Preferences for ChatGPT vs. Human Advisers: Do the Adviser and Cognitive Load Predict Risky Decision-Making?
Keywords
Loading...
Authors
Issue Date
2026-02-28
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
Participants aged 18–65 years old were recruited to participate in the Trust Game,
which was an online simplified version of the one used in Klingbeil et al. (2024). In half of
the trials, they received advice from ChatGPT, and in the other half, it was from another
participant. The first block did not induce cognitive load, and in the second block cognitive
load was induced by including time constraints. Across all four conditions, the number of
times they took the given advice was measured. No significant effect on rates of advice taking
was found for source of advice, cognitive load, or for their interaction. These results did not
replicate most earlier findings. It is possible that factors like the way we manipulated
cognitive load or the adjustments to the experimental design, as well as the research sample,
account for these differences.
Keywords: decision-making, social risk, advice-taking, following advice, ChatGPT, cognitive
load, time pressure, financial decisions, trusting behaviour, algorithm aversion, algorithm trust
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen