AI Fatigue: Why Keeping Up With AI Feels Like a Second Job

AI fatigue is the exhaustion of trying to keep up with artificial intelligence: the weekly model releases, the new tools at work, the pressure to upskill, and the steady rise of AI-generated content you have to wade through. It is not a failure of curiosity or a sign you are falling behind. It is a predictable response to a demand that has no finish line.

Most writing about AI and stress focuses on one of two things. Either the fear that AI will take your job, which we covered in AI anxiety and the nervous system, or the acute overload of supervising AI tools all day, which we covered in AI brain fry. AI fatigue is the third thing. It is the treadmill.

What is AI fatigue?

It helps to separate the three kinds of tired, because they need different responses.

| | What it feels like | Where it comes from | What helps | |---|---|---|---| | AI anxiety | Dread about the future, job insecurity | Threat of replacement | Safety, agency, realistic assessment | | AI brain fry | Fog, headaches, slow decisions by mid-afternoon | Supervising too many tools at once | Fewer tools, batched review, a hard stop | | AI fatigue | Overwhelm, cynicism, feeling permanently behind | The endless demand to keep up | Selective ignorance, a slower intake rhythm |

You can have all three. But AI fatigue is the one most people misread as a personal shortcoming. If everyone else seems to be on top of the latest release, the reasoning goes, the problem must be you.

It is not you. It is the pace.

AI fatigue tends to show up in a few recognizable ways:

• A small sinking feeling every time someone says "have you tried the new one?"
• Saved tutorials, bookmarked threads and unread newsletters piling up like a second inbox.
• Cynicism about every launch, even ones that might genuinely help you.
• Reluctance to learn anything new because it will be out of date by spring.
• A sense that rest is irresponsible while the field is moving this fast.
• Feeling behind even though, by any reasonable measure, you are keeping up fine.

That last one is the tell. Fatigue that persists regardless of how much you do is rarely about effort. It is about a target that keeps moving.

The keep-up treadmill

Consider what the market is telling workers. Microsoft and LinkedIn's 2024 Work Trend Index found that two-thirds of leaders said they would not hire someone without AI skills. In the same report, only 39 percent of people using AI at work said their company had trained them. The expectation arrived. The support did not.

The Upwork Research Institute found a similar gap that year. Nearly half of employees using AI, 47 percent, said they did not know how to achieve the productivity gains their employers expected, while 96 percent of C-suite leaders said they expected AI to raise productivity.

So the learning gets pushed onto evenings and weekends. Workers watch tutorials at night, test tools on Sunday, read newsletters about newsletters about models. Upskilling becomes unpaid second-shift work. And because the tools change every few weeks, the learning never compounds into rest. You master something just in time for it to be superseded.

There is an irony here worth naming. Learning can be restorative. Sabine Sonnentag and Charlotte Fritz (Journal of Occupational Health Psychology, 2007) identified mastery experiences, challenging yourself with something new outside work, as one of four experiences that help people recover from work, alongside psychological detachment, relaxation and control. But the same research treats control, choosing how you spend your own time, as its own ingredient. Learning you chose in the evening can refill you. Learning you were told to do, in the evening, to stay employable, is just more work wearing a hobby's clothes.

This is how a technology sold as freeing up time ends up consuming it. We have seen the same pattern in productivity dysmorphia: no matter how much you learn, the reference point moves.

Why does constant change wear you down?

Is it strange to feel tired from something that is supposed to be exciting? Not according to the research on technostress.

Monideepa Tarafdar and colleagues (Journal of Management Information Systems, 2007) surveyed technology users across 223 organizations and found that technology-driven stress reduced productivity and increased role stress, partly offsetting the productivity gains the technology was meant to deliver. Their work, and a follow-up led by T. S. Ragu-Nathan in Information Systems Research (2008) with 608 users, gave names to the specific pressures. Two map closely onto AI. Techno-overload is the sense that technology forces you to work faster and longer. Techno-uncertainty is the unsettled feeling that comes from tools that keep changing and keep needing to be relearned. In the 2008 study, these stress creators were linked to lower job satisfaction and, through it, lower commitment.

