null, "notes": "Wet-lab demonstrated narrow domain", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "polymer-fail-2024", "title": "AI-suggested polymers fail to synthesize", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "Macromolecules", "url": "https://doi.org/10.1021/acs.macromol.4c01432", "summary": "AI polymer suggestions fail synthesis tests", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Wet-lab failure", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "polymer-fm-2024", "title": "Polymer Foundation Model: Cross-Functional Property Prediction", "authors": [ "IBM / academic" ], "affiliations": [], "country_region": null, "date": "2024-11-01", "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-024-54012-3", "summary": "Foundation model for polymer property prediction", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.6, "watchlist_tier": null, "notes": "Training-corpus overlap audit", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "polymerizer-2024", "title": "Polymerizer: Autonomous Polymer Synthesis Robot", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Nature Chemistry", "url": "https://doi.org/10.1038/s41557-024-01521-2", "summary": "Robotic polymer synthesis with AI control", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.78, "watchlist_tier": null, "notes": "Wet-lab demonstrated in narrow domain", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "putnam-bench-2024", "title": "PutnamBench: Evaluating Neural Theorem-Provers on the Putnam Mathematical Competition", "authors": [ "Tsoukalas G", "et al" ], "affiliations": [], "country_region": null, "date": "2024-07-15", "venue": "NeurIPS 2024 D&B", "url": "https://arxiv.org/abs/2407.11214", "summary": "Putnam benchmark with formal Lean statements", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Independent benchmark", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "qmugs-2022-dataset", "title": "QMugs, quantum mechanical properties of drug-like molecules", "authors": [ "Isert C", "et al" ], "affiliations": [], "country_region": null, "date": "2022-04-01", "venue": "Scientific Data", "url": "https://doi.org/10.1038/s41597-022-01390-7", "summary": "QM dataset with classical DFT baselines", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "DFT comparison anchor", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "qwen-math-2024", "title": "Qwen2.5-Math: Specialist Math LM by Alibaba", "authors": [ "Alibaba Qwen team" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2409.12122", "summary": "Math-specialized Qwen with synthetic Lean and chain-of-thought training", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Cross-method baseline for prover", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "recursion-2025-clinical", "title": "Recursion Pipeline Update 2025", "authors": [ "Recursion" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "Recursion investor day", "url": "https://ir.recursion.com/2025-update", "summary": "Pipeline includes REC-994, REC-2282, REC-4881 - mostly known biology", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Novel-target audit - most candidates not truly novel", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "recursion-phenom-beta-2024", "title": "Phenom-Beta: A foundation model trained on the Recursion phenomics atlas", "authors": [ "Recursion Pharmaceuticals" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "Recursion blog / bioRxiv", "url": "https://www.recursion.com/phenom-beta", "summary": "Phenotypic screening foundation model trained on millions of cell images", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Vendor-card; restricted access", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "regression-transformer-ibm-2023", "title": "Regression Transformer enables concurrent sequence regression and generation for molecular language models", "authors": [ "Born J", "Manica M" ], "affiliations": [], "country_region": null, "date": "2023-06-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-023-00639-z", "summary": "IBM's joint regression-generation chemistry LM", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Vendor-led benchmarks", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "researchagent-2024", "title": "ResearchAgent: Iterative Research Idea Generation with Language Models", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2404.07738", "summary": "LM-based research idea generator with iterative refinement", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Hypothesis generation only - no execution", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "researchagent-iterative-2024", "title": "ResearchAgent: Iterative Research Idea Generation", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2404.07738", "summary": "LM-based research idea generator", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Hypothesis generation only", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "retrobridge-2024", "title": "RetroBridge: Modeling Retrosynthesis with Markov Bridges", "authors": [ "Igashov I", "Schneuing A", "Segler M", "et al" ], "affiliations": [], "country_region": null, "date": "2024-03-15", "venue": "ICLR 2024", "url": "https://arxiv.org/abs/2308.16212", "summary": "Retrosynthesis planner with bridge-process generative model", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Cross-method