Perceived risk negatively predicts use behavior and operates as a stage-sensitive deterrent, activated at submission and evaluation rather than initial exploration¶

Claim · All claims
Evidence · 2 studies · 1 associational r2, 1 qualitative r2 · q2–q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? Perceived risk significantly and negatively predicted use behavior (β = −0.384, p = 0.022), supporting H7. → Liu Y 2026 q2 i? Interviews indicate risk rarely barred initial experimentation but became most salient when AI-assisted work moved to submission, critique, or formal evaluation. → Liu Y 2026 (2)

Evidence¶

Liu Y 2026¶

Liu Y, Meng M and Zhang Y (2026) The competence paradox: negotiating ease, risk, and creative identity in text-to-image generative AI use among art and design students. Front. Psychol. 17:1858187. https://doi.org/10.3389/fpsyg.2026.1858187

q3 · i? · associational · r2

Structural model of the 400-student survey: PR→UB printed β = −0.384, SE = 0.168, p = 0.022; the same sentence reports the FC→UB null (β = −0.002, p = 0.980), so H4b was rejected.

"For downstream outcomes, PR significantly and negatively predicted UB, supporting H7, whereas FC showed no significant direct effect on UB, leading to the rejection of H4b."

Liu Y 2026 (2)¶

Liu Y, Meng M and Zhang Y (2026) The competence paradox: negotiating ease, risk, and creative identity in text-to-image generative AI use among art and design students. Front. Psychol. 17:1858187. https://doi.org/10.3389/fpsyg.2026.1858187

q2 · i? · qualitative · r2

Thematic analysis of the 20 follow-up interviews distinguished evaluation-oriented risk (submission legitimacy, external judgment) from development-oriented risk (weakened originality, loss of foundational skills).

"the interviews suggested that students rarely described risk as a barrier to initial experimentation; instead, it became most salient when AI-assisted work moved from exploration to submission, critique, or formal evaluation."

Discussion¶

Related Claims¶

• A negative behavioral path through frequent resubmissions counteracted the affective benefit of layered feedback — related
• Perceived independence did not mediate a significant indirect effect of layered AI feedback on learning performance — related
• Exploratory multi-group analyses suggest competence-related paths differ by educational level and gender, with heterogeneity concentrated downstream of intention — related
• Personal innovativeness did not significantly influence behavioral intention to use ChatGPT — related