AE Studio
@AEStudioLA
Frontier AI Alignment Research and High-Stakes Production AI
We're hiring:
108 Following 3.4K Followers
As I've said before, there is a Pointless War between Anthropic and everyone else
In the @WSJ today, I outline a solution:
AI Alignment R&D solves risks & helps us win, including on open source
🧵
Show more
Our SCML 2026 paper, Cluster-Weighted EDMD, jointly learns a soft partition of the state space and local Koopman operators to capture dynamics that vary across regions.
What interests me more broadly about Koopman theory is the possibility that complex dynamics may admit simpler descriptions through the right observables and spectral modes.
I am a research lead at Fractal, @AEStudioLA’s skunkworks research division focused on neglected approaches in AI and alignment. We are interested in exploring whether Koopman-inspired and related operator-theoretic methods can be useful for studying language and reasoning.
If you are working on related ideas, let’s talk.
Paper:
Show more
We’re pleased to have collaborated with AE Studio on this research.
Read more here:
New research! Some AI capabilities are both helpful and dangerous. E.g., knowledge of virology can be used to create life-saving vaccines or deadly pathogens. We introduce GRAM, a training method that puts dual-use capabilities (like virology) into removable modules.
Show more
Exactly what we experience @AEStudioLA building AI with big companies, usually very afraid at first:
"The firms with the most to gain from the rebuild are the ones whose current economics make the rebuild most painful.
They wait, and the waiting is rational until it is fatal."
Show more
Are you concerned that AE Studio has been too focused on AI alignment?
Then we have the perfect job for you!!:
Role: Pirate
This is a "neglected approach" to "earn to give."
Any candidate recommendations are much appreciated:
Show more
