Skip to content
Ideas for a more human education
Teacher's hands shield a brain-shaped plant held by a student from a reaching robotic arm.
Illustration: Human Learning Review, made with AI. It does not depict real people or events.
Opinion · Teaching & Learning

Protecting Student Cognition in the Age of AI

Source: EdSurge, . Summarized by Human Learning Review.

Summary

This EdSurge essay is a first-person account by a New York City teacher with 13 years in English and special education. It asks how schools should respond as generative AI takes on the thinking students most need to practice, and argues that educators should shift from preventing cognitive offloading to preserving critical thinking.

The author opens with 11th graders who stall on an analytical question about Kate Chopin, and with take-home work far more polished than what the same students produce by hand. The essay connects this to wider evidence: 2024 NAEP results showing 30% of eighth graders at Proficient in reading and one-third Below Basic, a 2024 Microsoft Research and Carnegie Mellon study linking heavier AI reliance with less critical evaluation among adult workers, faculty reports that incoming college students are less prepared, and a Brown economics professor whose class averaged nearly 96% on a take-home midterm and under 49% on a proctored final. The author says responsibility for preparing students rests with K-12 educators, and notes that many districts have expanded handwritten work and device limits while still teaching largely through laptops.

In the author's own classroom, the changes include:

  • Students annotate, discuss, gather evidence and form a claim before opening a device.
  • AI then generates counterarguments or flags reasoning gaps, and students critique its answers against their own.
  • Handwritten brainstorming, printed texts, device-free discussion and periodic device-free writing return as regular practice.
  • Students learn about AI through digital citizenship, which the author extends to what she calls cognitive citizenship.

Students resisted at first, the author reports, but over time they defended their reasoning, revised more thoughtfully, self-assessed before turning to AI and began trusting their own thinking. The essay concludes that older practices such as annotation and productive struggle should be paired with AI so it supports human learning instead of replacing it.

Main points

  • A New York City teacher with 13 years in English and special education argues that AI's greater risk is the erosion of students' ability to think independently, more than plagiarism.
  • The author cites 2024 NAEP data showing only 30% of eighth graders reached Proficient in reading, and says AI arrived just as these foundational skills were already weakening.
  • The author points to a Brown professor whose students averaged nearly 96% on a take-home midterm but under 49% on a proctored in-person final, suggesting polished take-home work can mislead.
  • A 2024 Microsoft Research and Carnegie Mellon study cited in the piece linked heavier AI use with less critical evaluation among workers, though the author notes it studied adults.
  • The author has students build their own claims before using AI to test them, and reports they grew more willing to defend their reasoning and trust their own thinking.

How the author's students bring AI into assignments

  1. Annotate the text independently

    Students mark up the reading on their own before opening any computer.

  2. Discuss and gather evidence

    Students talk through ideas with classmates and collect support from the text.

  3. Develop an original claim

    Each student forms their own argument while working through uncertainty.

  4. Consult AI as a challenger

    AI offers counterarguments, reasoning gaps, extra evidence, or alternative interpretations.

  5. Compare and decide

    Students critique AI responses against their own, and the final decision stays theirs.

Data and findings: EdSurge, “Protecting Student Cognition in the Age of AI.” Graphic: Human Learning Review.

The editor’s perspective

Which part of a student's thinking would I miss first if a machine did it for them?

That's the question under this EdSurge piece, which argues that educators should stop trying to prevent cognitive offloading and start protecting the thinking skills AI could replace.

I agree with the turn. Policing every handoff is a fight schools will lose, and I'd have to start by policing myself.

The issue is that we can only check work we could have done ourselves.

Hiring someone to do your taxes is sensible. Signing the return without being able to read it is a different decision, and from the outside the two look the same.

So the skill worth guarding is judgment, the ability to look at what came back and know whether it's right. Those who favor strict limits have a fair point that judgment grows from doing the work the slow way at least once. I'd keep that slow work, and I'd aim it at the moments where students have to decide.

This week, pick 1 assignment, have students write their own answer, then have them grade an AI's answer against it.

This page is a short summary. The full piece belongs to EdSurge.

Read the original at EdSurge
A little room for reflection

A more considered
view of education.

Ideas worth sitting with. Practical questions worth asking. Human Learning Review, in your inbox.