Technolandy

Day 166 (of 2025/26) first little “test” of AI & learning this #tEChursdAI

Sigh. A happy sigh, but still a sigh.

I ran a small test around AI and learning this week, using an “AI as primary” collaboration model, and I’m left sitting with mixed feelings about the results.

The learner outcome was… not good.

A student was willing, even happy, to lean heavily on AI for a required graduation course they strongly disliked. (A teacher colleague immediately flagged what might be problem #1: perhaps these inquiries need learners who actually like, or at least tolerate, the subject matter.)

For one unit, the process was intentionally simple:

Read the question.

Prompt an AI tool for the answer.

Read the response.

Submit it to the online portal.

That was the entire workflow.

It raised a question I was genuinely curious about: would that kind of repeated exposure, reading and reproducing AI-generated information, create enough familiarity for short-term recall?

To stir the pot further, the student was asked not to review any material before the test, and they would not have access to the usual supports and resources often available in our online environment.

They were not thrilled.

I principal-promised the assessment would only help them. This unit had become less about their course completion and more about what my education team and I might learn from the process.

The result?

Absolutely abysmal.

The assessment happened roughly 72 hours after final submission, and recall was remarkably weak.

I wanted to know whether simple exposure, read, paste, submit, would create enough cognitive residue to move the needle for a disengaged learner in a course they openly despised. A course they were only completing because it was required for graduation.

The answer, at least in this tiny test case, appears to be no.

And that leaves me oddly conflicted.

Part of me is relieved. It reinforces what I have believed all along: AI is a remarkable collaborator with learning, but not an automatic shortcut to learning when it is doing the cognitive heavy lifting for the learner.

Part of me is disappointed.

If I’m honest, there was a small part of me that hoped the student would perform similarly to when they had worked more independently. That perhaps mere exposure might create some measurable retention.

My teacher collaborator sees this differently. They view it as a useful baseline: if this is the floor, an unmotivated learner, a disliked course, minimal engagement, then what happens when curiosity is present? What happens when the course matters to the learner? What happens when AI is used as a thinking partner rather than a vending machine for answers?

That feels like the real next question.

Still, today’s little experiment left me more convinced that while AI can absolutely amplify learning, it cannot replace the messy, uncomfortable, cognitively demanding work of actually doing the learning.

Turns out there’s still no copy-and-paste shortcut through understanding.

At least not yet… 😜