This is the authors' abstract. We don't add key points for this paper.
A number of factors, including different task characteristics, influence trust in human vs. AI decision support. In particular, the aspect of information quantifiability could influence trust and dependence, especially considering that human and AI support may have varying strengths in assessing criteria that differ in their quantifiability. To investigate the effect of information quantifiability we conducted an online experiment ( N = 204 ) with a 2 (support agent: AI vs. human) × 2 (quantifiability: low vs. high) between-subjects design, using a simulated recruitment task. The support agent was manipulated via framing, while quantifiability was manipulated by the evaluation criteria in the recruitment paradigm. The analysis revealed higher trust for human over AI support. Moreover, trust was higher in the low than in the high quantifiability condition. Counterintuitively, participants rated the applicants as less qualified than their support agent’s rating, especially noticeable in the low quantifiability condition. Besides reinforcing earlier findings showing higher trust towards human experts than towards AI and showcasing the importance of information quantifiability, the present study also raises questions concerning the perceived leniency of support agents and its impact on trust and behavior. • Higher trust and perceived reliability in human experts vs. AI support. • Higher trust and perceived reliability in the low vs. high quantifiability condition. • Participants rated the applicants as less qualified than the rating of the support.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Roesler et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: