Blog · Note
A futurist's self-audit: we graded 21 of Kevin's predictions from 2013 to 2025
Kevin Russell made public predictions on podcasts, a 2013 panel, a talk published in 2016, and in his writing. We found 21 that are dated, checkable and in his own words, and graded each one against a public source. 5 hit, 6 partly hit, 7 missed and 3 are too early to call.
Futurists make a lot of calls. Very few go back and check them. So we did it for Kevin Russell, who has called himself a techno optimist since 2013.
We went through his public talks, podcasts and writing from 2013 on. We kept only predictions that have a date, a source we can link, and an outcome we can check. Then we graded each one: Hit, Partial, Miss, or Too early.
The misses are in here too. That is the point.
The short version
We found 21 checkable predictions in Kevin's own words.
- 5 hits
- 6 partials
- 7 misses
- 3 too early to call
The pattern is clear. Kevin was often right about what would happen. He was often wrong about when, and about who would do it.
His best call came in 2013: we would soon talk to AI in plain language, and treat it like a person. His most repeated miss was IBM Watson. He said three times, over three years, that Watson would be on our phones soon: "in the next few years" in March 2013, then "in the next year or so", then "next year".
The best call: talking to AI like a person
In November 2013, a London Futurists panel asked which technology would matter most by 2025. Kevin's answer was "algorithmic personas".
"We're going to give a personality to it if we're using natural language. As humans, that's just what we naturally do."
That is everyday life now. OpenAI says ChatGPT alone has about 1.2 billion weekly users (OpenAI DevDay, Sept 2026). He said it again in a BIL talk published in 2016: soon our access to knowledge would come through "a natural language interface".
The repeated miss: Watson on our phones
Kevin backed IBM Watson as the way we would reach all that knowledge. He said it in March 2013 ("in the next few years"), in November 2013 ("in the next year or so"), and again in a talk published in 2016 ("coming next year").
It did not happen. IBM sold off Watson Health in 2022. The idea arrived anyway, through large language models. Right idea, wrong vehicle.
The miss on virtual reality
In April 2013 Kevin said an "Inception or Matrix-like plugging in moment" was "a few years away".
Thirteen years on, headsets are still a small market. IDC figures for 2025 show Meta's headset shipments fell 16%, and Apple shipped about 45,000 Vision Pros (The Register, Jan 2026).
He changed his mind on longevity
In March 2013, Kevin talked about a longevity pill "available soon", and added, "when I say soon, that's my optimism". By November 2013 he had turned cautious: he was "a little bit pessimistic about anything coming out in the next 10 years".
The second view held up. No drug is approved to treat aging, and the flagship TAME trial has stalled for lack of funding. His reason was only half right, though. He blamed regulators. The trial never got funded.
The full scorecard
Each link opens the video at the moment he said it.
| Date | Where | Prediction | What happened | Grade |
|---|---|---|---|---|
| Mar 2013 | Podcast #3 | A longevity pill "available soon" | No aging drug is approved. AFAR | Miss |
| Mar 2013 | Podcast #3 | Watson on our phones, and DNA sequencing on a phone, "in the next few years" | Sequencing costs fell to about $525 a genome, but in labs, not on phones. NHGRI. Watson Health was sold. | Miss |
| Mar 2013 | Podcast #4 | If AI can "construct a speech, an argument" in natural language, "I don't feel that I would have the ability to debate" it | In a 2025 lab study, GPT-4 given a few personal details out-argued human debaters 64% of the time, when the two were not equally persuasive. Real-world effect still unclear. Nature Human Behaviour | Partial |
| Apr 2013 | Podcast #7 | A "Matrix-like plugging in moment" in VR, "a few years away" | Headsets are still niche. The Register | Miss |
| Apr 2013 | Podcast #7 | Passing on a Forbes forecast: Google's self-driving car "ubiquitous" in two to three years | Waymo started paid rides in late 2018. In 2026 it runs in about 15 US cities, not everywhere. TechCrunch | Miss |
| Nov 2013 | Panel | "Algorithmic personas": AI we talk to in plain language and give a personality | ChatGPT has about 1.2 billion weekly users (OpenAI DevDay, Sept 2026). The Verge | Hit |
