cp

Human & the Loop

[cp @humanandtheloop]

My categorical imperative for using AI:

Use it always in such a way that the result of your work makes you proud to have accomplished it.

[cp @humanandtheloop]

@hexorer @Nicolas_Colin Policy choices. For ex: Increase bargaining power for highly productive workers. Reduce costs of running businesses. Make scaling easier across Europe (depends a lot on national tax systems).

[cp @humanandtheloop]

This is exactly what happened at Aleph Alpha before they were sold.

[Thibault Schrepel @ProfSchrepel]

GREAT piece of journalism. I sum up:
- @MistralAI is moving away from the desire to be a frontier lab
- The company is now focusing on deployment within big companies (including the deployment of Chinese models).
- These companies CEOs find Mistral strategy hard to follow… which they don’t like (makes it for an uncertain business partner).

https://t.co/sv6noJt4LM

[cp @humanandtheloop]

@nicolenotdunn and lastly - it may cross regulatory boundaries which are yet to fully hit the token economy (export restrictions, KYC etc)

[cp @humanandtheloop]

@nicolenotdunn I agree this is surprisingly simplistic. Enterprises are actually very unhappy with token overspend (undermining the premise); "spending" tokens is in most cases to a single "trusted" counterparty (AI Lab) whereas agentic commerce is likely involves "untrusted" 3rd parties;

[cp @humanandtheloop]

@hormetic @QiaochuYuan @tak3sh8 Turing's Cathedral is great!! The issue with places like IAS is not simply embracing computation, it's that all of institutional math is built around the bottleneck of proof which is now disappearing.

[cp @humanandtheloop]

@QiaochuYuan @tak3sh8 absolutely the case. two events epitomise this: 1) the IAS banishing the computer after JvN's death 2) the math world ignoring NN basically throughout the 2010s bc you couldn't (still can't) prove much about their general properties.

[cp @humanandtheloop]

100%

[Christian Szegedy @ChrSzegedy]

If I did not believe in the intrinsic and underrated usefulness of mathematics, I would be somewhat sad about what AI is doing to it.

I think Jevons paradox will strike again: much cheaper and deeper mathematics will become even more of the fundamental infrastructure of the future than it is today.

Math will no longer be centered on conjectures posed by humans. It will be continually postulated, solved, processed, and utilized by AI systems as they work on the hard scientific, engineering, and computing problems of the future.

[cp @humanandtheloop]

@lugaricano @boazbaraktcs @asving94 @scottnarmstrong @henryquantum @robinhouston

[cp @humanandtheloop]

When mathematicians move the goalposts of what can be imagined — our world changes with it. AI systems solving Fermat and Navier–Stokes promise to put the tools for playing with abstraction in everyone's hands. https://t.co/y04f9TGPtO

[cp @humanandtheloop]

@Nicolas_Colin @robin_j_brooks The required discourse for these policy changes isn't happening at all. Although the slogan would be pretty obvious, and actually support a broad coalition: Make Germany attractive for foreign investment & raise German wages.

[cp @humanandtheloop]

When this happened last time, it started with non euclidean geometry and we ended up with space time geometry, it started with infinity and we got the computer and the nuclear bomb.

[cp @humanandtheloop]

First Fermat, then Navier Stokes. Setting aside the personal politics: We have created a tool which has the ability to plumb the deepest depths of problems that humans can fathom.

[cp @humanandtheloop]

We are living through an era, that will lead to a complete rethink of how humans relate to the abstractions with which they see the world.

[cp @humanandtheloop]

Irreverent joke: the dirty secret about navier stokes is that everyone thinks people in regularity theory for nonlinear pdes are the math version of accountants.

[Konsti Wohlwend @konstiwohlwend]

drama summary for those confused:

- Aug 15th: Tristan Buckmaster & Levent Alpöge make progress on a few important math problems

- they do NOT have a proof for the $1,000,000 Millenium Prize problem. BUT, they do claim to have a proof for a similar (non-Millenium) Navier Stokes problem that could help lead the way there

- Levent works at Anthropic, but this research was independent of his work there, with a mix of GPT and Claude models. Tristan is not related to Anthropic.

- Early Sep: Rumor spreads to OpenAI that Anthropic has solved a major problem. Tristan emails OpenAI to clarify. without revealing the problem they solved or how they did it.

- After hearing of the rumor, OpenAI started researching Navier Stokes with a new internal model.

- Sep 6th: OpenAI's Sebastien Bubeck tells Tristan that they solved the $1,000,000 Millenium Prize Navier Stokes problem. The approach is very similar to Tristan & Levent's approach to the non-Millenium problem.

- Tristan is suspicious of the timing, as only few others were trying this approach. OpenAI says the model didn't access his user data directly, but leaves unanswered whether Tristan's chat conversations were part of the training.

- OpenAI says they would partially credit Tristan for the $1,000,000 discovery (even though Tristan did not solve the $1,000,000 problem) — but only if they remove Levent as an author, as he works for Anthropic.

- Sep 8th: Tristan refuses to remove Levent, and rushes to publish their results independently.

Currently unclear is whether Anthropic had a separate solution for the $1,000,000 problem, or whether the rumor was about Tristan & Levent's independent research.

[cp @humanandtheloop]

@paulg The real impact is through AI diffusing throughout mathematics. Labs' attention moving on would mean that the course of the discipline would depend less on them. But teaching, evaluation, collaboration are likely to be transformed.

[cp @humanandtheloop]

@Scholars_Stage @ginevlily Immanentizing the eschaton. That's never worked as intended.

[cp @humanandtheloop]

@Leonard41111588 @paulg The key insight from Kevin: "The ability to autoformalize hard material will ultimately make the review process for mathematics papers far less painful" - math as a discipline is organised around a disappearing bottleneck

[Leonardo de Moura @Leonard41111588]

I did not expect this. Anthropic just published a Lean proof of Fermat's Last Theorem.
The proof is +13M LoC, more than 5 times the size of Mathlib.

Kevin Buzzard's post: https://t.co/yVJ46QzsZn
The proof: https://t.co/PqAq7qc375
Anthropic's post: https://t.co/d4e1eNqlwQ

[cp @humanandtheloop]

Anthropic published a formalised proof of Fermat's Last Theorem in Lean today.

[cp @humanandtheloop]

What is the biggest box you can build that provably has no exit. Turns out this is a hard problem @wordgrammer

[wordgrammer @wordgrammer]

https://t.co/DEwsU0YzRE