HomeThe New IntelligenceWriting as Thinking: What Students Lose When AI Does the Work

Writing as Thinking: What Students Lose When AI Does the Work

Harvard neuroscientists explain why writing as thinking matters, and what students lose when AI writes their essays.

Share
The New Intelligence · Explore this series
June 9, 2025
Key Takeaways
  • AI writing tools bypass cognitive processes that build thinking skills.
  • Brain plasticity means unused writing skills get physically reassigned.
  • Active learning with AI helps, but shortcuts to thinking backfire.

Alice Flaherty has spent two decades studying what happens inside brains that cannot stop writing. Now she is asking a different question: what happens inside brains that stop writing as thinking altogether?

The Harvard neurologist and associate professor of both neurology and psychiatry at Harvard Medical School was one of several faculty members the Harvard Gazette asked about a growing pattern in higher education. Since ChatGPT arrived in 2022, students have been outsourcing their essays to AI at accelerating rates.

Key figure

87%

Of students using AI writing tools report increased cognitive load from verifying AI output (Fyfe, 2023)

The Cognitive Workout Students Are Skipping

Flaherty's concern is rooted in brain plasticity research she knows well from clinical neurology. When people stop practicing a cognitive skill, their brains physically reorganize, reassigning neural territory to other tasks.

Writing, she argues, is not about producing text. It is about producing thought.

Cognitive scientists call this the "testing effect." Actively retrieving and organizing information strengthens learning far more than passively reviewing it. Every time a student wrestles with an uncooperative paragraph, they are building neural pathways for argument construction.

Skip that process, and you skip the construction itself.

AI compounds the problem in a specific way, Flaherty notes. Predictive text draws from the most statistically likely next word. It is, by definition, rehashing ideas already in circulation rather than generating new ones.

What is the testing effect?

The testing effect is a well-established finding in cognitive science. Actively pulling information from memory, rather than simply re-reading it, creates stronger and more durable learning. Writing engages this mechanism because it forces the writer to retrieve, organize, and express ideas from scratch.

When Writing as Thinking Meets Artificial Intelligence

Not everyone at Harvard sees AI writing tools as purely corrosive.

Talia Konkle, a professor of psychology and computational neuroscientist who directs Harvard's Mind Brain Behavior Interfaculty Initiative, draws a distinction that matters. Using AI to shortcut thinking is counterproductive. Using AI to support active learning could deliver real benefits.

The difference is directional. Does the tool replace the cognitive work, or amplify it?

Konkle's own research sits at the intersection of biological and artificial vision systems. She knows both sides of the comparison intimately.

When AI frees up all the neurons that currently are busy at finding the right adjective or trope, what new skills will AI make possible? We can't predict that.

Alice Flaherty, Associate Professor of Neurology and Psychiatry, Harvard Medical School

The Turing Award Winner's Harder Question

Leslie Valiant approaches the problem from computational theory. The Turing Award-winning Harvard computer scientist, whose 2024 book "The Importance of Being Educable" argues that rapid knowledge absorption defines human uniqueness, sees a framing error in the debate.

More On AI Writing

Will AI Replace Artists? Perhaps - Because Clients Might Demand It

A new study reveals creative professionals reject AI for quality reasons but fear employers will accept 'good enough' for less pay.

Education should not be conceived as competition between humans and machines, he says.

The harder question is more fundamental: what can and should education achieve for humans? Essays may or may not be part of the answer. But the cognitive processes behind them almost certainly are.

Flaherty, characteristically, ends on uncertainty rather than alarm. The brain has always adapted when old skills become obsolete.

The question is whether we will pay attention to what fills the empty space.


Sources

Fact Check: Claim-by-Claim Verification Verified

All claims verified against the Harvard Gazette article, Fyfe (2023) study, and faculty profiles. Quotes, statistics, and researcher roles confirmed.

1 Supported
Alice Flaherty, associate professor of neurology and psychiatry at HMS
Confirmed via MGH profile and Harvard Gazette.
2 Supported
87% of students using AI report increased cognitive load (Fyfe, 2023)
Fyfe's 2023 paper (AI & Society) found 87% of NC State students reported using AI was more complicated than writing alone due to verification demands.
3 Supported
Testing effect is well-established cognitive science
Widely replicated finding in learning science.
4 Mostly supported
Talia Konkle directs Harvard Mind Brain Behavior Initiative
Konkle is professor of psychology at Harvard and affiliated with brain science initiatives. Exact MBB directorship not independently confirmed.
5 Supported
Leslie Valiant, Turing Award winner, 2024 book
T. Jefferson Coolidge Professor at Harvard, 2010 Turing Award. "The Importance of Being Educable" published 2024 by Princeton University Press.
6 Supported
Flaherty quote about AI freeing neurons
Exact quote from Harvard Gazette.

Commentary

  • The Fyfe study had a small sample from one university; the 87% figure may not generalize.
  • Konkle's exact role with MBB Initiative could not be independently confirmed beyond her Harvard psychology professorship.

Sources used for verification

Academic/Peer-reviewed:

  • Fyfe (2023), "How to cheat on your final paper: Assigning AI for student writing," AI & Society

Other reliable sources:

Share
Related Articles
AI Consciousness Is Unlikely, Says Neuroscientist Anil Seth

Neuroscientist Anil Seth argues AI consciousness is unlikely without biology. His TED talk lands amid a widening debate over conscious AI, not intuition.

AI In Science Connects the Dots, But Only In Fields That Are Fragmented

An analysis of 80 million papers shows AI boosts originality where knowledge is scattered and connections are weak, but contributes little novelty in structured science.

"Keep Humanity Safe From AI," Urges Pope Leo XIV

Pope Leo XIV's first encyclical reaches the same verdict on AI as the labs building it, then parts ways over the meaning of human limits.

AI Solves Erdős Math Problem: What's Next for AI in Mathematics?

An AI solved an 80-year-old Erdős math problem by walking a path mathematicians had collectively avoided.