The Superpower That's Burning Us Out
#040: How the novelty-reward loop that makes AI tools so useful for neurodivergent brains also exhausts us.
I’ve been noticing something about myself that I’m not sure I like.
The loop goes like this: I think of an interaction concept, feed it to an AI tool, and watch it take shape in minutes. The concept doesn’t even have to be good. It could be half-formed and ugly, but the speed of seeing an idea become something real, right now, in front of me, is intoxicating. One idea leads to three. Three leads to a rabbit hole that eats the rest of my evening. I look up and it’s 1 AM and I’ve built something I didn’t plan to build, forgotten to eat dinner, and I’m somehow both wired and completely drained. This happened to me the other night and then I couldn’t sleep; I went almost 24 hours straight without rest.
If that sounds familiar, there’s a decent chance you’re neurodivergent and you’re not alone. I have ADHD.
There’s a growing body of conversation, research, and personal testimony pointing to a pattern: neurodivergent people, particularly those with ADHD, are finding AI tools unusually magnetic — not just useful, but the kind of thing you have trouble putting down. The novelty-reward loop that AI enables maps almost perfectly onto how ADHD brains are wired to seek stimulation, and the result is a tool that feels like it was designed for you, right up until it starts wearing you out.
The interest-based nervous system meets infinite novelty
ADHD brains don’t run on importance, they run on interest. Motivation ignites through novelty, challenge, urgency, or personal fascination, not because something shows up on a to-do list marked “high priority.” This is well-documented in clinical research and anyone who’s ever spent four hours reorganizing their desk instead of answering one email already knows it intuitively.
AI tools slot directly into that wiring. Every prompt is a fresh stimulus. Every response is a small surprise. The iterative nature of working with an LLM, where you refine, redirect, and regenerate, creates exactly the kind of variable reward pattern that ADHD brains find almost impossible to disengage from. It’s the same mechanism that makes slot machines work, except this time the output is a functioning prototype or a first draft of something real, which makes it feel productive rather than wasteful, which makes it even harder to stop.
A UK Department for Business and Trade study found that neurodiverse workers were 25% more satisfied with AI assistants than their neurotypical colleagues and were more likely to recommend the tools to others. That’s a real gap, and it makes sense when you think about what these tools actually do. They reduce the friction that neurodivergent people have fought against their entire careers — the blank page, the executive function overhead of just getting started, the energy it takes to organize your thoughts before you can act on them. AI handles that scaffolding so the interesting part can start immediately.
Jeff Putz, a developer who was diagnosed with ADHD and autism spectrum disorder a few years ago, wrote about his experience coding with AI tools and landed on something that resonated: the tools help, but they also create new problems. During the 30-plus seconds it takes for code generation, he described the urge to do something else as “overwhelming,” estimating his attention shifted “at least a hundred times” within five minutes. The irony of needing coping strategies for a tool designed to help you cope wasn’t lost on him.
The productivity trap
Here’s where it gets uncomfortable. Adults with ADHD already lose an estimated 22 days of workplace productivity per year due to symptoms like disorganization, forgetfulness, and impulsivity. That’s essentially trying to squeeze a full year’s work into eleven months, which creates a baseline of chronic overextension that most neurodivergent professionals just live with. AI tools offer a way to close that gap, and they do, but they also introduce a new failure mode: the inability to stop when the gap is closed.
Because ADHD hyperfocus doesn’t come with a built-in off switch. When the task is interesting enough, you don’t stop at “done.” You keep going until the interest burns out or your body gives up first. And AI tools, by continuously generating new threads to pull on, keep the interest alive far longer than it would survive on its own. The technology that helps you function during the day can quietly become the thing that keeps you going at 2 AM when you should have been asleep three hours ago.
Some people have started building explicit boundaries. One person I came across in my research described a personal rule: no AI after 11 PM, because the tools would stimulate new ideas precisely when they needed to be winding down. That’s a coping strategy for a coping strategy, which tells you something about the dynamic at play.
The scaffold can become the cage
An EY study of over 300 neurodivergent employees found that 85% believe AI workplace tools can create more inclusive environments, with 91% viewing them as valuable assistive technology. Those numbers are encouraging and, I think, accurate. For people who’ve spent their careers fighting their own executive function just to do the baseline work, having a tool that handles the organizational overhead changes what’s possible.
But there’s a less-discussed failure mode that a piece on Medevel captured well: ADHD brains operate with lower baseline dopamine, which makes them more vulnerable to the variable reward patterns that modern tools are built around. The same mechanism that makes AI useful to you also makes you more susceptible to it. Extended use of highly stimulating tools creates dopamine downregulation, where normal, non-AI-assisted work starts to feel impossibly dull by comparison. That’s not laziness, that’s a neurological withdrawal state, and it’s something worth being honest about.
The design of AI tools doesn’t help. Most of them are built for engagement, not for sustainable use. There’s no “you’ve been at this for four hours, maybe take a walk” prompt. There’s no friction designed into the experience to help you disengage. These tools are built to keep you in the loop, because that’s what makes them useful, but for people whose brains already struggle with disengagement, “useful” and “compulsive” live uncomfortably close together.
What I’m still working out
I don’t have a clean resolution for this. I know that AI tools have made me measurably more productive. I know they’ve helped me build things I couldn’t have built alone, or at least not at the speed I’ve been building them. I also know I’m tired in a way that feels different from normal work fatigue, the kind of tired that comes from running your brain at redline for hours because the tool keeps feeding you reasons not to stop.
What I keep coming back to is that the conversation around neurodivergence and AI has been dominated by the “superpower” framing. And there’s truth in it. Neurodivergent brains are wired for the kind of associative, nonlinear thinking that gets the most out of these tools. But “superpower” is doing a lot of work in that framing, and it’s glossing over the part where the power source isn’t unlimited. Hyperfocus is a resource that depletes, and AI tools are unusually good at draining it.
The more useful framing might be this: AI tools are the best assistive technology many neurodivergent people have ever had, and they’re also the most seductive burnout accelerant. Both things are true, and pretending the first cancels out the second isn’t helping anyone.
If you’re in this boat, you probably already know. The question isn’t whether these tools work for you. It’s whether you’re working with them, or whether they’re working you.
If this resonated, I’d like to hear how you’re managing it. I’m still figuring it out myself.
Sources and further reading: