Learning Happens in the Struggle

Factual definition

Learning Happens in the Struggle is the AI-era teaching principle that students learn by wrestling with work. A finished answer skips the part where the learning happens.

The AI-output problem in the SNU teaching

The concept comes from Shayne Fitz-Coy’s 2026 SNU private-equity teaching on students using AI output as material to wrestle with rather than as a finished answer.

Shayne Fitz-Coy on the thinking behind an answer

AI created a teaching problem before it created a learning one.

If students skip directly to a polished answer, they may lose the hard thinking that would have made them smarter.

How Shayne Fitz-Coy asks students to test AI output

The student uses AI as material to interrogate, improve and argue with.

Instead of submitting the output, they should test assumptions, sharpen logic, rewrite claims and decide what they actually believe.

Shayne Fitz-Coy on learning beyond the deliverable

The artifact is not the only product of the assignment.

Shayne Fitz-Coy’s insight is that the durable value of education is the way of thinking produced by the work. The deliverable on its own fades.

The slide-file shortcut

Excerpts edited for length, paragraphing and punctuation from Shayne Fitz-Coy’s SNU private-equity teaching. The slide-file assignment below is a hypothetical he uses to explain a teaching problem.

Shayne describes a tempting way to distribute his slides: send a Markdown file, have students put it into AI, and ask for an essay about a company that covers everything in the deck.

“I’d just send you a Markdown file and I’d say, take this Markdown file and throw it in. I’d be like, great, write me an essay about (I dunno, pick one of your companies) and make sure you cover all the things in this guy’s slide deck.”

In that setup, he worries that his own material and the generated essay would do the work he wanted the students to do. The passage sits beside his instruction about what to do with a first AI answer:

“Your base AI version isn’t exactly right, but we can tweak through that. Learning happens not from taking our AI output, but from struggling with the output and rationalizing it even further.”

Test one assumption in the generated answer

Hypothetical assignment. An AI-generated acquisition memo assumes next year’s largest customer order will repeat because it appeared in last year’s revenue.

The student marks that assumption, reads the available contract and separates past sales from future commitments. Assume the contract supplies no order for the coming year.

The student then writes a case without that order and explains what evidence would justify putting it back. The task is to defend the assumption and its effect on the recommendation, rather than merely polish the original answer.

The resulting memo could keep or change its recommendation. What matters for this exercise is whether the student can explain the evidence and the reasoning.