Truace
Subscribe
68/100

Moderate evidence

Cite
More
Export CSVExport JSONEmailLinkedInX

AI health coach seemed more human and trusted with self-references; adding small talk backfired

Study title: The Role Of Self-Referential Social Cues and Small Talk In Large Language Model-Based Health Coaching: Enhancing Anthropomorphism and User Trust

Evidence type
Controlled or validated comparison
Sources
1 sourcePeer-reviewed
Published
1 min read
Truvace record
TRV-2026-2694cite or check it

Study title: The Role Of Self-Referential Social Cues and Small Talk In Large Language Model-Based Health Coaching: Enhancing Anthropomorphism and User Trust

A 2x2 experiment tested how two verbal social cues, small talk and self-referential statements, change the way people perceive a large language model acting as a lifestyle behavior change coach. Researchers measured perceived humanlikeness, trust and eeriness.

Read the study abstract

This paper investigates the role of anthropomorphic design in Large Language Models (LLMs), particularly in the context of lifestyle behavior change coaching. This study examines how specific verbal social cues – namely small talk and self-referential social cues – affect user perceptions of anthropomorphism, trust, and eeriness. Findings from a 2x2 experiment show that self-referential social cues enhance perceptions of anthropomorphism, which in turn positively influence user trust. At the same time, the combi…

56Gain side

How the two sides scoreClose call

57Problem side

The gain

In a 2x2 experiment on an LLM lifestyle-coaching chatbot, self-referential social cues raised users' perception of the coach as humanlike, which in turn increased their trust in it.

The problem

In the same 2x2 lifestyle-coaching experiment, combining self-referential cues with small talk had adverse effects on user perceptions, suggesting an uncanny valley effect in text-based chat.

A fixed formula sets both scores from the evidence and its sources. Reader votes don’t move them.

What the scores mean

G is the gain side, out of 100. It goes up when the gain is specific, when it measurably helped people and when the sources behind it are strong.

P is the problem side, out of 100. It goes up when the problem is specific, when it measurably hurt people and when the sources behind it are strong.

In brief

Self-referential cues made the coach seem more humanlike, and that perception was linked to greater trust. Using those cues together with small talk produced adverse effects, which the authors read as a possible uncanny valley effect in text-based chat.

The results bear on how health and lifestyle coaching chatbots are designed: more humanlike chat does not automatically build more trust. The measures are user perceptions, not health or behavior outcomes.

Main points

  1. The study used a 2x2 experimental design crossing small talk with self-referential social cues in an LLM-based lifestyle coach.
  2. It measured three user perceptions: anthropomorphism, trust and eeriness.
  3. Self-referential cues increased perceived anthropomorphism.
  4. Higher perceived anthropomorphism in turn positively influenced trust.
  5. Self-referential cues combined with small talk had adverse effects, suggesting an uncanny valley effect in text-based interaction.

The full story

The paper examines anthropomorphic design in large language models, focusing on lifestyle behavior change coaching. It asks how verbal social cues, specifically small talk and self-referential statements, shape user perceptions of the system.

In the 2x2 experiment, self-referential cues enhanced perceptions of anthropomorphism, and this increased perceived humanlikeness positively influenced user trust. The combination of both cues, however, had adverse effects, which the authors link to a possible uncanny valley effect in text.

Sources

1 on file
  1. Journal of the Association for Information Systemsaisel.aisnet.org
    Peer-reviewed2026-06-14

Talk it over

Four questions to start onThis record found a gain and a problem in the same place. Which matters more to you, and who might answer differently?

Start with a question

AI health coach seemed more human and trusted with self-references; adding small talk backfiredhttps://truvace.com/item/the-role-of-self-referential-social-cues-and-small-talk-in-large-language-model-

  1. This record found a gain and a problem in the same place. Which matters more to you, and who might answer differently?

    The gain In a 2x2 experiment on an LLM lifestyle-coaching chatbot, self-referential social cues raised users' perception of the coach as humanlike, which in turn increased their trust in it.

    The problem In the same 2x2 lifestyle-coaching experiment, combining self-referential cues with small talk had adverse effects on user perceptions, suggesting an uncanny valley effect in text-based chat.

    Answer in the thread

  2. How do we know? What can this kind of evidence show, and what can’t it?

    The evidence This rests on a controlled comparison, from Journal of the Association for Information Systems.

    Answer in the thread

  3. What does it leave open? What else would you want to know before acting on it?

    Answer in the thread

  4. Have you seen AI used this way where you study, work or live? Did it go the same way?

    Answer in the thread

The questions are the same on every record. The details under them are taken from this record as written.

What readers say

More from Health