Five “Good” Uses of AI That May Actually Harm Student Learning

by Michael Keany

on Tuesday

Five “Good” Uses of AI That May Actually Harm Student Learning

School Leadership 2.0 | Summary for Educators

Based on:

Original Article: 5 ‘Good’ AI Use Cases That Are Actually Bad for Students

Authors: Paige Tutt and Stephen Merrill Source: Edutopia | October 16, 2026

Category: Innovation & EdTech | Instruction & Pedagogy

THE BIG IDEA

Artificial intelligence can help students complete assignments faster, produce more polished writing, and find information almost instantly. But what happens when these conveniences eliminate the very thinking students need to learn?

Tutt and Merrill explore an increasingly important distinction: AI can improve the quality of student work while simultaneously weakening the learning process that produces it.

Activities such as brainstorming, developing thesis statements, researching, summarizing, and writing drafts are not simply preliminary steps toward a finished assignment. They are essential intellectual exercises that develop creativity, reasoning, memory, independence, and communication.

The authors caution that even seemingly responsible uses of AI may undermine learning when students outsource these foundational processes before mastering them.

The central question for educators is no longer simply whether students should use AI. It is the thinking tasks that students must learn to perform independently before AI becomes a productive partner.

FIVE KEY TAKEAWAYS

1. Brainstorming: Creativity requires productive struggle.

AI can rapidly generate ideas, but its suggestions often converge on predictable patterns. Students who rely on AI to brainstorm may miss opportunities to develop originality, explore unusual connections, and tolerate uncertainty.

2. Thesis writing: Developing an argument builds reasoning.

Asking AI to formulate thesis statements removes an important stage of interpretation and critical thinking. Students need to analyze evidence, establish positions, and decide what makes an argument defensible.

3. Conducting research: Finding information is not the same as evaluating it.

AI-generated research suggestions can contain unreliable sources or misleading claims. Students who allow AI to select their evidence may never develop the ability to investigate credibility, compare competing perspectives, or recognize weaknesses in an argument.

4. Summarizing: Remembering requires mental effort.

AI-produced summaries are convenient but can encourage passive studying. When students summarize material themselves, particularly from memory, they practice retrieving, organizing, and consolidating knowledge.

5. Writing first drafts: The struggle is part of learning.

Turning incomplete thoughts into coherent prose is a demanding intellectual exercise. Students who let AI compose their initial drafts risk becoming editors of machine-generated writing rather than independent thinkers and writers.

WHY IT MATTERS FOR SCHOOL LEADERS

The article challenges a growing assumption that using AI responsibly means avoiding plagiarism and verifying information.

Those safeguards matter, but they are insufficient.

A student may use AI transparently, follow classroom rules, and submit factually accurate work while still bypassing essential learning.

The distinction is especially important because education differs from professional work. An experienced adult may appropriately use AI to save time. A developing learner, however, may need to perform that same task independently to build competence.

The authors draw on research and expert commentary, including cognitive scientist Daniel Willingham and MIT education researcher Justin Reich, to emphasize that schools need clearer instructional guidance. The long-term effects of AI use are still being investigated, and educators should avoid treating all AI assistance as either beneficial or harmful.

LEADERSHIP ACTIONS

Principals and instructional leaders can translate these findings into practical school policies.

  • Establish “think first, AI second” expectations. Require students to generate initial ideas, attempt solutions, or produce first drafts before consulting AI.

  • Identify protected learning tasks. Have departments determine which skills students must demonstrate independently at each grade level.

  • Redesign assessments. Incorporate handwritten drafts, oral explanations, classroom discussions, and opportunities for students to defend their reasoning.

  • Teach appropriate AI assistance. Students might use AI to critique an argument, suggest revisions, or identify gaps in understanding after completing the foundational work themselves.

  • Provide teacher professional development. Help teachers distinguish between AI that supports learning and AI that substitutes for learning.

A useful guiding principle is: Use AI to extend student thinking, not replace it.

LEADER REFLECTION

Consider discussing these three questions at your next faculty or department meeting:

  1. Which intellectual tasks in our classrooms are students increasingly outsourcing to AI?

  2. How can we distinguish between students who produce better work and students who actually learn more?

  3. What evidence should students provide to demonstrate that the thinking behind their assignments is genuinely their own?

THE BOTTOM LINE

AI may become one of the most powerful educational tools of this generation. But its greatest danger may not be cheating or inaccurate answers. It may be the gradual disappearance of intellectual effort from everyday schoolwork.

Great teaching does not always make learning easier. Sometimes it protects the productive struggle that makes genuine learning possible.

Original Article

Source: Paige Tutt and Stephen Merrill, “5 ‘Good’ AI Use Cases That Are Actually Bad for Students,” Edutopia, October 16, 2026.

------------------------------ 

Prepared with the assistance of AI software OpenAI. (2026). ChatGPT (5.2) [Large language model]. https://chat.openai.com