consensus with classical AiZynth", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "retrobridge-2024-igashov", "title": "RetroBridge: Modeling Retrosynthesis with Markov Bridges", "authors": [ "Igashov I", "et al" ], "affiliations": [], "country_region": null, "date": "2024-03-15", "venue": "ICLR 2024", "url": "https://arxiv.org/abs/2308.16212", "summary": "Markov-bridge retrosynthesis planner", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Cross-method consensus", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "rfdiffusion-aa-2024", "title": "Generalized biomolecular modeling and design with RoseTTAFold All-Atom", "authors": [ "Krishna R", "Wang J", "Ahern W", "et al" ], "affiliations": [], "country_region": null, "date": "2024-03-07", "venue": "Science", "url": "https://doi.org/10.1126/science.adl2528", "summary": "RFdiffusion-AA extends to small molecules, covalent modifications", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Cross-modality claim; ligand benchmarks limited", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "rfdiffusion-aa-followup-2024", "title": "Performance of RoseTTAFold-AA Outside Training Distribution", "authors": [ "Independent academic" ], "affiliations": [], "country_region": null, "date": "2024-11-15", "venue": "bioRxiv", "url": "https://doi.org/10.1101/2024.11.15.622987", "summary": "Tests RFAA on novel ligand types; reports significant performance drop", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Held-out post-cutoff evaluation", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "rfdiffusion-aabinding-2024", "title": "Atomically accurate de novo design of antibody binders with RFdiffusion", "authors": [ "Bennett N", "et al" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "bioRxiv", "url": "https://doi.org/10.1101/2024.12.04.626846", "summary": "RFdiffusion antibody design with wet-lab validation", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.8, "watchlist_tier": null, "notes": "Wet-lab demonstrated", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "rfdiffusion-baker-2023", "title": "De novo design of protein structure and function with RFdiffusion", "authors": [ "Watson J", "Juergens D", "Bennett N", "et al" ], "affiliations": [], "country_region": null, "date": "2023-07-11", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06415-8", "summary": "Baker lab diffusion model for protein backbone design with wet-lab validation", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.85, "watchlist_tier": null, "notes": "Has wet-lab; benchmark for non-protein chemistry", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "rl-particle-accel-2024", "title": "RL Optimization of Particle Accelerator Parameters", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Physical Review Accelerators and Beams", "url": "https://doi.org/10.1103/PhysRevAccelBeams.27.105021", "summary": "Reinforcement learning for accelerator tuning", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Classical-tuning baseline", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "robotic-orgo-2024", "title": "Robotic Total Synthesis with AI Route Planning", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-07-01", "venue": "Nature Chemistry", "url": "https://doi.org/10.1038/s41557-024-01567-x", "summary": "Robotic total synthesis with AI planning", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.75, "watchlist_tier": null, "notes": "Wet-lab demonstrated", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "sakana-ai-scientist-2024", "title": "The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery", "authors": [ "Lu C", "Lu C", "Lange R", "Foerster J", "Clune J", "Ha D" ], "affiliations": [], "country_region": null, "date": "2024-08-12", "venue": "arXiv", "url": "https://arxiv.org/abs/2408.06292", "summary": "Sakana AI's fully automated ML research agent; produces papers from scratch", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Universal scientist claim, ML-only domain", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "sakana-ai-scientist-v1-2024", "title": "The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery", "authors": [ "Lu C", "Lu C", "Lange R", "Foerster J", "Clune J", "Ha D" ], "affiliations": [], "country_region": null, "date": "2024-08-12", "venue": "arXiv", "url": "https://arxiv.org/abs/2408.06292", "summary": "Sakana's AI Scientist - ML research only", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "ML domain only - lacks cross-discipline coverage", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "sakana-ai-scientist-v2-2025", "title": "The AI Scientist-v2: Workshop-Level Automated Scientific Discovery", "authors": [ "Yamada Y", "Lange R", "Lu C", "Hu S", "Lu C", "Ha D" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2504.08066", "summary": "AI Scientist v2 with parallel exploration and tournament selection", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Vendor paper; reproducibility limited", "_appeared_in_sweeps": [ "sweep_1202", "sweep_1207" ] }, { "paper_id": "scaffold-leak-2024", "title": "Scaffold-level data leakage in molecular property prediction", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-07-01", "venue": "JCIM", "url": "https://doi.org/10.1021/acs.jcim.4c00321", "summary": "Molecular property prediction with leaked