If AI weakens the brain
- Algopolio
- Dec 4, 2025
- 3 min read
Source: Corriere della Sera – “If AI weakens the brain”, 3 December 2025
The cognitive emergency of the AI era
The article highlights a crucial issue: the increasing use of artificial intelligence tools — particularly generative language models — may weaken essential cognitive skills, especially among younger generations. AI accelerates processes, provides shortcuts and offers immediate solutions, but this apparent efficiency risks eroding the mental effort necessary to build deep competencies. The digital environment does not train the mind: it replaces it.
The university experiment: when the brain stops “working hard”
The study referenced — conducted by MIT — divided participants into three groups:
the LLM group, which relied on generative AI;
the Brain-only group, which worked without technological assistance;
the Hybrid group, which used digital tools more moderately.
Results revealed clear divergences:
continuous use of AI correlated with a decline in neural connectivity, indicating reduced cognitive engagement;
participants who shifted from AI to autonomous work showed partial recovery, though with weaker connectivity;
the Brain-only group demonstrated the strongest cognitive performance, especially in memory recall and depth of processing.
The worrying conclusion is that repeated exposure to AI leads to cognitive “shortcutting” — a kind of mental atrophy.
Delegation becoming dependence
The core of the analysis is straightforward: the efficiency promised by AI comes at a cost.Whenever we delegate a task — drafting, reasoning, synthesising — we relinquish part of the cognitive training that builds autonomy.This can undermine:
critical analysis,creative reasoning,academic depth,the ability to form independent judgement.
The mind grows through exertion. Misused, AI removes that exertion.
The educational paradox: more AI = less thinking?
The authors note a paradox: the enthusiasm for AI within the academic world often obscures the real risk. If universities integrate AI without reshaping teaching methods, they risk producing a generation of students who achieve results but cannot generate meaning. Understanding how to use tools is not enough: we must question what is lost when tools perform the very processes that define identity, knowledge and autonomy.
The issue is not AI itself. The issue is an educational system that adopts it without teaching how to preserve cognitive independence.
The illusion of intelligence: AI does not think — it completes
AI’s apparent intelligence is misleading. Models do not possess intuition, intention or understanding; they generate statistically coherent completions. Confusing this with genuine reasoning creates two dangers:
granting AI undue epistemic authority, underestimating the human role in decision-making.
AI can produce plausible responses, but not reliable judgement. Discernment — the most human of faculties — cannot be outsourced.
Learning to think: a civic and political responsibility
The article concludes with a call to re-examine our relationship with technology. We need a new cognitive literacy that helps citizens:
recognise AI’s limits,preserve focus,cultivate creativity and imagination,train their abilities independently of machines.
AI can support learning, but it must not replace reflection: thinking remains a human responsibility.
What this means for Algopolio: defending cognitive sovereignty
This issue touches a central concern for Algopolio: the growing vulnerability of individuals in a technological ecosystem that risks eroding cognitive autonomy.Algopolio works to: protect citizens from the subtle harms of algorithmic dependence,support those affected by opaque or intrusive digital systems, promote awareness of the cognitive risks associated with excessive automation,ensure that technology remains a tool — not a substitute — for human reasoning.
In an age where thinking can be delegated to machines, protecting the sovereignty of the human mind becomes essential. Those who feel overwhelmed or disadvantaged by AI-driven dynamics can turn to Algopolio for guidance, analysis and concrete support.


Comments