13 November, 2026

AI always seems to have an answer and, more often than not, it’s a plausible one. But what happens to our own thinking when we increasingly rely on AI to fill the gaps? Two recent research papers offer different sides of the same story.

The first paper brings together evidence that as people use the same large language models to write, generate ideas and reason through problems, our outputs can become more alike. Language becomes more standardised, perspectives converge and creative ideas can become less diverse.

The second study found that access to AI made people much less likely to say “I don’t know”. Faced with difficult questions, participants with AI were more likely to give an answer, even though the AI advice was deliberately designed to be unreliable. They became more confident, but were less likely to be right. Those without AI were more comfortable recognising when they didn’t know.

The connection here is cognitive offloading. We have always used tools to reduce mental effort, from calculators to search engines, but large language models go further. They don’t just help us find information. They can formulate the argument, suggest the perspective and deliver a plausible conclusion. The first paper describes this as externalising not only memory, but the “articulation and justification of thought”. When the answer arrives already sounding convincing, some of the friction involved in reaching our own view disappears. We risk outsourcing not just the task, but part of the judgement that goes with it.

That makes human judgement more important, not less. If AI supplies the plausible answer, our role shifts towards deciding whether it is the right answer: bringing context, experience and alternative perspectives; noticing uncertainty; and questioning answers that sound convincing but may not stand up to scrutiny.

Perhaps good AI use requires preserving a little bit of that friction: thinking before prompting, challenging what comes back, and retaining the confidence to say “I don’t know.”

By Hillevi Fock