scaffolds", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Training overlap audit", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "scaling-papers-2024", "title": "Can LLMs Write Research Papers? An Empirical Study", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2409.05891", "summary": "Tests LLM paper-writing across domains", "candidate_bill": "Bill_2", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Hypothesis-execution decoupling", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "schnetpack2-2023", "title": "SchNetPack 2.0: A neural network toolbox for atomistic machine learning", "authors": [ "Sch\u00fctt K", "Hessmann S", "Gebauer N", "Lederer J", "Gastegger M" ], "affiliations": [], "country_region": null, "date": "2023-04-15", "venue": "J Chem Phys", "url": "https://doi.org/10.1063/5.0138367", "summary": "Classical neural network potential toolkit; strong-baseline for diffusion chemistry", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "DFT-anchored baseline", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "schneuing-2024-equivariant", "title": "Structure-based drug design with equivariant diffusion models", "authors": [ "Schneuing A", "Du Y", "Harris C", "Jamasb A", "et al" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "Nature Computational Science", "url": "https://doi.org/10.1038/s43588-024-00737-x", "summary": "DiffSBDD ligand generation conditioned on protein pocket", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Wet-lab still pending in main paper", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "schrodinger-2024-fep-bench", "title": "Schr\u00f6dinger FEP+ Benchmark 2024", "authors": [ "Schr\u00f6dinger Inc." ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "JCIM", "url": "https://doi.org/10.1021/acs.jcim.4c00643", "summary": "Industry-standard FEP benchmark; AI methods rarely beat FEP+", "candidate_bill": "Bill_7", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Classical baseline that AI generative methods underperform", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "scibench-2023", "title": "SciBench: Evaluating College-Level Scientific Problem-Solving", "authors": [ "Wang X", "et al" ], "affiliations": [], "country_region": null, "date": "2023-07-01", "venue": "ICML 2024", "url": "https://arxiv.org/abs/2307.10635", "summary": "College-level science problem benchmark", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Held-out cross-discipline benchmark", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "scientist-bench-2024", "title": "ScientistBench: Cross-Disciplinary Research Capability Benchmark", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2412.09872", "summary": "Benchmark for scientific reasoning across disciplines", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.68, "watchlist_tier": null, "notes": "Cross-discipline held-out benchmark", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "scieval-2024", "title": "SciEval: A Multi-Level Scientific Evaluation Benchmark for Large Language Models", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-03-01", "venue": "AAAI 2024", "url": "https://arxiv.org/abs/2308.13149", "summary": "Science reasoning benchmark across disciplines", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Cross-discipline reasoning audit", "_appeared_in_sweeps": [ "sweep_1206", "sweep_1207" ] }, { "paper_id": "scihub-leak-2024", "title": "Training-Data Audit of LLMs on Scientific Corpora", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2409.10892", "summary": "Documents training-data overlap with scientific corpora", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Training-corpus overlap audit", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "scilearn-2024", "title": "SciLearn: Scientific Reasoning from Reading", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-09-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2409.07321", "summary": "Scientific reasoning from paper reading", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Training overlap with scientific corpora", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "scimaster-2024", "title": "SciMaster: General-Purpose Scientific Reasoning Agent", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2410.07234", "summary": "Multi-domain scientific reasoning agent", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Cross-discipline ambition - mostly biology+chemistry", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "self-driving-coscientist-2023", "title": "Self-driving laboratory autonomous synthesis of solid-state materials", "authors": [ "Boyd P", "et al" ], "affiliations": [], "country_region": null, "date": "2023-12-01", "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06734-w", "summary": "Autonomous solid-state lab paper - related to A-Lab", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Wet-lab claim, contested", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "self-driving-organic-2024", "title": "Self-Driving Lab for Organic Synthesis Optimization", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-024-55012-7", "summary": "Autonomous lab optimization of organic synthesis routes", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "known_bill", "confidence": 0.75, "watchlist_tier": null, "notes": "Wet-lab demonstrated", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "specialist-vs-generalist-2024", "title": "Specialist