Claims

2,653 claims by 33 public figures and organisations, checked against the public record. Each has its quote, source, verdict and evidence. Read them without a key; browse by person.

28,548 claims · page 24 of 286

28,548 claims, sorted by newest said. Choose a claim to read it in full.
VerdictPersonKindSaidClaimTopic
Mixed Eliezer Yudkowsky Factual 20 Apr 2026 Yudkowsky himself observed the persona-selection-like hypothesis around the time of GPT 3.5. Eliezer Yudkowsky · Factual · 20 Apr 2026 AI safety and alignment
Not checkable Eliezer Yudkowsky Not checkable 20 Apr 2026 I may be less open to the airy, unadorned denial that any tension exists. Eliezer Yudkowsky · Not checkable · 20 Apr 2026
Not checkable Eliezer Yudkowsky Not checkable 20 Apr 2026 What I'm questioning is whether persona selection lets you solve alignment problems by picking an aspect of humanity you like, and durably… Eliezer Yudkowsky · Not checkable · 20 Apr 2026
Not checkable Eliezer Yudkowsky Not checkable 20 Apr 2026 An alien being harshly and strictly trained to exactly imitate a human would not feel like a human about that training or about… Eliezer Yudkowsky · Not checkable · 20 Apr 2026
Not checkable Rodney Brooks Not checkable 20 Apr 2026 Even his own toadies are realizing he can't be trusted to command out troops. Rodney Brooks · Not checkable · 20 Apr 2026
Mostly true Eliezer Yudkowsky Factual 20 Apr 2026 Presumably, at some point during RL training on code-passing tasks, LLMs learn a dispreference for failing tests, independent of how that is avoided or whether the user's real goal is met. Eliezer Yudkowsky · Factual · 20 Apr 2026 AI safety and alignment
Not checkable Eliezer Yudkowsky Not checkable 20 Apr 2026 One should of course be open to other hypotheses to explain these observations Eliezer Yudkowsky · Not checkable · 20 Apr 2026
Not checkable Eliezer Yudkowsky Not checkable 20 Apr 2026 They are not deeply tested nor yet based on interpretability results. Eliezer Yudkowsky · Not checkable · 20 Apr 2026
Not researched Demis Hassabis Factual 16 Apr 2026 Gemini 3.1 Flash TTS, designed to give builders granular control over AI-generated speech, is available in preview from 2026-04-16 via the Gemini API and Google AI Studio and for enterprises on Vertex AI. Demis Hassabis · Factual · 16 Apr 2026 AI capabilities
Not researched Demis Hassabis Factual 16 Apr 2026 Gemini 3.1 Flash TTS is Google's most expressive and steerable text-to-speech model yet. Demis Hassabis · Factual · 16 Apr 2026 AI capabilities
Unresolvable Demis Hassabis Factual 14 Apr 2026 Gemini Robotics-ER 1.6 enables robots like Boston Dynamics' Spot to read complex industrial gauges autonomously. Demis Hassabis · Factual · 14 Apr 2026 AI capabilities
Not checkable Demis Hassabis Not checkable 14 Apr 2026 Great to see our collaboration w/ @BostonDynamics unlocking new capabilities! Demis Hassabis · Not checkable · 14 Apr 2026
Mostly true Rodney Brooks Factual 12 Apr 2026 JD Vance campaigned for Viktor Orban in the Hungarian elections the week before 12 April 2026, and Orban lost the election. Rodney Brooks · Factual · 12 Apr 2026 Other
Resolved true Rodney Brooks Factual 12 Apr 2026 A Putin ally suffered a major defeat in the election after 16 years in power. Rodney Brooks · Factual · 12 Apr 2026 Other
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I would love to see international cooperation amongst the leading labs around these safety issues Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 With computational tools like AlphaFold 2, structural biology can now keep up with genetic sequencing. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Mostly true Demis Hassabis Factual 10 Apr 2026 Deep Blue was world-champion level at chess but could not do anything else. Demis Hassabis · Factual · 10 Apr 2026 AI capabilities
Not checkable Demis Hassabis Not checkable 10 Apr 2026 But we don’t have to wait till AGI arrives to start getting the benefits of AI. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 One is bad actors — whether that’s individuals or all the way up to nation states — repurposing these technologies that we’re trying to… Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 Neuroscience has so far not found any quantum effects in the brain. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not checkable Demis Hassabis Not checkable 10 Apr 2026 Done more things like AlphaFold, maybe cured cancer or something like that. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 Google can detect content made with its watermarked technologies and flag it as fake to users or governments. Demis Hassabis · Factual · 10 Apr 2026 AI safety and alignment
Not checkable Demis Hassabis Not checkable 10 Apr 2026 if you just make that 5% faster, that’s a huge cost saving, given tens of billions being spent on training. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 Demis Hassabis said his organisation had spun out a new company, Isomorphic Labs, that tries to build on AlphaFold to speed up drug discovery. Demis Hassabis · Factual · 10 Apr 2026 Their own work or organisation
