What do brain activity and memory tests show about students recalling their own essays after writing with versus without AI?

What the brain and memory tests actually showed

The most direct evidence comes from a study comparing students who wrote essays without AI against students who used a chatbot. The students who wrote using only their own thinking showed the strongest brain activity and the best memory, while the chatbot group displayed the least [1]. In plain terms, the more of the work the student's own mind did, the more the brain seemed to be involved and the more the student retained.

That pattern is suggestive rather than settled. The comparison is described at the level of groups and general outcomes, not as a detailed causal story, so it is fair to say the non-AI writers showed stronger engagement and recall, not that AI use was proven to cause weaker memory [1].

Why writing it yourself may leave a stronger trace

A plausible explanation is that composing an essay without assistance forces the writer to generate ideas, choose words, and organize structure — effortful mental work that may be what makes the material stick. When a chatbot supplies text, that generative effort is reduced, and with it perhaps the mental workout that supports later recall. The study's finding that the chatbot group showed the least brain activity and memory is consistent with this idea [1], though the study does not establish the mechanism.

This is a hypothesis about effort, not a demonstrated cause. The evidence links less self-generated writing with weaker brain activity and memory, but it does not isolate which part of the writing process matters.

What the current experiments can and cannot tell us

A second line of work is a randomized controlled experiment among college students that records brain activity with fNIRS — a non-invasive method that measures blood-flow changes in the brain — while students write with and without generative AI [2]. This design can compare cognitive effort and writing performance between the two conditions, which is useful for understanding how demanding each task is.

Crucially, that experiment does not include memory tests or evaluate whether students can recall their own essays [2]. So it can speak to effort during writing, but not to the recall question. The memory result rests on the first study [1], and the effort result will come from a separate experiment still being run [2]. Keeping those two questions apart matters: a difference in brain activity during writing is not the same as a difference in what students remember afterward.

How much weight the answer can bear

The honest summary is that the direction of the evidence favors unaided writing for both engagement and memory, but the base is small. One study reports the brain-and-memory comparison [1], and the other is a protocol that has not yet addressed recall [2]. That means the finding should be read as an early signal worth following, not a settled rule about AI and learning.

For a reader deciding how to study or teach, the practical takeaway is modest: writing an essay yourself appears to involve the brain more and may support better memory of the content, while chatbot-assisted writing appears to reduce that engagement [1]. Whether that translates into lasting learning differences is a question the next round of experiments can answer.

Sources used in this answer

Related research questions