Research map / Large language models / AI and society
Too Sure for Our Own Good: A User Study on AI Confidence and Human Reliance
Caterina Fregosi, Lucia Vicente, Andrea Campagner, Federico Cabitza
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use well-calibrated instead of miscalibrated AI confidence | AI decision support confidence scores · Logic puzzle user study (n=184) | mean accuracy improvement: 0.02 -> 0.20 | Improved |
| use miscalibrated confidence (vs no accuracy gain baseline) | AI decision support confidence scores · Logic puzzle user study (n=184) | mean accuracy improvement: +0.02, 95% CI [-0.00, 0.04] | No clear effect |
| use well-calibrated instead of miscalibrated confidence | AI decision support confidence scores · Logic puzzle user study (n=184) | appropriate reliance: 0.25 -> 0.47 | Improved |
| show high instead of low AI confidence | AI decision support confidence scores · Logic puzzle user study (n=184) | appropriate reliance: 0.27 -> 0.49 | Improved |
| miscalibrate AI confidence | AI decision support confidence scores · Logic puzzle user study (n=184) | automation bias (odds ratio): 0.14 -> 0.46 | Hurt |
| miscalibrate AI confidence | AI decision support confidence scores · Logic puzzle user study (n=184) | conservatism bias (odds ratio): 0.54 -> 6.29 | Hurt |
| show high instead of intermediate confidence | AI decision support confidence scores · Logic puzzle user study (n=184) | automation bias (odds ratio): 0.07 -> 0.46 | Hurt |
| show low instead of high confidence | AI decision support confidence scores · Logic puzzle user study (n=184) | conservatism bias (odds ratio): 0.53 -> 5.96 | Hurt |
| use calibrated instead of miscalibrated confidence | AI decision support confidence scores · Logic puzzle user study (n=184) | average perceived utility: 2.09 -> 2.21 | Improved |
| show high instead of low confidence | AI decision support confidence scores · Logic puzzle user study (n=184) | average perceived utility: 1.87 -> 2.48 | Improved |
Closest papers by meaning
- Learning to Decide with AI Assistance under Human-Alignment (NeurIPS 2026 · similarity 0.87)
- Trustworthy AI Must Account for Interactions (NeurIPS 2026 · similarity 0.86)
- Measuring Error Alignment for Decision-Making Systems (AAAI 2025 · similarity 0.86)
- U-trustworthy Models. Reliability, Competence, and Confidence in Decision-Making (AAAI 2024 · similarity 0.86)
- Decoding AI’s Nudge: A Unified Framework to Predict Human Behavior in AI-Assisted Decision Making (AAAI 2024 · similarity 0.85)
- Position: AI Evaluation Should Learn from How We Test Humans (ICML 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
