Gabriel Poesia Reis e Silva
Kempner Research Fellow
Contact Information
Areas I Research:
About
Gabriel Poesia is a Research Fellow at the Kempner Institute. Before Kempner, he received a Ph.D. in Computer Science from Stanford University, working under Noah Goodman. Gabriel grew up in Brazil, in Belo Horizonte (“Beautiful Horizon”). Outside of research, he enjoys running, making and listening to music, reading, cooking, hiking and fermenting.
Research Focus
Gabriel’s research is centered around building self-improving machines that are capable of formal reasoning, which includes proving mathematical theorems, conjecturing, decomposing problems, and developing increasingly higher-level abstractions over time. This goal has involved interfacing ideas from type theory (to define a game of formal theorem proving), reinforcement learning (to become steadily better at playing this game), language models (to represent policies, value functions, and leverage informal reasoning), program induction (to discover lemmas, create new tactics, or invent useful definitions), and the whole toolbox from game-playing AI, such as tree search and self-play.