Definition
Bayesian updating is the process of revising credences when new evidence arrives, by comparing how expected that evidence is under competing hypotheses.
Why It Matters
It is the actual mechanics of intelligence, whether in a human brain or an AI. By constantly reallocating confidence based on how the world pushes back, we ensure our internal maps stay aligned with the external terrain.
Core Concepts
- Evidence favors hypotheses under which it is more likely.
- Priors matter, but strong evidence can overwhelm weak priors.
- Updating is comparative rather than absolute.
- Rational disagreement can persist when priors or evidence assessments differ.