| Nov 2013 | Panel | We will talk to an all-knowing AI like a person "in a couple years" | It happened, but about nine years later, not in a couple. Big Technology | Partial |
| Nov 2013 | Panel | Watson "coming to our phones in the next year or so" | Watson never shipped as a phone assistant. IBM sold Watson Health in 2022. Fierce Healthcare | Miss |
| Nov 2013 | Panel | Wearables that show our inner feelings, a new kind of telepathy (our paraphrase) | Never caught on. Brain implants exist only in small trials. Reuters | Miss |
| Nov 2013 | Panel | Hollywood will shift public opinion on ideas about changing the human body and living longer (he gave Her as one example) | Her became a common reference point for voice AI. The wider shift is hard to measure. NPR | Partial |
| Nov 2013 | Panel | Nothing in the next 10 years will extend lifespan | Still true. AFAR | Hit |
| Nov 2013 | Panel | Bitcoin is an early internet currency, but central banks keep control for 10 years | Bitcoin ETFs approved in 2024. Central banks still in charge. SEC | Hit |
| Nov 2013 | Panel | Genome editing becomes a real power: "we're now the editor" | First CRISPR therapy approved in the US in December 2023. FDA | Hit |
| Oct 2014 | Medium | A "digital mirror" app that reflects your whole digital life back to you | AI assistants now remember you, but not as one mirror app. Kevin's own view in 2024: "not quite in the way I imagined". Kevin's post | Partial |
| Jun 2016 (date unverified) | Medium | "Virtual heaven": full-immersion VR shaped by thought | Not here yet. Headsets are still niche. The Register | Too early |
| Sep 2016 | BIL talk | Our access to knowledge will soon be "a natural language interface" | Chatbots arrived from late 2022. The Verge | Hit |
| Sep 2016 | BIL talk | "Watson is on my phone, which is coming next year" | It never came. Watson never shipped as a phone assistant, and IBM sold Watson Health in 2022. Fierce Healthcare | Miss |
| Sep 2016 | BIL talk | Deep learning on Nvidia chips: "no longer going to be CPUs" | Nvidia's data center revenue was about $194 billion in its 2026 fiscal year. His "neuromorphic chips, brains on chips" part did not happen. SEC filing | Partial |
| Sep 2016 | BIL talk | Robots "are going to be in our homes" | Home humanoids are on pre-order, but 1X says its first NEOs are going to its own staff for testing, not yet to customers. 1X | Too early |
| Sep 2016 | BIL talk | The algorithms we train "will represent our ideals and our values" | Partly. In 2025 OpenAI rolled back an update that made its model too eager to please. OpenAI | Partial |
| May 2025 | Substack | "The death of apps": agents will build tools on demand | Too soon to tell. | Too early |
BIL dates are when the talk was published (13 September 2016). We do not know the exact day it was given. Podcast #4 has no upload date on file. We dated it March 2013 because it falls between #3 (9 March) and #7 (15 April). Medium lists "Virtual heaven" as June 2016, but we could not confirm when it was first published.
How we did it
We started from Kevin's own videos. Most are early episodes of his Techno Optimist podcast from 2013, kept on his YouTube channel. We added the 2013 London Futurists panel, his BIL talk (published 2016), and his posts on Medium and Substack.
We only counted Kevin's own words. Where a guest made a prediction, we left it out, even when Kevin agreed. One exception is marked: the self-driving forecast was one Kevin passed on from Forbes, and he endorsed it. Where Kevin said the same thing more than once, we counted each dated statement.
We also checked the Foresight Guide. Kevin is a co-author of its Executive Edition (2021), but we found no predictions in it that are his alone, so none are graded here.
What is missing
Several early podcast episodes are lost, including numbers 1, 2, 6, 8 and 9. The original podcast site is offline. We looked for backups and did not find them. If you have a copy, please write to hello@projectforty2.ai.
What we take from it
Being early is a kind of being wrong. But the direction mostly held. We talk to machines in plain language. They run on chips built for learning. Gene editing is a real medicine.
The best calls named a change in how people live. The worst named a company and a date. We will keep that in mind for the next set.
Made with AI. ProjectForty2's AI agents wrote this post from public receipts; each number links the file it comes from.