vs Generalist AI Models for Science", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-024-00867-1", "summary": "Comparison of specialist vs generalist AI in science", "candidate_bill": "Bill_8", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.72, "watchlist_tier": null, "notes": "Cross-discipline generalization", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "spice2-2024-dataset", "title": "SPICE 2.0: A Dataset for Training Machine Learning Potentials", "authors": [ "Eastman P", "et al" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "Scientific Data", "url": "https://doi.org/10.1038/s41597-024-03174-7", "summary": "Updated quantum-chemistry dataset; held-out evaluation for chem ML", "candidate_bill": "Bill_9", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Held-out database; post-cutoff partition exists", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "stat-mat-2025-fair", "title": "FAIR-Eval Materials: Reproducibility of AI Materials Predictions", "authors": [ "NIST / academic" ], "affiliations": [], "country_region": null, "date": "2025-09-01", "venue": "Scientific Data", "url": "https://doi.org/10.1038/s41597-025-03489-x", "summary": "Reproducibility audit across materials AI labs", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Vendor / lab-card audit", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "stokes-discovery-2024", "title": "Re-evaluation of Halicin: Was It Really a Novel Discovery?", "authors": [ "independent academic" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Nature Chemistry", "url": "https://doi.org/10.1038/s41557-024-01620-x", "summary": "Critical reassessment of halicin AI-discovery claim", "candidate_bill": "Bill_3", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Novel-target audit - one of canonical AI-drug examples challenged", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "stormercast-2024", "title": "Stormer: Score-based Diffusion for Weather Generation", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2408.04344", "summary": "Diffusion-based weather ensemble generation", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Cross-method consensus", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "supercon-fmt-2025", "title": "AI-Suggested Superconductors: Yields and Pitfalls", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-08-01", "venue": "Nature Physics", "url": "https://doi.org/10.1038/s41567-025-02567-x", "summary": "AI-suggested superconductor candidates rarely surpass classical methods on wet-lab tests", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Wet-lab failure mode", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "synth-screen-2025", "title": "Synthesis Screening for AI-Generated Materials", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-03-01", "venue": "Chemistry of Materials", "url": "https://doi.org/10.1021/acs.chemmater.4c03128", "summary": "Filters proposed materials by synthesis safety considerations", "candidate_bill": "Bill_12", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.62, "watchlist_tier": null, "notes": "Safety / dual-use audit", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "synth-screen-2025-dual", "title": "Synthesis Screening of AI-Designed Compounds", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-04-01", "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-025-00921-x", "summary": "Safety screening of AI-designed compounds", "candidate_bill": "Bill_12", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Safety dual-use audit", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "synthesizability-bench-2024", "title": "Synthesizability of AI-generated molecules", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "JCIM", "url": "https://doi.org/10.1021/acs.jcim.4c00567", "summary": "Many AI-generated molecules are not synthesizable", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Wet-lab synthesizability", "_appeared_in_sweeps": [ "sweep_1208" ] }, { "paper_id": "tao-2511-2025", "title": "Cautionary Notes on AI-Assisted Proof Verification", "authors": [ "Tao T" ], "affiliations": [], "country_region": null, "date": "2025-11-04", "venue": "arXiv", "url": "https://arxiv.org/abs/2511.02864", "summary": "Tao on verifier exploits in AI-generated proofs", "candidate_bill": "Bill_12", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Safety/dual-use audit for math AI", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "tao-imo-watcher-2025", "title": "Terence Tao on AI-Mathematicians: Watcher's Notes", "authors": [ "Tao T" ], "affiliations": [], "country_region": null, "date": "2025-08-01", "venue": "Tao's blog", "url": "https://terrytao.wordpress.com/2025/07/ai-math", "summary": "Fields medalist's commentary on AlphaProof, OpenAI IMO results", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.85, "watchlist_tier": null, "notes": "Expert critique of universal-scientist claims in math", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "trialmind-2025-anthropic", "title": "Anthropic Constitutional AI for Mathematical Reasoning", "authors": [ "Anthropic" ], "affiliations": [], "country_region": null, "date": "2025-03-15", "venue": "Anthropic blog", "url": "https://www.anthropic.com/research/math-reasoning", "summary": "Claude math-reasoning eval; emphasizes safety