None of that research was about generative AI. It was about email, enterprise software and the early smartphone era. That is the point. The mechanism is old. The pace is new.

Information overload works the same way. Martin Eppler and Jeanne Mengis (The Information Society, 2004) reviewed decades of research across organization science, accounting, marketing and information systems and traced a consistent pattern: when the information arriving exceeds the capacity to process it, decision quality falls and stress rises. AI news is an overload machine. Every launch claims to change everything, and there is a new one before you have finished reading about the last.

Then there is the social layer. Andrew Przybylski and colleagues (Computers in Human Behavior, 2013) defined fear of missing out as a persistent worry that others are having rewarding experiences you are absent from, paired with a drive to stay continually connected to what they are doing. Swap "experiences" for "productivity gains" and you have the professional version. Your feed is full of people describing their new workflow. You are not seeing their failed experiments, their abandoned subscriptions, or their own private sense of being behind.

AI slop and the fatigue of filtering

Keeping up is one kind of tired. Filtering is another.

Merriam-Webster named "slop" its 2025 word of the year, defining it as low-quality digital content produced in quantity, usually by artificial intelligence. The dictionary pointed to a spike in lookups, and the word now covers a whole genre of the feed: mass-produced synthetic videos, junk books, glitchy advertising images, repetitive posts.

The workplace version arrived with its own name. In 2025, researchers from BetterUp Labs and the Stanford Social Media Lab described "workslop" in Harvard Business Review: AI-generated output that looks finished but does not actually move a task forward. In their survey of US desk workers, 40 percent said they had received workslop in the previous month. Each incident took an estimated two hours to deal with, and about half of recipients said they saw the sender as less creative, capable and reliable afterward.

Slop fatigue is the cost of constant discernment. Every email, document and post now carries a quiet extra question: did a person mean this, or did a machine produce it? That question takes attention. Asked a few hundred times a day, it takes a lot.

Selective ignorance is a skill

What if you are not supposed to keep up with everything?

Ralph Hertwig and Christoph Engel (Perspectives on Psychological Science, 2016) wrote a paper with a pointed title, "Homo Ignorans," about deliberate ignorance: the conscious choice not to seek or use certain information. They argued it is not a failure of rationality. People choose not to know for good reasons, including protecting their attention, their emotions and the quality of their decisions. Deliberate ignorance is a strategy.

Applied to AI, that looks like this:

• Pick depth over coverage. One or two tools that solve problems you actually have, learned properly, beat ten tools skimmed.
• Batch your intake. Review what has changed once a month, not every morning. Most launches that matter will still matter in four weeks. Most that do not will have been forgotten.
• Test against your real work. Ask whether a new tool improves a task you do every week. If you cannot name the task, it can wait.
• Keep a "not now" list. Write interesting things down and let them sit. The list absorbs the pull without obeying it.
• Mute the launch cycle. Unfollow accounts whose main product is urgency. Keep a few sources that explain rather than announce.

We made a similar case in our essay on news avoidance: the answer to overload is not total withdrawal but a deliberate diet, with fixed times, chosen sources and fewer alerts. The same logic applies here. You can stay informed without staying flooded.

Behind compared to what?

The phrase "falling behind on AI" assumes a race with a known finish line and a clear field of competitors. Neither exists. The tools are changing too fast for anyone to be fully current, including the people building them. What you are seeing on your feed is a highlight reel of everyone's best week.

This is the same distortion we see in feeling behind in life: the reference group is imaginary, and the comparison is rigged. And it connects to a broader pattern in digital overwhelm, where the problem is not any single input but the total volume.

Rest, in this context, is not opting out of the future. It is refusing to let the pace of somebody else's product cycle set the pace of your nervous system. You are allowed to learn at the speed of understanding. You are allowed to let a release go by. You are allowed to be the person who is good at a few things rather than anxious about all of them.

That is not falling behind. That is choosing where to stand.

---

Get about one essay a week on the science of rest and the cost of optimization culture. Subscribe to the newsletter. Free, about one essay a week, no upsell.

Related reading: AI Brain Fry · AI Anxiety and the Nervous System · News Avoidance · Productivity Dysmorphia