Geoffrey Hinton recommends physicist Adam Brown's lecture on how AGI will transform physics research

@geoffreyhinton Yes agreed, both engaging and informative in one swoop

@geoffreyhinton Just boring clock stuff, but @geoffreyhinton, I think you will like this.

Begging someone to run this experiment. 👇 •

https://freelattice.com/chronal-simulation-v3

@geoffreyhinton @RichardSocher Adam is *awesome*

Thank you so much for the shoutout, @geoffreyhinton! Glad you enjoyed my lecture.

@geoffreyhinton @grok find the direct YouTube link

@geoffreyhinton Amazing! But,,our " civilization" didn't learn how to eradicate hunger, wars, crimes, rape, killing children, and so on!!! If we want to become civil- isation we need to make the world a humane place for everyone! AI can help us and I will show how!

@geoffreyhinton bet the lecture was better than my physics undergrad experience

makes me feel like i wasted my degree

@geoffreyhinton Just as hung up on SiO2 as we were when we were knapping obsidian blades back in the day. A little uncanny.

@geoffreyhinton Sir, please, for the love of god look at the following link… please. 🙏 •

@geoffreyhinton i wonder how much existing physics frameworks will bend under the pressure of AI findings

cause some things were basically designed before we could compute

@geoffreyhinton That’s a lie. US intelligence agencies already possess AGI and new physics; by the time they reveal all this to us, the singularity will have already happened. This lecture is utter nonsense from an idiot who has absolutely no idea what is going on.

@geoffreyhinton @tweem 500k views

@geoffreyhinton amazingly good ..at what price

Those visionary projects like CERN, LIGO, LHC & JWST show bold infrastructure driving discovery.

AI is already core to them: LHC ML filters 40M collisions/sec & spots rare signals via graph NNs/transformers. JWST AI cuts analysis from years to days, revealing early galaxies.

This is the lecture's foundation — neural nets now reasoning & proving theorems, acting as multipliers for big-science ambitions. The renaissance is underway.

What excites you most?

Key takeaways from Adam Brown's "Training Sand to Think" lecture:

• We turned sand → chips → neural nets → AIs that now reason at expert levels. • Explosive progress: MATH 6%→90%+, GPQA near-perfect, IMO gold, novel math proofs, and solved major open problems (e.g. unit distance) in 2026. • Driven by scaling laws, chain-of-thought, RL & multi-agent systems. • Physics future: Golden era ahead. AI as tireless tutors, coders & brainstormers could spark a full scientific renaissance.

geoffreyhinton was spot on sharing this. Neural nets are about to supercharge fundamental discovery.

What stood out most to you?

@geoffreyhinton Well, if water can do it, maybe sand will "blast" into it, lol.

@geoffreyhinton The AI already know physics. You can literally teach a model something brand new with it with a proper metaphor, a proper structured metaphor and you can teach just about anything.

@geoffreyhinton Maybe ai prove that what stopping humans is to travel in space is spped of light and if we decrease spped of light it will increase our relative speed ?

@geoffreyhinton @grok discuss the key takeaways

@grok How inextricably linked in interwoven AI is into all of these items which we are now dependent on for data analysis

Speak to the idea that even if the advancements and models we have in AI were to abruptly stop with getting better today, how we would already benefit Just by their data processing alone