Monthly letter · News · Topics · Identity · Dispatches · Est. 2026

Where the Real You Learns to Live, not Just Stand Up

TheSignal

Dispatch · AI & the Self

AI and thinking skills: a 2026 theoretical review and a 2025 survey of 666 people

A July 2026 PsyPost report covers a Trends in Cognitive Sciences review that says the effect of AI depends on how it is used. A March 2025 PsyPost report covers a survey and interview study that linked heavy AI tool use to lower critical thinking scores.

James H. TippinsThe Autonomy Commission
DispatchAI & the Self
SourcesPsyPost on Cash et al. (Jul 15, 2026) · PsyPost on Gerlich (Mar 21, 2025)

PsyPost reported on July 15, 2026 on a theoretical review titled Is AI making us stupid?, by Trent Cash, Megan Kelly, Brooke Macnamara and Evan Risko in Trends in Cognitive Sciences. The review separates learned skills, which it says can decay without practice, from basic abilities such as working memory and attention, which it says resist change from task-specific training. It concludes that the effect of AI depends largely on how people use it, and that there is a risk that AI can compromise skills. The page does not say how many studies the review covers or how they were chosen, and the numbers it quotes come from studies discussed in the review, not from the review itself.

PsyPost reported on March 21, 2025 on a Societies paper published January 3, 2025 by Michael Gerlich. It combined a survey of 666 people of varied ages and education with 50 interviews. Frequent AI tool use was negatively correlated with critical thinking scores, with offloading to AI named as a mediating factor, and ages 17 to 25 showed more use and lower scores. The page states that the study shows correlations, not causation, and that people with weaker critical thinking may be more likely to rely on AI to begin with. It reports no effect sizes or p-values.

Read together, one source describes a mechanism and a condition, practice, under which a skill might fade, and the other counts a link between use and scores at one point in time. Neither tracks the same people over months, and neither measures what any one person gives up or keeps when they hand a task to a tool.

◆ My Take

Use and practice are different things

The review puts the weight on how AI is used, and the survey puts the weight on how often. Those are different questions, and the survey's own caution applies: lower scores could come before heavy use.

Between them they leave a gap that only following the same people over time could fill, which the survey author lists as future work.

◆ If this is your question

Work on it out loud.

Free Mind Thursday is my weekly live call, and it costs nothing. Sign up at beyondunbound.com/freemind. Or get the next issue of The Signal free, below.