Not researched Demis Hassabis Factual 10 Apr 2026 Hassabis said he thinks OpenAI has described ChatGPT's release as a kind of research experiment. Demis Hassabis · Factual · 10 Apr 2026 Their own work or organisation
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I think the way to do that is by using in silico methods Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 And I’ve always thought of AI as potentially the ultimate tool to do that. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 given how important AGI is and how transformative a technology is, maybe the most transformative one in human history Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 you can’t really fully understand your systems until they’re stress tested by millions of people. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I thought the kind of problem it was would be suitable for AI one day Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I think we need these types of ideas back in now with our foundation models Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I don’t actually have any pre-described notion of what the answer should be. I just want to know the answer. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 We can’t even describe something as fundamental as that Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I think a lot more research and a lot more effort needs to go into that from everyone. Demis Hassabis · Not checkable · 10 Apr 2026
Not yet due Demis Hassabis Prediction 10 Apr 2026 Within 50 years, AI-enabled abundance could lead to human flourishing, cured diseases, longer healthier lives and travel to the stars. Demis Hassabis · Prediction · 10 Apr 2026 AI timelines and forecasting
Not checkable Demis Hassabis Not checkable 10 Apr 2026 So I think the two things I just mentioned are the things that maybe the average person is not worrying enough about. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 Physics was Demis Hassabis's favorite subject at school. Demis Hassabis · Factual · 10 Apr 2026 Their own work or organisation
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I would hope that they would say that my life was of benefit and service to humanity. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 although I’m a scientist first and foremost, I’m also a pragmatic engineer. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 So we need to vastly improve that if we want to improve human health, I think. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I think governments should be using AI, and we want to support all sorts of democratically elected governments. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 So it just suddenly hit me, we should just do that. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 And that for me was the signal that we were ready to turn it to scientific problems like AlphaFold. Demis Hassabis · Not checkable · 10 Apr 2026
Unresolvable Demis Hassabis Factual 10 Apr 2026 Knowing a protein's structure is only one small part of the drug discovery process. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not researched Demis Hassabis Factual 10 Apr 2026 Demis Hassabis said that potentially a lot of neuroscientists, including him, think that maybe a good model for the brain is an approximate Turing machine. Demis Hassabis · Factual · 10 Apr 2026 Philosophy, ethics and society
Mostly true Demis Hassabis Factual 10 Apr 2026 Demis Hassabis decided from a very young age to work on AI because he was obsessed with big questions as a schoolchild. Demis Hassabis · Factual · 10 Apr 2026 Their own work or organisation
Not researched Demis Hassabis Factual 10 Apr 2026 Predicting a protein's 3D structure from its amino acid sequence was a 50-year grand challenge. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not checkable Demis Hassabis Not checkable 10 Apr 2026 Yeah, no, I think we are special. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 But I think that’s an empirical question. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 Roughly 6 months to a year after AlphaFold's release, some teams used it with experimental data to work out the structure of what he called a 'gateway protein', which he described as one of the biggest proteins in the body. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not yet due Demis Hassabis Prediction 10 Apr 2026 Science-fiction developments such as asteroid mining become very plausible in the next 50 years. Demis Hassabis · Prediction · 10 Apr 2026 Technology
