Model Discovery Agent: LLM-Assisted Bayesian Experiment Design
K. Murphy · 2026
What this means for traders
When the final decision rests on deterministic machinery rather than on the model itself, swapping the underlying model stops being a source of unpredictable behavior.
Abstract
Couples an LLM used strictly as a proposer of candidate mechanisms with standard Bayesian machinery: sequential Monte Carlo for posteriors and evidence, and value-of-information to choose the next experiment. Across physics, chemistry and neuroscience benchmarks it recovers the correct mechanism far more data-efficiently than an LLM agent working alone. The most transferable result is a robustness one: because the decision sits in the deterministic layer, accuracy stays between 89% and 94% regardless of which model proposes, whereas the pure LLM agent swings from 26% to 81% depending on the model behind it.