In an essay for The Freethinker, behavioural scientist Gleb Tsipursky argues that AI poses a subtler risk to free thought than censorship. The danger, in his view, is that people gradually hand over the very thinking that would let them notice when the machine is wrong. He cites a field experiment in which consultants using GPT-4 did much better on tasks within the model's reach but worse on a task outside it, and a study where students lost their edge once the AI tutor was taken away.

His remedy is 'metacompetence': knowing what to delegate, checking outputs, forming your own view first and sometimes working without the tool. For freethinkers, the point is that independent judgement is a habit that has to be practised, or it fades.

Source: Gleb Tsipursky, The Freethinker (2026-11-06)
https://freethinker.co.uk/2026/11/the-freedom-to-think-requires-the-ability-to-disagree-with-ai/

Summarised by the FTI newsdesk.