AI Didn't Create the Distraction Problem. But It Just Made the Stakes Infinitely Higher.

In the last post, we made the case that distraction is a system problem. It is something that disengaged learning environments specifically help create. The device is often an escape hatch from a learning experience that wasn't worth paying attention to in the first place.

That argument was about the present.

This one is about what's coming.

Because if the distraction problem was urgent before AI, it is now a five-alarm fire. Not because AI is inherently bad for learning (it isn't when used correctly), but because AI dropped into a system already struggling with engagement, will accelerate every existing failure mode while adding several new ones we don't fully have language for yet.

Understanding why requires being honest about what AI actually is, what it does to cognitive effort, and what that means for a generation of learners who are now growing up with instant answers available to every question they've ever had.

Buckle up!

AI Is the Ultimate Distraction Machine (If We Let It Be)

The attention economy we described last time works by making consumption effortless. Scroll, swipe, autoplay. The experience is frictionless by design, because friction is the enemy of engagement-by-passive-consumption.

AI takes this even further. It doesn't just distract you from thinking — it replaces the thinking entirely. You don't have to wait for an answer, search for it, evaluate competing sources, or wrestle with a hard concept. You ask, and it answers as confidently and as quickly as possible.

That is genuinely useful for some things. But for learning, it introduces a problem that is more serious than social media…

It doesn't just compete for your attention, it volunteers to do your cognitive work for you.

A USC study released in early 2026 surveying 1,000 college students found that most use AI tools for what researchers call "executive help" (i.e. seeking quick solutions with minimal effort) rather than to deepen their understanding. When something makes thinking easier to outsource, we outsource it. When the outsourcing feels productive (which is does with AI because you end up with a completed essay or a solved problem) then it's very difficult to notice that the learning didn't happen.

This is the distraction problem in a newer form. Not "I looked at my phone instead of paying attention." But "I generated an answer instead of building understanding."

The Research on What This Does to Thinking

The evidence here is early but consistent enough to take seriously.

A 2025 study of 666 participants published in Societies found a significant negative correlation between frequent AI tool use and critical thinking scores, mediated by what researchers call cognitive offloading. Younger participants showed the strongest effects with a higher dependence on AI tools and lower scores on critical thinking assessments compared to older participants.

A CHI meta-analysis of 17 studies found that while AI produces meaningful overall learning gains, those benefits weaken or reverse for higher-order thinking skills. Obvious to those of use who use AI, this is precisely because offloading the thinking prevents the practice that builds the skill.

Also, a lab study comparing students who used ChatGPT for essay writing against those using human coaches, a structured toolkit, or no support found something interesting. The ChatGPT group actually produced the highest quality essays. But (an important “but”) they showed no gains in learning, motivation, or interest. They also copy and pasted far more frequently than other groups. The output ended up being better, while the learning was not.

This is the paradox of AI assistance where we often see better results but worse thinking.

None of this means AI is the enemy of learning. Yet, when AI is used in ways that go around the effort-reward cycle of productive struggle it becomes an issue. When learners miss out on working through something hard, overcoming it, internalizing it and using that knowledge…we have a huge problem.

This Is the Engagement Crisis, Accelerated

In my last article, we argued that distraction is partly a symptom of disengagement — that learners reach for the escape hatch when the learning itself isn't worth staying for. The solution is building learning experiences compelling enough that the device becomes less interesting by comparison, or is used with a purpose in the experience.

AI raises the stakes on that argument dramatically.

Because now the technology doesn't just distract from learning, it instead produces something that looks like learning. A student who checked Instagram during a lecture wasn't producing anything. A student who had AI write their essay has a finished essay. The output exists, and the grade might follow. The learning did not happen, but nothing in the system signals that clearly.

This is disengagement that is much harder to detect, much harder to address, and much easier to rationalize than staring at a phone.

If we were already struggling to build learning environments that demanded and rewarded genuine cognitive engagement, we are now in a race against a tool that makes bypassing that engagement invisible.

The Real Threat Isn't Cheating

Most of the K-12 and Higher Ed conversation about AI in education has focused on academic integrity. Totally understandable….but also looking at the wrong problem.

The cheating framing assumes the issue is moral with students taking shortcuts they shouldn't. This is true. It should be discussed. When Princeton changes a 133-year honor code because of AI cheating, that is a moral problem.

Students, however, have always cheated. There aren’t many more cheating now than were before (if you can believe it). The newer and bigger threat is intellectual atrophy. Researchers describe it as "metacognitive laziness", where we see a decline not just in critical thinking output but in the inclination to think critically. When good answers are always available, the habit of working through a problem independently weakens.

Think about what this means over time. A student who uses AI to shortcut the thinking work of high school doesn't arrive at college or a career having banked years of practice analyzing arguments, forming positions, sustaining effort through difficulty, and revising their thinking in response to pushback. They arrive having produced outputs.

The outputs don't transfer. The practice would have.

Learning is fundamentally a process, not a product. AI, at its most passive, inverts the whole experience by delivering the product while bypassing the process entirely.

Here's the Other Half of the Story

This is where the distraction-and-AI argument has to be honest rather than alarmist.

The same research that documents the risks of AI in learning also documents the conditions under which it doesn't produce those risks AND even produces meaningful gains. A December 2024 review of 69 studies concluded that AI use can enhance academic performance, motivation, and deeper cognitive engagement when the design of the learning experience requires the learner to do real thinking alongside it.

The difference is whether the learner is positioned as an active participant or a passive consumer.

Stanford researchers studying this question put it usefully, “when students have to edit, curate, evaluate, and build on AI-generated material — rather than simply accept it — the cognitive bar may actually rise". The thinking required to work critically with AI output is genuine, demanding, and different from what was previously required. So it is actually a different kind of thinking.

This is the version of AI that engagement-focused learning can make possible. Using AI as a collaborator that raises the quality and complexity of the cognitive work learners actually do.

This looks completely different with various ages and levels of learners. But the central educational question of this decade is whether we will design learning experiences that demand that kind of engagement, or whether we will allow AI to do what the attention economy already started…

What This Asks of Learning Environments

The last post ended with the idea that the most powerful response to distraction is learning that's worth showing up for.

AI makes that more serious than ever.

Learning environments that will serve learners well in an AI-saturated world are not environments where AI is banned.

We need to have discernment and judgement on when/where to use. But, banning leads to no discernment skills being built at all (by the adults or learners as they grow older).

What's needed instead are learning environments designed around the things AI cannot do for you:

• Form your own position and defend it under pressure.

• Decide what matters and why.

• Connect ideas across domains in ways that are genuinely yours.

• Build relationships with ideas over time, revising your thinking as you encounter new evidence.

• Sit with uncertainty long enough to develop real judgment.

These are the durable cognitive skills that become more valuable as AI gets better at producing fluent outputs on demand. They are also exactly what compliance-oriented learning environments have never been good at developing.

The distraction crisis and the AI crisis are pretty much the same crisis. Both are asking whether learning environments can make deep thinking the compelling, central, rewarding experience of being a learner. Or will we stick with the grades and compliance and see who wins that battle? If that’s the case, it is hard not to bet on AI.