Prior genAI access shows only exploratory evidence of poorer subsequent unaided reasoning, within the no-incentive condition, and no preregistered main effect¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r3 · q3 · i1 small
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person

Subclaims¶

q3 i1 Within the no-incentive condition, participants with prior genAI access had lower odds of answering a Stage 2 network-reasoning question correctly than participants who had worked without genAI. → Dawid Ratajczyk 2026 q3 i? The preregistered test found no significant main effect of prior genAI access on Stage 2 accuracy. → Dawid Ratajczyk 2026 (2)

Evidence¶

Dawid Ratajczyk 2026¶

Dawid Ratajczyk, Agata Dymarska, Adam Matłoka, Michał Tomczyk, Małgorzata Wiącek. (2026). Thinking With AI, Reasoning Without It: Cognitive Offloading to Generative AI Weakens Effort Regulation. Preprint. https://osf.io/59qtm/ doi:10.31234/osf.io/83jum_v1

q3 · i1 · causal · r3

Exploratory item-level mixed-effects logistic regression on Stage 2 accuracy in the preregistered lab experiment. The quoted estimate shows prior genAI users had OR = 0.68, approximately 32% lower odds of a correct response in the no-incentive condition.

"Within the no-incentive reference condition, prior genAI access was associated with lower odds of answering a Stage 2 question correctly (𝑏= −0.392,𝑆𝐸= 0.195,𝑧= −2.01,𝑝=.045,𝑂𝑅= 0.68)."

Dawid Ratajczyk 2026 (2)¶

Dawid Ratajczyk, Agata Dymarska, Adam Matłoka, Michał Tomczyk, Małgorzata Wiącek. (2026). Thinking With AI, Reasoning Without It: Cognitive Offloading to Generative AI Weakens Effort Regulation. Preprint. https://osf.io/59qtm/

q3 · i? · causal · r3

Preregistered 2×2 ANOVA on Stage 2 network-reasoning accuracy in the same lab experiment. The quoted tests show no significant genAI, incentive, or interaction effects, so H1a, H1b and H4 were not supported. The lower odds of a correct answer appear only in the exploratory item-level analysis, within the no-incentive condition; the genAI × incentive interaction was not significant there either, and the authors say this conditional effect "should not be interpreted as evidence that the effect of genAI differed between incentive conditions". GenAI access was assigned; the study does not separate those who used the tool from those who had it.

"There was no significant main effect of prior genAI access (𝐹(1,153) = 1.84,𝑝=.177,𝜂 2 𝑝 =.011), no significant main effect of incentives (𝐹(1,153) = 1.14,𝑝=.287,𝜂 2 𝑝 =.007), and no significant interaction between prior genAI access and incentives (𝐹(1,153) = 2.22,𝑝=.138,𝜂 2 𝑝 =.014)."

Discussion¶

Related Claims¶

• Participants deprived of relevant relational knowledge in Stage 1 take longer on the subsequent network-reasoning task — related
• Prior genAI use and incentives show no clear effects on later memory retention — related
• GenAI use weakens effort adjustment to incentives on subsequent unaided reasoning — related
• Access to an LLM assistant during practice did not significantly lower subsequent unaided test performance — related
• Greater answer offloading during learning is associated with lower subsequent unaided performance (exploratory) — related
• Review reports a field experiment in which GPT access improved supported practice but was followed by poorer unaided test performance, mitigated by a guarded tutor — a broader claim this one bears on
• In one preregistered experiment, metacognitive feedback on learners' own LLM use lowered the odds of answer offloading and raised the odds of a correct unaided answer on one-sided tests, with 95% intervals reaching no effect — related