Loading activity…
@soniajoseph_
co-founder of World Mechanics @world_mech_ai | prev world models / JEPA, interpretability @AIatMeta @Mila_Quebec @MATSprogram @Princeton
AI
The Information's Oct. 19 episode features World Mechanics CEO Sonia Joseph and Rocket Drew discussing interpretability.
Technology
The Information's episode considers whether interpretability could improve training and why language alone may not be enough for physical AI.
AI
A post announces a first blog post on interpretability, describing it as critical to scaling physical intelligence.
AI
A co-founder says the lab wants to make interpretability an objective during training, rather than analyze models only afterward.
AI
A guest says @climatepaige’s latest podcast discusses how billions of neurons in AIs and brains learn to make sense of the world.
AI
Reddy said in an X post that his role will focus on safe, capable, open LLMs and building ties between Mistral, McGill, Mila, and the wider Canadian AI ecosystem.
AI
A user says they spent three of their last five Saturday mornings on email and still had more than 90 unread messages—all individual or specific, not spam.
AI
World Mechanics says the SF Bay Area roles will help shape a research agenda that makes interpretability and safety priorities from the earliest stages of pretraining.
