Evidential Claims

Building trust by being transparent about evidence

An analysis of how honesty about evidence shapes a source’s credibility.

The challenge

Is factual accuracy enough to establish trust, or does transparency about how we know things matter just as much?

This work shows how misrepresenting information sources undermines trust and informs principles for communicating evidence provenance in human and human-AI systems.

My role & methods

At the Institute of Communication and Cognition Sciences, I designed and implemented an online experiment (Qualtrics), recruited participants (Prolific), ran data collection, and analyzed the results.

Participants evaluated a collaborator (“Peter”) who made either a strong evidential claim (“I saw that X”) or a weak hearsay claim (“Somebody told me that X”). Both statements were accurate, but their claimed evidence was false.

Key insight: Transparency about evidence is essential for credibility.

A speaker was penalized more for overstating evidence (“I saw it”) than for understating it (“I heard it”).

This finding highlights that for human and human-AI information systems, trustworthiness depends not only on transparency about what is known, but also on how it is known.