HUGE RESULT FOR FORMALIZATION! in only 6h, using relatively modest compute capacity, the Grothendieck-Riemann-Roch theorem was fully verified in Lean by orchestrating a couple hundreds of agents. this is one of the main results of Alexander Grothendieck, and arguably the best result in term of a single theorem. the entire codebase is made open source, but also the processes by which the verification was done. this provides not only a result, but the reasoning behind it.
this was made possible in such a short time because a lot of the components needed from algebraic geometry were made public by @axiommathai and in this spirit, i highly encourage these kinds of initiatives of open-sourcing the basic modules of the core fields of maths: algebra, geometry, number theory, and others. this will be useful not only for the formalization of known results, but also in enabling the inference of new results which can be verified fast enough.
huge progress! that's the biggest formalization work accomplished through grothendieck so far and it's the wonderful Grothendieck-Riemann-Roch theorem!
the full work took around 6h.
we make public not only the finish result, but the process by which it was done. going forward, we will opt for increasingly more transparency in sharing more and more details about the making of our results.
this paves the way to a much more structured approach of our goal, which is to build your interface to higher intelligence.
this is well beyond what i have compute to do right now, but it would be cool if someone formalized the Atiyah-Singer Index theorem, and make available the modules on elliptic (pseudodifferential) operators on manifolds, topological K-theory, Bott periodicity, Thom isomorphism, de Rham cohomology of smooth manifolds, Chern-Weil theory, ...
mathematics has never been more alive! the top 3 anthropic models formally verified the Riemann mapping theorem. what matters most is not just that they did it well, but that they can explain their thought process, where the failed, their assumptions and how they actually did it.
enjoy the nice explainer!
this is one of the most fundamental convictions behind everything im currently doing. take a proof, it's just a logical conclusion given some initial axioms. today, we have to find them and put a lot of them together in a certain way to define some theory. the theory itself evolves as if it's filling some kind of conceptual shape. very soon, theories will be self-refining, self-redefining to serve particular conceptual objectives.
yesterday, we managed to get opus 5.5 verify formally a proof of the Brouwer's fixed-point theorem. while this is a simple task, it begs the question of what this looks at scale. of course we can point it towards much more complicated questions, but what happens when we formally verify entire fields of mathematics and completely solve all the open conjectures in certain fields in thousands and millions of pages of proofs, incomprehensible by humans?
the biggest bottleneck is not going to be getting the verification right. the real issue will be how do we get to interface with what super-intelligence is doing.
the future of the production of and the interaction with knowledge is going to be amazing. grothendieck will be your interface to higher intelligence.
in the next 6 months, this will be the best tool you use to get higher bandwidth with (super)intelligence. i’m heading to Singapore soon to get it up and running. i’m very excited! ai shouldn’t handle all of the civilization’s intellectual work without us being able to understanding it. my goal is to reduce the gap between what AI can do and the speed at which we can understand it.
dear algorithm: please show my feed to people who are tired of reading AI slop and asking the question of how they are going to interface with higher intelligences.
Listening to "Introduction", the introduction episode to what should be a long road to making education a perfect configuration at https://t.co/7umV3gtJa9
We strongly believe that the way knowledge is transmitted and organized today effectively correlates with what the future will look like, so we make it our mission to structure and optimize it.
Et pourquoi l'éducation ne peut pas être personnalisée ? Pourquoi étudier doit être si pénible ? Pourquoi l'histoire doit encore se répéter ? On a des moyens encore plus que jamais de remédier à cela. Nous vivons l'ère de l'intelligence artificielle quand-même. #insighteducati
Nous croyons que la transformation de l'éducation représente la prochaine étape de l'application de l'intelligence artificielle. Une éducation décentralisée, accessible de partout, intelligente et dotée de puissantes métriques aussi qualitatives que quantitatives.