World Mechanics builds founding team to make physical AI interpretable from day one
soniajoseph_ · x · 2026-09-15
- Sonia Joseph announced the founding research team at World Mechanics, a frontier neolab building interpretable models of the physical world.
- Thesis: the physical world offers ground truth that language lacks, enabling interpretability and pretraining methods that capture causal physical dynamics—making interpretability and safety first-class objectives from the earliest pretraining stages.
- Target domains: robotics, autonomous driving, manufacturing, and scientific discovery.
- Hiring two Founding Research Scientists in interpretability, training dynamics, and physical AI, based in the SF Bay Area.
More from Companies & People
- Cohere CEO Aidan Gomez: Terminator-style AI takeover talk shouldn't enter public debate — cohere · 2026-09-15
- Observer: Alpha must control its narrative or its ambitions collapse — benjaminjriley · 2026-09-15
- OpenAI engineers: kernel optimization cut GPT-5.6 Sol serving cost by 20% — TheTuringPost · 2026-09-15
- Companies rolling out Claude to everyone get 'just a chat subscription' — missing harness and evals, not weak models — alex_verem · 2026-09-15
- Dario: Anthropic controlling AGI would be like Hitler getting the atomic bomb first — teortaxesTex · 2026-09-15
- OpenAI President Brockman: The Most Important AI Stories Are the Ones Nobody's Telling — a16z · 2026-09-15