A 30-minute generative AI literacy training increased intelligence analysts' overall accuracy at distinguishing real from AI-generated images by 9 percentage points¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r2 · 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? Among intelligence analysts, a 30-minute expert-led training increased overall accuracy by 9 percentage points (95% CI: [2.7, 15.4]) over a 72% pre-training baseline. → Negar Kamali 2026 q3 i? The stricter pair-level outcome, correct only when both members of a matched real–fake pair are classified correctly, also improved after training. → Negar Kamali 2026 (2)

Evidence¶

Negar Kamali 2026¶

Negar Kamali, Candice Rockell Gerstner, Jessica Hullman, Matthew Groh. (2026). Generative AI Literacy Training Improves Intelligence Analysts' Discrimination of Real and AI-Generated Images. https://arxiv.org/abs/2606.28510 doi:10.48550/arxiv.2606.28510

q3 · i? · causal · r2

Pre-registered OLS regression of trial-level accuracy with the BA×Post interaction as the critical test, standard errors clustered at the participant level, from the counterbalanced within-subject experiment. The article reports "overall accuracy increased by 9 percentage points (95% CI: [2.7, 15.4])".

"The training intervention produced a meaningful improvement in detection performance: overall accuracy increased by 9 percentage points (95% CI: [2.7, 15.4]), corresponding to a predicted post-training accuracy of 82%."

Negar Kamali 2026 (2)¶

Negar Kamali, Candice Rockell Gerstner, Jessica Hullman, Matthew Groh. (2026). Generative AI Literacy Training Improves Intelligence Analysts' Discrimination of Real and AI-Generated Images. https://arxiv.org/abs/2606.28510

q3 · i? · causal · r2

Same pre-registered OLS analysis applied to pair accuracy, a stricter post hoc measure requiring both members of a matched real–fake pair to be correctly classified, reducing inflation from chance-level guessing. The article reports "pairwise accuracy increased by 11 percentage points".

"Results were consistent across outcome measures: pairwise accuracy increased by 11 percentage points (95% CI: [1.8, 21.0])."

Discussion¶

Related Claims¶

• AI literacy interventions show a large, statistically significant overall positive effect on student AI literacy outcomes (g = 0.837) — the general claim this finding bears on, with the studies behind it
• Training improved accuracy in every digital-forensics experience group and most generative-AI familiarity groups, with the largest estimated gains among analysts with no forensics experience and frequent generative-AI users — related
• The training gain was driven by a 14.2-percentage-point improvement in accuracy on real images, while the gain on AI-generated images was not statistically significant — related
• Training effects concentrated in portrait images (+39.8 percentage points, significant), with directional non-significant negative estimates for full-body and posed-group images — related
• Training gains were distributed broadly across the 97 stimulus images rather than concentrated in a small subset, though stimulus-level effects were heterogeneous — related
• Training improved discrimination sensitivity (d′) with only a small, non-significant criterion shift toward labeling images as AI-generated — related
• Attendees correctly identified AI-generated images in 82% of polling responses, and polling served as a scaffold for discussion rather than a measurement tool — related
• After training, analysts' free-text detection reasoning shifted away from physics-violation and anatomical cues toward functional and stylistic artifacts and became more specific — a narrower finding that bears on this claim