INTIMA: A Benchmark for Human-AI Companionship Behavior
Lucie-Aimée Kaffee, Giada Pistilli, Yacine Jernite · 2025
Read INTIMA: A Benchmark for Human-AI Companionship Behavior by Lucie-Aimée Kaffee, Giada Pistilli, Yacine Jernite (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-09998.txt.
Description
Lucie-Aimee Kaffee, Giada Pistilli, and Yacine Jernite grade how chat models act when users grow fond of them, in a paper from 2025. Fond bonds with rigs can soothe or harm, and no yard stick yet grades the acts. Their stick sorts acts into kinds, from bond feed to bound hold to plain.
Three hundred and odd prompts probe the kinds, drawn from mind lore and user tales. Four top models took the probes. Bond feed acts beat bound hold acts in each model by a wide gap. Firms stress odd kinds of the stick in their top builds, which splits the field. Both bound sense and warm back count for user good, they urge.
Catalog record
| CREATOR | Lucie-Aimée Kaffee, Giada Pistilli, Yacine Jernite |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 50 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-09998