Partnership with the Machine: Why We Need "Productive Friction"

Fifty-four participants wrote essays over several sessions, some with the help of a language model, some with a search engine, and some with their own minds alone, while their brain signals were recorded. This is the paper "Your Brain on ChatGPT", published in June 2025 by researchers led by Nataliya Kosmyna. It contains a point worth reading calmly.

What did the study find?

Brain connectivity, measured by electroencephalography, was strongest among those who wrote without tools, intermediate among search-engine users, and weakest among model users. Model users reported the lowest sense of ownership of their essays and had difficulty quoting what they had written. The researchers called what they saw "cognitive debt": comfort now that may be offset by a cost later.

But let us weigh the evidence. There were 54 participants, only 18 completed the fourth session, the task was essay writing of one kind, and the paper is a preprint whose page does not indicate that it has been peer-reviewed. It is a signal that calls for more research, not proof that the mind is atrophying.

Trust in the tool or trust in oneself?

Closer to professional reality is a survey presented by researchers from Carnegie Mellon University and Microsoft at the CHI conference in 2025, of 319 knowledge workers who described 936 tasks. They found that higher confidence in generative AI went with less critical thinking and a sense that it takes less effort, while higher confidence in oneself went with more critical thinking and more effort. The data are self-reported and correlational, so they do not say the tool is the cause; but they point out that excessive trust in a polished answer may silence the doubt we need.

In an older field, a study of about fifty drivers indicates that those who had used navigation systems more over their lives had weaker spatial memory when finding their way without help. The follow-up, however, covered only 13 participants, and the authors themselves caution against firm conclusions. The pattern is familiar, but the case still needs wider studies.

Productive friction

We use the phrase "productive friction" here as a description, not a scientific term: it is the mental effort we expend when we wrestle with an idea, shape a sentence or settle a hard decision, and it is where a faculty grows. This does not mean abandoning tools; delegating repetitive tasks is a blessing to be thankful for. We simply ask: which effort do we want to keep because it builds us?

World Economic Forum data (2025) indicate that analytical thinking is the most sought-after core skill, with about seven in ten employers considering it essential, and that employers expect about 39% of a worker's skills to change or become outdated by 2030. In our assessment, what becomes outdated is the procedure, while critical judgement remains in demand.

Small habits

1. Think first: write your own draft, opinion or plan, then ask the tool to review it.
2. Ask for critique, not a replacement: "What are the weaknesses in my argument?" is more useful than "write this for me".
3. Treat the answer as a draft: check figures and sources before relying on them.
4. Keep human spaces: a letter to someone you love, an educational conversation with your children, an ethical dilemma you ponder yourself.
5. Teach the young to question: digital literacy today should include telling the reliable from the fabricated, not only mastering tools. See the article "The End of the Age of Rote Instruction?"

The mind is a trust and a blessing, and we give thanks for it by using it well and developing it. The tool can speed up our work, but responsibility for decision and judgement stays with the one who will be asked about them.