Not checkable Demis Hassabis Not checkable 10 Apr 2026 AlphaFold is one of the linchpins — that’s the structure of the protein. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 Most protein structure prediction algorithms before AlphaFold were quite slow. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not checkable Demis Hassabis Not checkable 10 Apr 2026 But we still need this ability to search and think and reason on top of those models. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 we try to do that by advancing the frontier, but also trying to be as responsible as we can as we deploy these very powerful technologies… Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Rodney Brooks Not checkable 10 Apr 2026 But, you know, automatic enrollment to vote would be full of fraud Rodney Brooks · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 the downside of it is that we’re in this sort of ferocious commercial pressure race that everyone’s sort of locked into currently. Demis Hassabis · Not checkable · 10 Apr 2026
Unresolvable Demis Hassabis Prediction 10 Apr 2026 There could be unique things, and unique connections between humans, that AI systems will never replicate. Demis Hassabis · Prediction · 10 Apr 2026 Philosophy, ethics and society
Not checkable Demis Hassabis Not checkable 10 Apr 2026 humanity could benefit from the proceeds of that, like cures for cancer, or maybe new energy sources or new materials. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 AlphaZero plays chess ideas that brute-force engines like Stockfish had not discovered. Demis Hassabis · Factual · 10 Apr 2026 AI capabilities
Not researched Demis Hassabis Factual 10 Apr 2026 Hassabis said that "we could still" create many types of AlphaFold-like systems and efforts like Isomorphic Labs while building AGI in a careful scientific way, and added that he is still doing this, describing it as part of what would have been the ideal way for things to play out. Demis Hassabis · Factual · 10 Apr 2026 Their own work or organisation
Not checkable Demis Hassabis Not checkable 10 Apr 2026 AlphaFold was one of those, right? Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 Malaria, Chagas disease and leishmaniasis affect hundreds of millions of people worldwide. Demis Hassabis · Factual · 10 Apr 2026 Biosecurity
Resolved true Demis Hassabis Factual 10 Apr 2026 Over 3 million scientists are now using AlphaFold. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Unresolvable Demis Hassabis Factual 10 Apr 2026 Transformers, invented by Google colleagues, together with some reinforcement learning, proved sufficient to crack language tasks without further major breakthroughs. Demis Hassabis · Factual · 10 Apr 2026 AI capabilities
Not checkable Demis Hassabis Not checkable 10 Apr 2026 And I want to use AI as a tool to help us understand the nature of reality. Demis Hassabis · Not checkable · 10 Apr 2026
Mixed Demis Hassabis Factual 10 Apr 2026 Hassabis got into AI originally because he was interested in big questions such as the nature of reality and consciousness. Demis Hassabis · Factual · 10 Apr 2026 Their own work or organisation
Resolved true Demis Hassabis Factual 10 Apr 2026 Only about 10% of drugs get through all the clinical stages. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not researched Demis Hassabis Factual 10 Apr 2026 AlphaGenome, recently released by Google DeepMind, is the best system in the world for predicting whether single-letter genetic mutations are harmful. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Unresolvable Demis Hassabis Prediction 10 Apr 2026 Hassabis thinks the types of technologies behind the drug programs his team is working on should eventually be able to help across almost every therapeutic area. Demis Hassabis · Prediction · 10 Apr 2026 Science and research methods
Not checkable Demis Hassabis Not checkable 10 Apr 2026 giving them the structures of the proteins that are involved in, say, malaria is a huge boon for them too Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 A lot of the reasons that I got into AI 30+ years ago now is to advance science and medicine. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I thought it would be best to approach these kinds of the sort of latter stages of building it, which we’re in now, using the scientific… Demis Hassabis · Not checkable · 10 Apr 2026
Mostly true Demis Hassabis Factual 10 Apr 2026 A protein's 3D structure is what determines, or partially determines, its function. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not checkable Demis Hassabis Not checkable 10 Apr 2026 it turns out that things like language were a lot easier than we were all expecting. Demis Hassabis · Not checkable · 10 Apr 2026
Not yet due Demis Hassabis Prediction 10 Apr 2026 Within 50 years, humanity could get safely through the AGI moment, with AGI built and helpful to society. Demis Hassabis · Prediction · 10 Apr 2026 AI timelines and forecasting
