I have read with particular interest the article by Eduardo López-Collazo, published in El Español, which examines the impact of artificial intelligence on writing and cognition. It stands out in a debate often dominated by exaggeration by combining empirical evidence with philosophical depth.
The piece draws on an MIT experimental study comparing writing processes with and without AI assistance. The findings are sobering. Intensive use of tools such as ChatGPT accelerates production but appears to reduce cognitive engagement, memory retention, and neural activity associated with learning. The resulting texts are grammatically correct and structurally coherent, yet often lack intellectual depth — technically sound but conceptually thin.
Perhaps the most revealing detail is that the strongest results come from those who first think and draft independently, and only then use AI as a tool for revision. The sequence matters.
López-Collazo enriches this empirical discussion by invoking Plato’s Phaedrus, where writing itself is presented as a technology that would transform memory. Plato feared that reliance on writing would weaken internal recollection. History showed that writing did not diminish intelligence; it reconfigured how memory operates.
The analogy is instructive.
If writing altered what we remember, artificial intelligence may alter how we think. Not because it imposes decisions, but because it can habituate us to delegating the very acts that shape thought: hesitation, reformulation, doubt, structural organisation, and conceptual struggle.
The risk is not technological determinism. It is an anthropological substitution.
When AI serves as a prosthesis — extending our cognitive reach after disciplined effort — it strengthens reasoning. When it serves as a substitute — replacing the formative struggle — it weakens the habits on which judgement depends.
The distinction between producing more and understanding better is becoming blurred. Productivity metrics can mask intellectual erosion.
For educators and institutional leaders, the implication is clear: the decisive question is not whether AI should be used, but how it is integrated into the learning process. Reflection must precede automation.
Artificial intelligence may change our tools.
Whether it changes our thinking depends on how we form the habits that underpin it.