aceModerate evidence
More
In a 163-person test, privacy assurances and warm wording each raised trust in a bank chatbot
Study title: Privacy and empathy in AI chatbots: A dual-route analysis of cognitive and emotional trust in financial services
- Evidence type
- Controlled or validated comparison
- Sample
- n = 163
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-2687cite or check it
Study title: Privacy and empathy in AI chatbots: A dual-route analysis of cognitive and emotional trust in financial services
A peer-reviewed experiment with 163 participants tested how an AI investment chatbot could earn trust in a simulated banking scenario. Chatbot messages that stressed privacy raised cognitive trust, which concerns the chatbot's competence and integrity. Messages written in an empathetic style raised emotional trust, which concerns warmth and benevolence.
Read the study abstract
Trust remains a critical barrier to the adoption of artificial intelligence (AI) chatbots in financial services, yet extant research has not systematically examined how distinct communication strategies differentially shape cognitive and emotional trust. Grounded in Dual Process Theory and enriched by Theory of Mind, this study investigates how privacy communication and empathetic language influence users’ trust formation through complementary cognitive (System 2) and affective (System 1) pathways. Privacy commu…
The gain
In a 163-participant experiment with a GPT-based investment chatbot in a simulated banking scenario, privacy communication raised cognitive trust and empathetic language raised emotional trust.
In brief
The effect sizes were similar: F = 15.43 (η2 = .088) for privacy on cognitive trust and F = 15.76 (η2 = .090) for empathy on emotional trust. The two factors worked independently, with no significant interaction, so the effects added together rather than amplified one another.
The authors say trust remains a barrier to adopting chatbots in financial services and suggest that banks can combine privacy assurances with empathetic language in interface design. The evidence comes from a simulated scenario, not from real customers handling real money.
Main points
- The study used a 2 × 2 between-subjects design with 163 participants, crossing privacy communication with empathetic language.
- The chatbot was a GPT-based investment assistant used in a simulated banking scenario.
- Privacy communication significantly raised cognitive trust (F = 15.43, p < .001, η2 = .088).
- Empathetic language significantly raised emotional trust (F = 15.76, p < .001, η2 = .090).
- Both factors strengthened overall trusting beliefs, with no significant interaction between them.
The full story
The researchers drew on Dual Process Theory and Theory of Mind. They treated privacy communication as a structural assurance that prompts analytical (System 2) evaluation of a chatbot's competence and integrity. They treated empathetic language as a cue that triggers intuitive (System 1) impressions of warmth and benevolence.
Participants were split across four conditions in a between-subjects design. Privacy messaging moved cognitive trust and empathetic wording moved emotional trust. The absence of an interaction supports what the authors call an additive dual-pathway model.
The authors say the findings offer interface-design guidance for banking executives who want to build user confidence through both rational and relational trust mechanisms.
Sources
1 on file- Journal of Global Scholars of Marketing Sciencedoi:10.1080/21639159.2026.2660934
Talk it over
Who gets this benefit, and who was left out of the study or story?
In a 163-person test, privacy assurances and warm wording each raised trust in a bank chatbothttps://truvace.com/item/privacy-and-empathy-in-ai-chatbots-a-dual-route-analysis-of-cognitive-and-emotio
Who gets this benefit, and who was left out of the study or story?
The gain In a 163-participant experiment with a GPT-based investment chatbot in a simulated banking scenario, privacy communication raised cognitive trust and empathetic language raised emotional trust.
Answer in the threadHow do we know? What can this kind of evidence show, and what can’t it?
The evidence This rests on a controlled comparison with a sample of 163, from Journal of Global Scholars of Marketing Science.
Answer in the threadWhat does it leave open? What else would you want to know before acting on it?
The record’s own caveat The experiment ran in a simulated banking scenario with 163 participants, so the results do not show how trust would change with real customers and real money.
Answer in the threadHave 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