Not researched Demis Hassabis Factual 10 Apr 2026 AlphaGenome is probably not yet good enough, though a future version might be accurate enough to show the effect of a particular mutation in combination with another one. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not checkable Demis Hassabis Not checkable 10 Apr 2026 Yes, I love games. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 One of the biggest proteins in the body acts as a gateway that opens and closes to let nutrients in and out of the cell nucleus. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not checkable Demis Hassabis Not checkable 10 Apr 2026 obviously AlphaFold was our first and, so far best expression of that, let’s say. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 Google DeepMind works on SynthID, an AI watermarking system. Demis Hassabis · Factual · 10 Apr 2026 AI safety and alignment
Own records only Demis Hassabis Factual 10 Apr 2026 Public users may have access to cutting-edge AI technology perhaps only 3 to 6 months behind what is in the labs. Demis Hassabis · Factual · 10 Apr 2026 AI policy and governance
Own records only Demis Hassabis Factual 10 Apr 2026 Hassabis said he thinks his team is working on about 18 or 19 drug programs across a range of areas, from cardiovascular disease to cancer to immunology. Demis Hassabis · Factual · 10 Apr 2026 Their own work or organisation
Not researched Demis Hassabis Factual 10 Apr 2026 Go has too many possible positions (about 10^170) to be solved by brute force as chess was. Demis Hassabis · Factual · 10 Apr 2026 AI capabilities
Not checkable Demis Hassabis Not checkable 10 Apr 2026 So a kind of combination of things like AlphaGenome and CRISPR could be incredibly powerful. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 It was in the minds of the chess grandmasters and the programmers. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 In Go and chess, about 16 or 17 generations of AlphaZero self-play improvement go from random play to better than world champion. Demis Hassabis · Factual · 10 Apr 2026 AI capabilities
Not checkable Demis Hassabis Not checkable 10 Apr 2026 And then only at the final stage do you check it in the wet lab. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 Computational tools can check very quickly, in a matter of hours or within a few minutes, how a given compound attaches to the other 20,000 proteins in the human body. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not checkable Demis Hassabis Not checkable 10 Apr 2026 one of my favorite series was the Culture series by Iain Banks. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 because it’s unprecedented to create technology like that. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 Google DeepMind's AI applied to Google's data centers saved 30% of the energy used for cooling systems. Demis Hassabis · Factual · 10 Apr 2026 AI compute, hardware and data
Not checkable Demis Hassabis Not checkable 10 Apr 2026 Now, it didn’t happen like that because technology is unpredictable. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 On average it takes about 10 years to develop a drug. Demis Hassabis · Factual · 10 Apr 2026 Science and research methods
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I’m quite sanguine about whatever the answer might be. Demis Hassabis · Not checkable · 10 Apr 2026
Not researched Demis Hassabis Factual 10 Apr 2026 The main cost of space travel, including SpaceX launches, is still rocket fuel, i.e. energy cost. Demis Hassabis · Factual · 10 Apr 2026 Technology
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I think they will be the biggest issues that we’re going to have to contend with if we’re going to get through the AGI moment in a way… Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 I would advocate that all companies working on generative AI should build in something, some kind of technology like that. Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 But I think that still pales, as a small issue, compared to some of these bigger issues around AGI itself becoming very capable Demis Hassabis · Not checkable · 10 Apr 2026
Not checkable Demis Hassabis Not checkable 10 Apr 2026 It’s not the way I dreamed about years ago, where we would be sort of contemplating this philosophically and carefully considering each… Demis Hassabis · Not checkable · 10 Apr 2026
Back to the list

Demis Hassabis

Open its page

Factual AI capabilities

“Also, it’s playing interesting new chess that even chess computers like Stockfish — the more expert system brute force ones — haven’t discovered.”

Demis Hassabis, CEO of Google DeepMind 10 April 2026 HUGE Conversations w/ Google DeepMind CEO Demis Hassabis (Transcript) (The Singju Post)

AlphaZero plays chess ideas that brute-force engines like Stockfish had not discovered.

Not researched Checkable, but outside the sample that was researched.

Said for an organisation they lead or work for; it counts on both records.

Cite this claim

Demis Hassabis, “AlphaZero plays chess ideas that brute-force engines like Stockfish had not discovered.” (said 10 April 2026). Verdict: not researched. Reliability API, claim c_65ff8aa9bb06, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_65ff8aa9bb06

Verdicts are made automatically; how verdicts and scores are made. The same data is on Hugging Face under CC BY 4.0.