CONCEPT
Testimony and Trust
Ricoeur's analysis of testimony as a speech act requiring the witness's
attestation—staking credibility on a claim others cannot verify—and the framework revealing why AI's confident output is testimony without attestation, requiring hermeneutical discipline to evaluate.
Testimony occupies a unique position in Ricoeur's epistemology: it is neither knowledge (the recipient cannot verify) nor faith (it is defeasible by evidence), but a third category Ricoeur called
credence—the willingness to accept another's word as provisionally true. Trustworthy testimony requires internal coherence, consistency with other evidence, susceptibility to challenge, and—crucially—the witness's
attestation:
I was there, I saw this, I stake my reliability on this claim. AI-generated output functions as testimony in every collaboration: the machine reports claims about the world the builder cannot always verify. But the testimony lacks attestation—the machine risks nothing, possesses no credibility to damage, makes no self-commitment. The result is testimony that can be confident, coherent, eloquent, and wrong—requiring the builder to supply the verification the witness cannot provide.
In The You On AI Field Guide
Ricoeur developed his theory of testimony across multiple works—Essays on Biblical Interpretation, The Just, and the second volume of