Researchers Link Heavy AI Use To Weaker Thinking

ItSecurityNews reports that researchers and educators are documenting possible cognitive effects from increased everyday use of AI tools. The article cites a study that analysed responses from over 650 individuals aged 17 and older in the UK and reports a correlation between heavy AI reliance and reduced recall or critical-thinking performance in some participants, according to ItSecurityNews. ItSecurityNews also describes anecdotal observations by Nataliya Kosmyna: she found multiple internship cover letters with similar structure and polished language suggestive of large language model assistance, and while teaching at the Massachusetts Institute of Technology she observed students having more difficulty recalling material compared with prior years.

Industry-pattern observations: research on tool-assisted cognition typically draws a distinction between externalising memory (offloading facts to tools) and degrading the retrieval pathways that reinforce learning. Studies in adjacent literatures on calculator use and search-engine dependence provide methodological templates, such as controlled recall tests and longitudinal follow-ups. For practitioners designing AI-assisted workflows, the relevant technical considerations include prompt design, transparency about source provenance, and features that encourage active recall rather than passive acceptance.

For practitioners and educators, reported correlations between heavy LLM use and weaker retention matter because they affect measurable outcomes like recall and critical evaluation. Reporting by ItSecurityNews adds qualitative signals, hiring managers seeing homogenised application text and instructors reporting reduced retention, that complement survey data. These signals do not establish causation; they indicate areas where experimental, controlled studies would be useful.

Observers should look for peer-reviewed replications, experimental interventions that test active-learning integrations with AI, and vendor features that enable citation, transparency, or "explain-your-reasoning" modes. Public statements or published studies from universities and educational researchers will be key to moving from correlation to actionable guidance.