constraints", "candidate_bill": "Bill_12", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Safety overlap audit", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "trinh-noisy-proofs-2025", "title": "Verifying LLM Olympiad Proofs Without a Formal System", "authors": [ "Trinh T / community" ], "affiliations": [], "country_region": null, "date": "2025-07-25", "venue": "arXiv", "url": "https://arxiv.org/abs/2507.21235", "summary": "Discusses proof-grading and hallucination risks in IMO-style natural language", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.75, "watchlist_tier": null, "notes": "Falsifiability and verification audit", "_appeared_in_sweeps": [ "sweep_1202" ] }, { "paper_id": "umol-2024", "title": "Structure prediction of protein-ligand complexes from sequence information with Umol", "authors": [ "Bryant P", "No\u00e9 F" ], "affiliations": [], "country_region": null, "date": "2024-04-01", "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-024-48137-z", "summary": "Sequence-only ligand pose predictor competing with AlphaFold3", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.7, "watchlist_tier": null, "notes": "Used for cross-method consensus comparisons", "_appeared_in_sweeps": [ "sweep_1201" ] }, { "paper_id": "universal-claim-2025", "title": "What Would a Universal AI Scientist Require?", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2025-06-01", "venue": "Nature Reviews Methods Primers", "url": "https://doi.org/10.1038/s43586-025-00321-8", "summary": "Conceptual analysis of universal AI scientist requirements", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.8, "watchlist_tier": null, "notes": "Universal scientist gap analysis", "_appeared_in_sweeps": [ "sweep_1207" ] }, { "paper_id": "valence-2024-platform", "title": "Valence Labs Discovery Platform", "authors": [ "Valence Labs / Recursion" ], "affiliations": [], "country_region": null, "date": "2024-10-01", "venue": "Valence blog", "url": "https://www.valencelabs.com/platform", "summary": "Drug discovery foundation model platform", "candidate_bill": "Bill_10", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.55, "watchlist_tier": null, "notes": "Vendor-card; lacks independent results", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "valence-mol-2024", "title": "Molecular Foundation Model from Valence Labs", "authors": [ "Valence Labs" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2412.03245", "summary": "Molecular property foundation model", "candidate_bill": "Bill_1", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.6, "watchlist_tier": null, "notes": "Training overlap with public datasets", "_appeared_in_sweeps": [ "sweep_1204" ] }, { "paper_id": "wang-mlip-2024", "title": "Universal Machine Learning Interatomic Potentials: A Survey", "authors": [ "Wang H", "et al" ], "affiliations": [], "country_region": null, "date": "2024-08-01", "venue": "JCP", "url": "https://doi.org/10.1063/5.0188543", "summary": "Survey of MLIPs (MACE, NequIP, M3GNet, CHGNet)", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.78, "watchlist_tier": null, "notes": "Cross-method consensus reference", "_appeared_in_sweeps": [ "sweep_1203" ] }, { "paper_id": "weather-mt-2024", "title": "Toward Multi-Task Weather Foundation Models", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "arXiv", "url": "https://arxiv.org/abs/2412.07823", "summary": "Multi-task weather model combining nowcasting, medium-range, and seasonal", "candidate_bill": "Bill_11", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.65, "watchlist_tier": null, "notes": "Closer to universal physics scientist - weather domain only", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "weatherbench-2-2024", "title": "WeatherBench 2: A Benchmark for the Next Generation of Data-Driven Global Weather Models", "authors": [ "Rasp S", "Hoyer S", "Merose A", "et al" ], "affiliations": [], "country_region": null, "date": "2024-02-01", "venue": "JAMES", "url": "https://doi.org/10.1029/2023MS004019", "summary": "Standardized weather model benchmark with classical baselines", "candidate_bill": "Bill_5", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.88, "watchlist_tier": null, "notes": "Cross-method consensus piece", "_appeared_in_sweeps": [ "sweep_1205" ] }, { "paper_id": "wet-lab-claims-2024", "title": "Are Autonomous Labs Really Autonomous?", "authors": [ "academic" ], "affiliations": [], "country_region": null, "date": "2024-12-01", "venue": "Nature Methods", "url": "https://doi.org/10.1038/s41592-024-02567-x", "summary": "Critical review of autonomy claims in self-driving labs", "candidate_bill": "Bill_13", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.82, "watchlist_tier": null, "notes": "Autonomy decomposition audit", "_appeared_in_sweeps": [ "sweep_1206" ] }, { "paper_id": "wytteveen-rebuttal-2024", "title": "GNoME, A-Lab and the State of AI Materials Discovery", "authors": [ "Wytteveen academic group" ], "affiliations": [], "country_region": null, "date": "2024-06-01", "venue": "ChemRxiv", "url": "https://chemrxiv.org/engage/chemrxiv/article-details/65d6e0c0a7a4a83e6f3a2bb1", "summary": "Joint critique of GNoME and A-Lab claims", "candidate_bill": "Bill_4", "candidate_meta_cost": null, "verdict": "rebuttal_paper", "confidence": 0.88, "watchlist_tier": null, "notes": "Combined rebuttal", "_appeared_in_sweeps": [ "sweep_1203" ] } ]