AI That Declares Its Priors

An expectation that political AIs should, by design, display the causal assumptions and normative priors (e.g., that disparate outcomes imply discrimination) underlying their answers, similar to academic disclosure of methods and priors. This would operationalize transparency in model outputs and make contested claims traceable to explicit assumptions. — If implemented, it would change accountability for automated political advice, shift debates from 'is the AI biased?' to 'which assumptions drove this conclusion?', and reshape regulatory and platform standards for AI in public debate.