Press Space to continue
Every sample at @periodiclabs sits in a circular holder, tracked
Press Space to continue
Every sample at @periodiclabs sits in a circular holder, tracked by position. One day, someone made a small mistake: the whole set got shifted by two, counterclockwise. Each sample was now mislabeled, and nobody knew why. A human researcher could burn hours retracing a mistake like that. Instead, the LLM diagnosed it on its own, worked out the rotation, and relabeled every sample correctly. A human could have solved it too, co-founder @ekindogus says, but it would have taken hours of…
Mario Gabriele@mariogabrieleSep 29
"A system that can literally engineer matter is going to be of key importance to everybody." @LiamFedus and @ekindogus are the co-founders of @periodiclabs, a San Francisco-based startup building AI systems and autonomous laboratories to discover new materials, starting with a higher-temperature superconductor. Fedus previously helped build ChatGPT and ran post-training at OpenAI; Cubuk led materials and chemistry research at Google DeepMind. (3:14) What it takes to build a synthetic superintelligence (16:44) Dogus and Liam's path to AI (23:28) Lessons from launching ChatGPT (35:10) What AI learns across experiments (49:23) The difference in LLM performance on math vs. science (54:09) The competitive landscape (55:39) Final meditations
View