AI probably inhibits flow states

Earlier this year, I wrote a series of brief white papers at work that attempt to contextualize AI in education through several foundational, well-studied phenomena of educational psychology. The goal of these papers was to provide our school division’s leadership with hype-free, unsensationalized lenses through which to consider both the affordances and drawbacks of AI in the classroom.

Well, I’m presenting this work at the upcoming MERC Summit on June 25th (if you live in the Richmond, VA area, you should definitely consider going – it’s one of my favorite conferences), and so while I’ve been prepping for that presentation, I figured why not turn those briefs into blog posts.

So this post will be the first in a series that does just that. I’m going to start with flow theory, but before I get into it, some disclaimers:

1. For all of the ed psych nerds: these are going to be short blog posts. Not full-ass scholarly articles. I’m going to simplify and omit some details. Hopefully not the most important ones.
2. Relatedly, I finished my PhD in 2019, and although I still dabble in the literature when it’s relevant to me, I’m not, like, reading current issues of Educational Psychology Review while eating my yogurt every morning. I kinda doubt there have been radical changes to these theories since 2019, but please let me know if that’s not true.
3. These blogs are mostly going to be centered around student learning and motivation, and will therefore focus on student use of AI, although I might touch on teacher use if it feels relevant.

With that out of the way, let’s go. I’ll start by providing a broad overview of flow theory, then I’ll describe how AI might interact with some of the foundational concepts of flow. 

Flow is the psychological state of total immersion in an activity. When in a state of flow, a person feels focused, energized, and intrinsically motivated. Flow is an active state (rather than a passive one) where the activity one is engaged in is rewarding in and of itself.

Mechanistically, flow occurs when a task consumes all of a person’s (finite) attention. Like, literally all of it. Since humans can only process a limited amount of information at once, this complete devotion of attention leads to flow’s characteristic “loss of self” (e.g. “where did that last hour go?”). To maintain a state of flow, a person needs to feel agency – a sense of control – over the task they’re engaged in. Without this sense of agency, the state of flow collapses. So, for instance, although someone might binge five episodes of The Pitt in one night, I wouldn’t characterize this as being in a flow state since they aren’t exercising agency and since they’re not actively engaged.

For a person to enter a flow state, the activity must meet three conditions:

1. It must have clear goals and progress. A person must understand what needs to happen at each step. E.g. I’m writing an essay, so I need to plan, draft, then revise.
2. It must provide clear and immediate feedback. This allows for real-time adjustments. E.g. This sentence sounds better after I revised it.
3. It must strike a balance between the task’s perceived difficulty and the person’s perceived skill. It should stretch a person’s skill without snapping it.

Flow can be a tenuous state, though. In academic settings, one of the biggest threats to flow is that the balance between challenge and skill is, well, imbalanced. When that happens, it can lead to:

1. Anxiety (if challenge far exceeds skill);
2. Boredom (if skill for exceeds challenge);
3. Apathy (if skill and challenge are low)

Flow and AI

Overall, AI tools seem far more likely to inhibit flow in students than to promote it. But let’s start with the positive.

The most likely way in which AI could promote flow is by providing scaffolding to get a student back into the optimal “challenge vs skill” balance, particularly if this balance is disrupted by a peripheral challenge (rather than one central to the task at hand). If used appropriately, an AI tool could help a student overcome a peripheral sticking point so they can more fully engage in the core challenge. For example, if a student forgets a specific exponent rule while working on an activity involving factoring equations, a tool like ChatGPT can provide instant feedback, allowing them to continue with the assignment without being derailed by forgetting a single rule.

Conversely, here are two ways that AI seems likely to inhibit flow state in students.

First, if used inappropriately, AI tools can reduce a student’s sense of agency. In the most extreme cases, this leads to students having AI tools simply do their work for them. But even when students use these tools “appropriately,” they are necessarily ceding some agency to the tools. If they’re not very very intentional about how they interact with AI during an activity, it can be easy to cede too much agency to the AI, which will lead students to become more of a passenger than the driver of the activity. If you’ve ever used any of these tools yourself, you know how easy this is. You think, “oh, I just want to check to see if the logic of this paragraph is sound,” and then, without intending to, you’ve been chatting with Gemini for 30 minutes and it has rewritten half of your argument. It’s also told you how insightful your questions are (several times).

Second, the mechanics of interacting with most AI tools – opening tabs, prompting, copy/pasting responses, asking for clarification, etc – necessarily incur some “switching cost.” Because human attention is a finite resource, these secondary tasks – even simple ones like opening a new tab in Chrome or clicking a bookmark – divert attention away from the primary task (e.g. drafting an essay). Recall that a flow state requires total immersion, and so any diversion of attention necessarily inhibits entering a state of flow.

Of course, “entering a state of flow” isn’t the end-goal of any learning activity. Flow is a form of deep engagement with a task, and engagement supports learning, but it’s obviously not the case that students can only learn if they’re experiencing flow. Nevertheless, it’s something we probably want students to experience when they’re working deeply, and so before we allow them to use AI to support writing that essay or coding that app or making sense of that physics experiment, we ought to acknowledge that AI tools are likely a net negative for flow.