Token Anxiety
The new burnout is quiet, wired, and hiding behind your productivity.
There's a new kind of anxiety moving through knowledge work, and it has nothing to do with being behind. It happens when you're ahead.
You open your AI tools in the morning. Within an hour you've spawned three agents, started a fourth chat, and left two long tasks running in the background. You feel sharp. You feel fast. But underneath the sharpness there's a pulse of something else — a quiet pressure that every second you're not dispatching another task, another token of your subscription is dying on the vine. You could be building a company right now. You could be writing a book. The compute is literally idle. So you push another button. And another.
I've started calling this token anxiety: the wired, slightly compulsive state that arises when AI's raised ceiling makes every unused moment feel like a wasted opportunity.
It's not in your head
A Berkeley research team published an eight-month ethnography of a 200-person tech company in Harvard Business Review this February. Nobody was pressured. Nobody was told to hit new targets. People just started doing more, because the tools made more feel doable. The findings are almost banal in how predictable they are in hindsight: work bled into lunch breaks and late evenings, and to-do lists expanded to fill every hour AI freed up — and then kept going. The researchers were blunt about where this trajectory ends: workload creep, cognitive fatigue, burnout, and weakened decision-making. The productivity surge turns into lower-quality work and turnover.
A parallel study from BCG named the symptom directly: AI brain fry. When workers had to constantly supervise multiple AI systems or juggle several tools at once, mental strain spiked. When workers used AI to simply offload repetitive tasks, stress levels dropped.
Read that twice. The difference between AI as a relief and AI as a torment isn't the tool. It's the posture. Offloading calms you. Supervising empties you.
Why it burns you out, specifically
Work used to have two things going for it psychologically: friction and finitude. Friction meant your hands were in the thing. Finitude meant eventually you were done. AI quietly removes both.
The research on what happens when you strip effort out of work is unambiguous. The IKEA effect — coined by Norton, Mochon, and Ariely — describes how people consistently overvalue what they build with their own hands, even when the objective result is worse. It's downstream of the effort heuristic: the brain reads its own exertion as evidence that the output matters. Pull the exertion out and the reward circuit misfires. You get the object. You lose the meaning.
This is the hidden injury of token anxiety. You're not just tired. You're doing more work than ever and feeling less pride about any of it. That gap — big output, empty reward — is exactly what burnout is made of. It isn't a volume problem. It's a meaning-delivery problem.
There's a second, subtler injury. Your role has quietly shifted from maker to supervisor. Supervising is high-vigilance, low-agency work. You can't flow in it; you can only review. And because the AI works at machine speed, you are now the slowest component in your own workflow. Anxiety follows bottleneck status automatically, and no amount of clicking "continue" dissolves it. It compounds.
The token trap
The deepest trick of token anxiety is that it hides inside language you've always trusted. Productive. Leverage. Output. These words used to describe things you built. Now they describe things you merely started.
The unit of your day used to be the task. Then it became the hour. Now it's the token — and tokens don't know what they're for. They run regardless. They accumulate regardless. Your subscription doesn't care if you made something meaningful today. It just wants to be used.
So the diagnostic question that cuts through all of it is this: when was the last time you made something you were proud of, start to finish, with your own attention on it the whole way? Not approved. Not orchestrated. Made.
If the honest answer is "I can't remember," you're not under-productive. You're under-contact with your own work.
Fall in love with the friction
The way out isn't a productivity system. It's a relationship with friction.
Some work should be done by hand on purpose — not because AI can't do it, but because you need the exertion. The writing that forms you. The code you want to deeply understand. The ideas you want to actually own. Protect those. They are how you stay in contact with your own life.
Tokens aren't the metric. Hours aren't either. The only honest question at the end of a day is: did I make something I'm proud of? One essay written in friction beats forty approved in a fugue.
The voice that whispers "you could be doing more right now" is not the voice of ambition. It's downstream of a belief that your worth scales with throughput. It doesn't. Throughput was never the point. The texture of what you choose to care about is.
And the ceiling on your life isn't your capacity. It's your attention. Spend it on fewer things, deeper.
The cure for token anxiety is to fall in love with the friction of being imperfect — to build the imperfect thing slowly, to write the sentence only you could write, to let some of your compute sit idle while you live the day.
Not every second is an asset to deploy. Some are yours.