Struggle by Design: Protecting Productive Difficulty in an Age of Instant Answers

Ask a student why they used an AI tool for a task and the answer is often reasonable: it was faster, and the result was good enough. Ask a teacher what that student learned and the answer is less comfortable. Much of the learning that lasts is produced by effort that feels inefficient at the time. When instant answers remove the effort, they can remove the learning with it.

This is the central design problem for schools in an age of instant answers. The goal is not to ban the tools. It is to decide, on purpose, where difficulty should stay.

Why difficulty matters

Cognitive science has described this for decades under the name desirable difficulties. Retrieving something from memory, writing an explanation from scratch, trying a problem before seeing the method and spacing practice over time all feel harder than rereading or copying. They also produce more durable learning. The effort is not a side effect; it is part of how knowledge is formed.

Instant answers attack exactly these mechanisms.

• They replace retrieval. If the answer is one prompt away, the brain has less reason to store it.
• They skip the struggle. A student who never faces a blank page never builds the ability to start one.
• They create false confidence. A polished answer on screen feels like understanding, even when the student could not reproduce it alone.

I looked at the memory side of this in Memory Augmentation as a Learning Tool. The same logic applies to every task that AI can complete in seconds.

The equity dimension

Students with strong study habits, supportive homes and good schools tend to use AI as a second opinion. They attempt the work first and then check it. Students without those supports are more likely to use it as the first and only step, because it is the quickest route to a finished task and the one that least exposes what they do not know.

The gap that results is not a gap in access. It is a gap in how the tool is used, and schools can narrow it only by setting the conditions of use rather than leaving them to individual habit.

A simple way to decide where AI belongs

A single school-wide rule will not work, because the right answer depends on what the task is for. A practical test is to ask what the task is meant to build.

1. Foundational skill-building. If the aim is to build the skill itself (basic number sense, sentence construction, core vocabulary, first drafts of an argument), keep AI out, or limit it to feedback after the attempt.
2. Retrieval and recall. Low-stakes quizzes, closed-book recall and spaced review should happen without AI, because the act of recalling is the learning.
3. Applying and extending knowledge. Once the foundation exists, AI can be allowed as a collaborator, with students required to explain and justify what they used.
4. Production and polish. For formatting, translation support and presentation, AI can be an ordinary tool, as a calculator is once arithmetic is secure.

The same task may sit in different categories for different year groups. A tool that is fine for a Grade 12 research project may be harmful in a Grade 5 writing lesson.

What schools should do

1. Write down where AI is off, limited and open. By subject and year group, in plain language that students and parents can read.
2. Attempt first, then consult. Make “try it yourself before asking AI” a routine, and require students to show their first attempt.
3. Build in unassisted retrieval. Regular short, closed-book checks in class, not as high-stakes tests but as practice.
4. Assess process as well as product. Draft histories, annotated notes, oral explanation and live problem-solving give evidence of learning that a finished artefact cannot.
5. Teach students why. Explain the difference between feeling fluent and being able to do it. Students cooperate more readily when they understand the reason.
6. Protect time for slow work. Struggle needs minutes and sometimes days. If the timetable has no room for it, no policy will help.
7. Review each term. Ask teachers where students are plainly relying on AI to avoid thinking, and adjust the boundaries.

This links directly to the assessment question in Assessment After AI.

The question to ask in your next department meeting

Name one task in your subject where the struggle is the point. Is AI currently allowed in it? If nobody has decided, the students have.

Related Reading

• Is AI Making Us Smarter or Lazier?
• The Disappearing Peer
• AI Grading Tools: Efficiency Gain or Pedagogical Retreat?
• AI in Schools: Ban It or Teach It? The Question Is Wrong