Perceived humanness and empathy are dissociated in relationship advice generated by a large language model

Bennett Kleinberg, Jari Zegers, Jonas Festor, Stefana Vida, Julian Präsent, Riccardo Loconte, Sanne Peereboom · 2026

Read Perceived humanness and empathy are dissociated in relationship advice generated by a large language model by Bennett Kleinberg, Jari Zegers, Jonas Festor, Stefana Vida, Julian Präsent, Riccardo Loconte, Sanne Peereboom (2026) 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/pmc-13635098.txt.

Description

Kleinberg and six co-authors test whether people can tell machine-written relationship advice from human advice, and what makes text feel human. Across five studies, people and GPT-4 wrote advice, sometimes told to sound human. Fresh readers then guessed each text's source.

Telling the model to sound human worked: readers rated its text as more human, erasing the usual human edge. The trick held even when writers were told not to sound like AI. One study showed the model can sound warm without sounding human, and human without warmth. Its human feel came from casual, chatty style. Human writers barely changed their style when asked.

Catalog record

CREATORBennett Kleinberg, Jari Zegers, Jonas Festor, Stefana Vida, Julian Präsent, Riccardo Loconte, Sanne Peereboom
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE87 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC13635098

Browse the full library · About the Tower of Babel