Easy to lump these together because they often show up in the same person at the same time, using the same tools. Worth pulling apart anyway, because the fix for one doesn't touch the other.
Dependency (most of what this site is about elsewhere) is a skills and habits problem: relying on AI to the point where your own debugging, comprehension, or verification muscles have atrophied. The fix is deliberate, unaided practice — rebuilding a capability that's still there but out of shape.
Burnout is a different animal entirely: chronic depletion from the pace and texture of the work itself, independent of whether your underlying skills are intact. And AI-heavy coding workflows produce a specific, somewhat under-discussed flavor of it.
Why AI-assisted work burns people out in a distinct way
Traditional coding burnout usually comes from sustained cognitive load — hard problems, long hours, high-stakes deadlines. AI-assisted burnout often comes from something almost the opposite: a review treadmill. Reading a constant stream of generated code, prompting, re-prompting, evaluating output, correcting subtle mistakes — hour after hour of shallow-but-relentless attention, rarely the deep, absorbing focus that used to come with solving something difficult yourself. Less intense in any given moment, but harder to switch off from, because there's always another generation to check.
There's also a quieter contributor: the loss of the specific satisfaction that used to come from figuring something out yourself. Review work doesn't deliver that the same way, even when it's objectively productive — and a job that's lost its satisfying parts wears on people faster than one that's simply hard.
How to tell which one you're actually dealing with
Ask what a day off from AI tools would do to you. If the honest answer is relief — genuinely lighter, even if slower — that points toward burnout from the review treadmill, not skill loss. If a day off would leave you stuck, unable to make progress at all, that points toward dependency, not burnout. Plenty of people have some of both, but they respond to different fixes, and treating burnout with a "build your skills back up" plan just adds more grind on top of what's already exhausting someone.
What actually helps burnout specifically, as distinct from dependency
Structuring in real breaks from the review-and-correct loop, not just from AI tools generally. Deliberately doing at least some work that isn't just verification — a task worked through by hand, start to finish, which restores some of the "figuring it out" satisfaction that review work doesn't provide. And being honest that speed isn't free even when it doesn't cost you a visible skill — it can still cost you the parts of the job that made it worth doing.
If it's dependency, not burnout, the test can tell you where →
Related reading: AI Coding Addiction Symptoms · The AI Coding Detox Challenge