Does frequent AI tool use reduce our critical thinking?

A mixed-method study surveyed 666 valid participants in the United Kingdom (669 recruited) using a 23-item questionnaire built on validated scales, including the Halpern Critical Thinking Assessment (HCTA) and Terenzini's self-reported critical-thinking measures, plus semi-structured interviews with a subset of 50 participants. Gerlich reports "a significant negative correlation between frequent AI tool usage and critical thinking abilities, mediated by increased cognitive offloading," calling the mediation finding itself "a novel contribution of this study." Age and education moderated the pattern: "younger participants exhibited higher dependence on AI tools and lower critical thinking scores compared to older participants," while "higher educational attainment was associated with better critical thinking skills, regardless of AI usage."

The paper frames the pathway through Risko and Gilbert's definition of cognitive offloading — delegating cognitive tasks to external tools to reduce working-memory load — which "can free up cognitive resources" but also risks "a decline in cognitive engagement and skill development." Gerlich ties the mechanism to trust: as "users develop greater trust in AI, they are more likely to delegate cognitive tasks to these tools," producing a reinforcing cycle in which routine delegation deepens dependence. The paper connects this to Carr's argument in The Shallows that technology creates "cognitive shortcuts that reduce the need for deep thinking."

This is cross-sectional self-report evidence for a mechanism that Does AI assistance weaken our brain's ability to think independently? measures directly in neural data, and that Does ChatGPT help writers think less deeply about revision? measures through an essay-writing RCT: three different methods — survey, EEG, and randomized experiment — converge on the same offloading-reduces-engagement pattern. It also extends Does AI assistance actually harm the way developers learn? beyond a professional-skill context (developers learning a new library) to general-population critical thinking, and adds demographic structure — age and education — that those experimental studies do not test. It also stands against a possible remedy: Can metacognitive feedback stop students from offloading to AI? shows one way the offloading this survey measures can be reduced.

The design is cross-sectional and correlational: AI usage, cognitive offloading, and critical-thinking scores were all captured by self-report in the same survey wave, so the study cannot rule out reverse causation — people with weaker critical thinking turning to AI tools more — or a shared cause behind both. Gerlich's own limitations section names "its reliance on self-reported measures and the potential for sample bias." The finding supports treating frequent AI reliance as a correlate of lower measured critical thinking in this UK sample, not as established proof that AI use causes the decline; that stronger claim would need the longitudinal or experimental designs used in the EEG and essay-writing studies above.