
Philosopher @GoogleDeepMind · Assoc Prof @Cambridge_Uni @LeverhulmeCFI | Consciousness, Machine Minds, AGI, Human-AI Relationships | Views my own | 🦐🌱
Cambridge, England
Joined November 2008
- Tweets20.1K
- Following10K
- Followers39.5K
- Likes143K
Pinned Tweet
Okay, here it is, the Bad AI Consciousness Takes Bingo blog post, with rebuttals for each and every square and some additional honourable mentions!
(I'd like to thank @united for making this post possible, and for the extra pretzels)
open.substack.com/pub/polytr…
Henry Shevlin retweeted
Like most mathematicians, I believed the Riemann Hypothesis to be completely out of reach at present (even with the best AI tools), because I expected it to require substantial theory-building, probably in arithmetic geometry. The kind of new theory-building that fills entire books, and that AI isn't so good at yet.
Mathematicians have a notion of how deep a problem is. A sufficiently deep problem is generally intractable until enough of the theory is in place. If difficult problems are like valuable mineral reserves buried miles underground, you first need to build the infrastructure to reach them.
What Levent's proof seems to have done instead is something completely different. It takes an incredibly complicated tour through the infrastructure built for a hundred other mineral reserves, situated far from the destination, and somehow create a wormhole to the original destination.
This is not theory-building in the usual sense. It is something much weirder and more alien. The best human mathematicians know how to build theory, painstakingly, to reach a goal. But no human brain, on its own, is capable of creating a wormhole that requires a 25,000 page long tour through existing infrastructure with no obvious or realistic connection to the goal. But huge credit to Levent here: I think no current AI, on its own, is capable of this either.
I had never worked on or thought too much about the Riemann Hypothesis: it seemed very far removed from my own area of number theory.
But the astonishing thing about Levent's claimed proof of RH (which I think has a very good chance of being correct) is that it uses a huge number of precisely the things that are central to my own corner of number theory:
- Theta functions and other automorphic forms on unitary groups
- Gan–Gross–Prasad period formulas
- Relative trace formulas
Etc., etc.
A problem that I had always assumed would require new theories in arithmetic geometry may instead turn out to be tractable through an inhuman chain of existing (deep) results in higher-rank automorphic forms (and a lot else).
Henry Shevlin retweeted
Good morning, human civilization was a 10,000 year sociological study conducted by the University of ☡ ☲ ⚟ ⚟ ⛮ to derive a solution to the Riemann Hypothesis. We are now complete with our study. Your universe will shortly be terminated. Thank you for your time and effort.
Henry Shevlin retweeted
"What happens after AGI?"
institute.deepmind.com/essay…
A great essay by Tim Genewein -- it concisely summarizes our paper "From AGI to ASI", in case you weren't brave enough to read the full 61-page report :)
Signal boosting folks who made good calls on this
Henry Shevlin retweeted
Riemann Hypothesis by the end of 2026.
If we don't solve it by 2027, we will look back at it as a "we didn't point the model at the problem the right way" situation, not as a problem of needing more AI training breakthroughs
Post-May developments have *not* changed my timeline
Henry Shevlin retweeted
An LLM will prove the Riemann hypothesis before a world model can generate better games than the Unreal Engine.
Henry Shevlin retweeted
BREAKING: A mathematician at Anthropic claims to have used the company's AI to solve the Riemann hypothesis, the most notorious open problem in all of math
sciam.com/article/anthropic-…
Related: when I started in AI academia, I'd often meet academics who knew far more about AI than anyone in business outside tech. Now I routinely meet bankers, brokers marketers etc with sophisticated agentic workflows. AI academics need to be serious about using the tools!
We are living in a funny time where plots like this around citation hallucinations by LLM feel super obvious to anyone using AI tools regularly, yet still surprise many scientists whose view of AI is shaped by a bad experience they or a colleague had in the past.
There is also not enough forward thinking in the field. Whenever we encounter a limitation or bug, we should always ask how likely it is to still be there in 6 months or a year.
AI today is as bad as it will ever be.
Henry Shevlin retweeted
I agree with Scott Armstrong here. I think there is still a lot of cope going around, especially among those who subscribe to some version of the (largely incoherent) "convex hull" theory of what counts as a new idea.
As I was explaining to a notorious philosopher of biology yesterday, the distinction between a genuinely new idea and a recombination of old ones is ultimately arbitrary and subjective, and says more about your own judgment and tastes than about the idea itself. More importantly, however you draw such a line, LLMs have clearly crossed it.
And I can guarantee that, without LLMs, no human would have proved the Navier–Stokes blowup, the quasi-Riemann hypothesis, the irrationality exponent of pi being 2, or the Unique Games conjecture in the coming months or years.
Henry Shevlin retweeted
gta 6 is insane. you can even start a b2b saas company where you turn unstructured data into actionable insights
Henry Shevlin retweeted
I think we now have to accept that the crazy AI people (I use the term affectionately) are, at least, directionally correct.
- There is at least a small chance that recursive self-improvement (RSI) in the next couple of years leads quickly to “superintelligence”, which in turn leads to extremely rapid transformative effects, including a possibility of total disaster.
- Even if, like me, you are very sceptical of the most aggressive forecasts, they are directionally correct. You can quibble with the timelines of RSI, with the precise meaning of “superintelligence”, and with the pace of transformation and diffusion, but the basic story is right.
Short of political interventions, we will build systems with capabilities superior to ours across all domains within the next couple of decades. The limitations of current systems look solvable, progress has been shockingly fast, and the resources being thrown at AI are enormous. This process will likely be greatly accelerated once AI systems outcompete us at designing better AI systems, and it will lead to a speed and scale of transformation, as well as novel catastrophic risks, that almost no mainstream elites or normies are pricing into their models of the world.
That, at any rate, is what I now believe with a high degree of confidence.
Henry Shevlin retweeted
Curious about AI, superintelligence and consciousness? We can recommend this excellent series of podcasts by former Corpus JRF @danwilliamsphil and @dioscuri
The last five podcasts of my AI podcast with @dioscuri:
Is AI a technology, maybe even a normal technology, or is it something radically different? Henry and I discuss this with Arvind Narayanan (@random_walker), co-author of "AI as Normal Technology" and one of the most important voices writing about AI today. (1/5) youtube.com/watch?v=7oHsQuuk…
Henry Shevlin retweeted
The last five podcasts of my AI podcast with @dioscuri:
Is AI a technology, maybe even a normal technology, or is it something radically different? Henry and I discuss this with Arvind Narayanan (@random_walker), co-author of "AI as Normal Technology" and one of the most important voices writing about AI today. (1/5) youtube.com/watch?v=7oHsQuuk…























