Access to AI advice roughly doubled reported confidence in answers in one experiment¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r3 · q3
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person

Subclaims¶

q3 i? AI advice nearly doubled reported confidence on a 0–100 scale (29.6 vs. 75.9 without stakes; 37.7 vs. 73.5 with stakes). → Chiara Marcoccia 2026 q3 i? Stakes raised confidence only when AI advice was unavailable; the confidence increase disappeared when AI advice was available (significant interaction). → Chiara Marcoccia 2026 (2)

Evidence¶

Chiara Marcoccia 2026¶

Chiara Marcoccia, Walter Quattrociocchi, Valerio Capraro. (2026). AI advice suppresses people's willingness to say "I don't know", even when the advice is wrong and accuracy is incentivized. https://osf.io/nwx6r/overview?view_only=a03ae0ca587c4c84a442c44d55ab97b6 doi:10.31234/osf.io/5y6m4_v1

q3 · i? · causal · r3

Study 2 secondary measure: after each question participants rated confidence on a continuous 0–100 scale. The article reports a "large main effect of AI advice, which nearly doubled reported confidence"; no standardized effect size printed.

"Confidence judgments revealed a large main effect of AI advice, which nearly doubled reported confidence (29.6 vs. 75.9 without stakes; 37.7 vs. 73.5 with stakes)"

Chiara Marcoccia 2026 (2)¶

Chiara Marcoccia, Walter Quattrociocchi, Valerio Capraro. (2026). AI advice suppresses people's willingness to say "I don't know", even when the advice is wrong and accuracy is incentivized. https://osf.io/nwx6r/overview?view_only=a03ae0ca587c4c84a442c44d55ab97b6

q3 · i? · causal · r3

Study 2 simple slopes for the significant stakes × AI interaction on confidence (b = −10.509, p = .006): stakes raised confidence only "when participants could not seek AI advice". No standardized effect size printed.

"Monetary stakes increased confidence when participants could not seek AI advice (29.6 vs. 37.7; b = 8.083, SE = 3.163, t = 2.556, p = .011), but this effect disappeared when AI advice was available (75.9 vs. 73.5; b = −2.426, SE = 2.216, t = −1.095, p = .274; see Figure 2b)."

Studies bearing on this claim¶

Does working with AI make learners' judgments of their own knowledge, effort and performance less accurate, and what restores them? This table gathers the studies behind this claim and the 12 narrower claims that bear on it: 8 distinct studies (6 causal, 1 theoretical, 1 review), q2–q4, 3 of 8 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Discussion¶

Learner Variables¶

• Metacognition — outcome: instruction changes it

Related Claims¶

• Self-efficacy judgments correspond less well to behavior in learning settings than in performance settings — the general claim this finding bears on, with the studies behind it
• Merely having access to AI advice nearly eliminates people's willingness to suspend judgment under uncertainty, even when the advice is wrong — related
• Monetary incentives for accuracy improve answer correctness specifically when AI advice is available, by reducing reliance on that advice — related
• Monetary stakes for accuracy act independently of AI availability on judgment suspension, with no significant AI × stakes interaction — related
• AI advice dramatically reduces judgment suspension even when it is displayed automatically rather than actively requested — related
• Participants seek AI advice less frequently when accuracy carries monetary stakes — related
• AI assistance can inflate confidence even after errors and reduce accuracy when AI is faulty (cognitive surrender) — related
• 88% of AI-assisted students reported increased self-efficacy and confidence during independent practicals — related
• Practicing with AI did not create an illusion of mastery: participants reported learning and skill levels similar to other conditions — related