https://gfrm.in/categories/philosophy/Recent content in Philosophy on Guy FreemanHugoenSat, 31 Jan 2026 00:00:00 +0000https://gfrm.in/posts/bayesian-agent-part2/Sat, 31 Jan 2026 00:00:00 +0000https://gfrm.in/posts/bayesian-agent-part2/<p>In <a href="https://gfrm.in/posts/bayesian-agent/">Part 1</a>, I built an agent that learns which foods are safe through Bayesian inference. It starts ignorant, observes outcomes, updates its beliefs using exact conjugate mathematics, and eventually acts with something resembling competence. Clean code, sound theory, and those belief distributions converging in real-time remain genuinely satisfying to watch, in the way that all correctly implemented mathematics is satisfying to watch.</p> <p>Something has been nagging at me, though.</p> <p>The agent learns <em>what</em> to believe. I designed <em>how</em> it believes. I chose the variables it perceives. I specified the structure of its world-model. I set the prior hyperparameters. The agent&rsquo;s entire cognitive architecture &mdash; the shape of its epistemic machinery &mdash; came from me, handed down like tablets from a mountain. The agent had no say in the matter. As someone who spent years doing Bayesian statistics, I should know better than to treat the model structure as given. The prior over model structures is the prior that actually matters, and I skipped it entirely.</p>