Frank Ramsey's Forgotten Philosophy Offers New AI Induction Model
A pre-war philosopher's view on universal statements could reshape how AI handles induction and uncertainty.
Frank Ramsey's 1920s philosophy offers a third path between Bayesian and Popperian approaches. Unlike Bayesians who assign probabilities to universal statements or Popperians who treat them as falsifiable propositions, Ramseyians treat universal statements as 'variable hypotheticals' — rules for judging without truth value or probability. This dissolves the induction problem by focusing on singular propositions rather than universal ones, and resolves Popper's objection that no probability can be assigned to universal laws.
- Ramseyians treat universal statements as 'variable hypotheticals' (rules for judging) with no truth value or probability, unlike Bayesian models.
- Popper's objection that universal laws have zero probability on finite evidence is avoided because Ramseyians only assign probabilities to singular propositions.
- Induction becomes evaluation of rule reliability over time, not inference of general truths — potentially improving AI robustness.
Why It Matters
This third viewpoint may resolve long-standing philosophical tensions in AI reasoning and enable more efficient uncertainty handling.