Where the Evidence Lives: Auditing AI Companions' Self-Descriptions

Abstract:Companion agents describe themselves: they remember, they understand their users, the relationship has changed them. We argue that such accounts, and the experience ratings that seem to confirm them, are checkable by users only where the evidence is theirs: in the agent's behavior, or in themselves. Where the evidence lives in the machinery, fluent self-description and moderately positive ratings do not establish that the mechanisms behind them ran. We demonstrate an audit procedure that sets an agent's self-description against its users' judgements and its implementation records, reporting each claim as supported, contradicted, or unresolved, and apply it to Lita, a proactive companion we built and deployed for a month with nine colleagues. Participants endorsed stylistic claims, withheld endorsement from relational ones, and rated memory at or above midpoint, while two of three memory layers had never executed their accumulation step. Memory-bearing agents should report what their self-descriptions cannot establish.

17 pages, 8 figures, 9 tables. Ancillary files contain the LLM judge prompt (Japanese original and English translation) and the probe statements

Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)

arXiv:2609.38753 [cs.HC]

(or arXiv:2609.38753v1 [cs.HC] for this version)

https://doi.org/10.48550/arXiv.2609.38753 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

judge_prompt_en.txt

judge_prompt_ja.txt

nda_statements.md

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.