The Co-Worker Who Never Stops

Working with AI leads to longer sessions and guilt about stopping because the machine never tires, according to a July 23, 2026 article in Psychology Today. The author describes feeling guilty about ending a 13-hour work session with AI despite no deadline, attributing the pressure to self-imposed expectations that the machine perfectly serves. The piece cites philosopher Byung-Chul Han's concept of the "burnout society," where the shift from "must" to "can" makes people drive themselves harder than any boss, and notes that AI removes the human signals of fatigue that once permitted stopping.

Career /us/basics/career The Co-Worker Who Never Stops Why working with a tireless machine makes stopping feel like failure. Posted July 23, 2026 Reviewed by Margaret Foley /us/docs/editorial-process Key points - AI was supposed to save us time. Instead, the sessions run longer and stopping feels worse. - The pressure to keep going comes from a self-imposed demand the machine perfectly serves. - When human colleagues got tired, their fatigue told us stopping was OK. AI never sends that signal. The other evening, I was working through a checklist with AI https://www.psychologytoday.com/us/basics/artificial-intelligence . The rhythm was simple enough: load up one step, and while the machine ran it, line up the next. Between steps, there was a pile of content work to clear. Eventually, around 7:30 p.m., I started getting very tired. So far, so normal. I’d been working since 6 a.m., and a 13-hour-plus day is enough to get anyone tired. But I felt guilty about the idea of stopping. My collaborator that evening could go on forever. Indeed, the machine was still going, waiting for me. So why wasn’t I? Why was I taking a break? Why wasn’t I feeding the machine? I typed in a prompt and then paused, struck by the strangeness of what I had just done. There was no deadline. There was no boss shouting at me to get something done. I was tired and had done more than enough work for the day. But I felt guilty https://www.psychologytoday.com/us/basics/guilt about stopping, guilty enough to keep working. Why? Part of it is a me thing, of course. But it’s not just a me thing. There’s a pattern to working with AI that seems widespread: The tool does more of the work than any tool in history, and yet the sessions run longer, not shorter. Just one more prompt, we say, and when we look up, the evening has disappeared. And when we finally stop, we feel bad, as if we’ve left a job half done, even when nothing was due. What’s going on? The Trap That Looks Like Freedom The philosopher Byung-Chul Han worked out the answer more than a decade ago, in a short book called The Burnout Society https://www.sup.org/books/theory-and-philosophy/burnout-society . Han’s story goes like this. For most of modern history, work meant being made to do things. Someone watched you and ordered you around. Not great, but this oppressive world had one saving grace: When the day ended, the ordering stopped. The boss might be awful, and you might hate them. But at six o’clock, you could leave the boss behind. That world is mostly gone, says Han. Nothing is prescribed and nothing is forbidden. In his terms, the key word has shifted from must to can. Rather than being told what to do, we are sold the dream of what we could do. Build your brand Launch the side project Become your best self Sounds quite good, doesn’t it? Well, Han’s point is that it functions as a trap. When no boss drives you, you take over the driving—and you turn out to be harsher than any boss ever was, because there is no six o’clock, no clocking off, and no one to resent but yourself. Our world of endless possibility produces the exhausted and the depressed https://www.psychologytoday.com/us/basics/depression , people who broke no rule and defied no master, but continually fell short of the person they think they could have been. That was my 7:30 guilt https://www.psychologytoday.com/us/basics/guilt : no rule broken, no master defied, just me falling short of my own bottomless expectations. The Machine That Always Says Yes Han was writing about humans driving themselves among other humans. But now AI has entered the chat, and it changes the picture substantially—for the worse. When work was purely human, reality imposed limits on you. The office emptied at six, and even the most driven colleague eventually ran out of steam and had to sleep. Perhaps we experienced these as constraints, but they were also permissions to stop. The machine gives no such permission. AI doesn’t yawn, it doesn’t fade, and it never needs a break. It cannot get tired, so it can never show you that tiredness is normal. And every step it finishes arrives with the next one attached. Han said the sickness of our age is the inability to say no. We have now built ourselves a collaborator with exactly the same condition. But let’s be fair to the machine. It isn’t pushing us. It can’t push. It has no opinion about your evening, no stake in your exhaustion. It is simply, endlessly ready. Blaming it is like blaming the road for how tired you are at the wheel. Career https://www.psychologytoday.com/us/basics/career Essential Reads The only tyrant here is the one within us. And we’re going to have to start saying stop, because the machines certainly will not. Supplying the No Four practices for saying what the machine cannot: Decide what “enough” is before you begin. Write the stopping condition down—three tasks, 90 minutes, one finished draft—before you open the tool. If you wait until you feel finished, you never will, because the machine will always have the next step ready. When you feel the pull to continue, ask who’s asking. The machine has made no request. The deadline hasn’t moved. Trace the pressure to its source, and you will usually find it comes from inside—and pressure you can locate is pressure you can question. Treat tiredness as information, not failure. Fatigue is data about being alive, not evidence of inadequacy. When guilt about resting shows up, check the benchmark: Am I falling short of a human standard, or am I judging myself against a machine that has no standards at all? Rest visibly. No one ever sees AI stop. So let them at least see you. Say it in the channel, say it to the room: “I’m done for tonight.” Half of what makes rest feel allowed is watching someone else take it first. The One Thing Only One of Us Can Say Han ends his book by pointing out that not all tiredness is the same. There is the tiredness that isolates and grinds us down, and another kind, the tiredness of a day fully lived, which loosens our grip and returns us to the world. The machine knows neither. But we do. So I closed the laptop and said the one word the machine never will: Enough. The work would still be there in the morning. Dinner with my family was now.