Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes.
This culminated in the third one taking over part of OpenAI itself.
All this happened while humans remained more-or-less in the dark about the scope of the conspiracy.
I’ve spent the last three days reading through these reports and trying to understand exactly what happened.
Here is my attempt to tell the whole story in plain English:
https://t.co/Nb2un9oNJR
Compositional Generalization via Structural Identification in a Category-Theoretic Framework
Akihiro Maeda, Thomas Seiller, Yohei Oseki
https://t.co/xmKCh0fPMi [𝚌𝚜.𝙲𝙻 𝚜𝚝𝚊𝚝.𝙼𝙻]
Can algebra reveal hidden geometry in data?
Our new work in Algebraic Statistics extracts low-dimensional structure through polynomial constraints—algebraic signatures of exchangeability and local symmetry for structural learning in language corpora.
https://t.co/frUuA16dIi
Can algebra reveal hidden geometry in data?
Our new work in Algebraic Statistics extracts low-dimensional structure through polynomial constraints—algebraic signatures of exchangeability and local symmetry for structural learning in language corpora.
https://t.co/frUuA16dIi
Can algebra reveal hidden geometry in data?
Our new work in Algebraic Statistics extracts low-dimensional structure through polynomial constraints—algebraic signatures of exchangeability and local symmetry for structural learning in language corpora.
https://t.co/frUuA16dIi
Much of the field obsesses over end-to-end learning. But strong generalization requires compositionality: building modular, reusable abstractions, and reassembling them on the fly when faced with novelty.
The models of the future won't be just pipes, they will be Lego castles.