There's a particular sound an office makes when another AI tool gets announced. It isn't excitement. It's the small, collective sigh of people who've sat through three of these already this quarter and know exactly how it'll go. Big launch. Mandatory webinar. A fortnight of activity. Then digital tumbleweeds rolling across the usage dashboard.

When the silence sets in, most leaders reach for the same word. Resistance. People are change-averse, set in their ways, dragging their heels, so the fix is obvious: push harder.

That word is wrong. Your people aren't resisting AI. Most of them are knackered by it. Those are two completely different problems with two opposite fixes, and if you muddle them, everything you do next makes it worse. The mandate. The reminder. The breezy "exciting update". All of it.

This is AI fatigue. It's worth understanding properly, because the data has stopped being subtle.

What AI fatigue actually is

AI fatigue is the overwhelm and quiet opt-out that sets in when you ask people to take on more AI than they've got the headspace or the motivation for. More tools. More change. More pressure to "adopt". Past a certain point, they just stop.

It's tempting to file this under "anti-AI", but that's not what's happening, and the difference is the whole point. We cram three separate things into one word. Scepticism is not believing the tool works. Resistance is not wanting to change. Fatigue is the odd one out. These are people who'd happily change, who probably use ChatGPT at home most evenings, but who are running on empty by the time your rollout reaches them.

There's a sharper, more clinical version doing the rounds too, nicknamed "AI brain fry": the cognitive strain of babysitting too many AI tools at once. More on the grim numbers behind that in a minute. The headline is the hopeful bit. A tired workforce isn't an unwilling one. You can't do much with people who flat-out don't want the thing. You can do an enormous amount with people who want it and are simply too frazzled to act.

Why it looks exactly like resistance

The reason it gets misread is simple. From the boardroom, fatigue and resistance look identical. Both show up as the same flat line on the dashboard. So leaders see the line, assume foot-dragging, and start solving a problem they don't actually have. The research says they're miles off.

Start with the gap between how leaders think their people feel and how they actually feel. In a survey of 1,400 US employees, 76% of executives reckoned their teams were enthusiastic about AI. Only 31% of those teams agreed. [1] The authors' verdict: leaders are "more than two times off the mark". Put plainly, the people at the top can't see how tired the people below them are.

It gets worse the closer you get to the desk. When the research firm Section surveyed 5,000 white-collar workers, around 70% of executives said AI made them "excited". Almost 70% of non-managers said it made them feel "anxious or overwhelmed", and four in ten said it had saved them no time at all. [2] Same technology. Completely different lived experience, depending on whether you're announcing it or absorbing it.

This isn't just a mood. Boston Consulting Group studied 1,488 workers and found something counter-intuitive. Productivity climbed as people picked up their first few AI tools, then fell off a cliff once they were juggling four or more. The heavy-oversight group reported 14% more mental effort, 12% more mental fatigue and 19% more information overload. And here's the figure that should stop you cold. Among the people experiencing this "brain fry", 34% were actively planning to quit, against 25% of those who weren't. [3]

"People were using the tool and getting a lot more done, but also feeling like they were reaching the limits of their brain power — like there were too many decisions to make." — Julie Bedard, Managing Director & Partner, Boston Consulting Group

None of that is a workforce digging in its heels. It's a workforce running on fumes. And in more cases than you'd like, it's quietly updating its CV while it does.

It's change fatigue wearing a new badge

If this feels oddly familiar, it should. AI fatigue isn't really new. It's the latest strain of plain old change fatigue, and change fatigue was at crisis point long before anyone switched on a copilot.

Gartner has the receipts. Employees' willingness to get behind organisational change dropped from 74% in 2016 to just 38% by 2022, while the number of major changes the average worker had to swallow each year went from two to ten. [4] AI didn't cause that exhaustion. It landed on top of it, then poured petrol on the pile.

Then there's the sheer number of tools. The average company now runs more than a hundred separate software apps. A hundred. 28% of enterprises are wrangling more than ten AI apps on their own, and that sprawl is already holding back AI integration at 70% of them. [5] No wonder people spend only about 45% of the working day on actual productive work. The rest leaks away switching between disconnected systems and stitching them together by hand, doing the "grey work" nobody ever put on a job description. [6]

Now picture the launch schedule sitting on top of all that. Every vendor you already pay bolted "AI" onto its product this year. IT pushed out a copilot. Someone announced an "AI-first" strategy at the town hall. A shiny new usage policy hit everyone's inbox. Every one of them billed as a revolution. Not one of them sequenced, prioritised or properly explained. Pile that on a workforce faster than anyone can keep up, and fatigue isn't a risk. It's the only possible outcome.

Treating it as resistance only makes it worse

This is where good intentions start doing real damage. Flat usage feels like something you fix with pressure, so leaders push. Mandate the tool. Monitor who's logging in. Have a quiet word with the stragglers. Every one of those moves treats a tired workforce like a defiant one, and pours on more of the exact thing that caused the fatigue.

Mandates buy compliance, not adoption. Logins go up and nothing changes about how people actually work. They also backfire harder than leaders expect. By early 2026, something like 80% of white-collar workers were pushing back on, or flat-out ignoring, AI adoption mandates. [7] Monitoring breeds fear, and fear is what tips a tired person from "not right now" to "never". More training? That just dumps another load on people who are already maxed out. You're prescribing more of the disease and calling it the cure.

Someone who's exhausted doesn't need a harder shove. They need a reason that's actually worth the effort, a path that's clear instead of cluttered, and permission to ignore the noise. That's not a management problem. It's a marketing one. It always has been.

The 5 drivers of AI fatigue (and the marketing fix for each)

Strip back almost any fatigued AI programme and the same five culprits are underneath. Not one of them is fixed by buying more technology. Every one is something a half-decent marketer would get right on instinct.

1. You launched everything at once

Launch ten things at once with equal billing and people can't tell what matters, so they tune out all of it. No marketer worth their salt would put ten products live in a single breath. Yet that's roughly what most AI rollouts do. The fix is sequencing. One hero tool. One clear use case. One message at a time. Plus the nerve to say "not yet" to everything else.

2. Nobody made the case for it

People were told to adopt. Nobody told them what was in it for them, in their actual job, on an actual Tuesday. Training shows them the buttons. It never sells the benefit. So the one question they really care about, "why should I spend my last drop of energy on this?", goes unanswered, and they answer it themselves. Usually with a shrug.

3. Every new tool lands as extra weight

Past the third tool, one more isn't progress. It's the brain-fry cliff from earlier. "Here, have another AI tool" feels like a gift to the person handing it over and like a burden to the person catching it. Good marketers obsess over the single next action they want from someone. Be just as ruthless. Curate. Stop piling things on.

4. There's fear sitting under the tiredness

Exhaustion plus a whiff of threat equals shutdown. Around a quarter of the global workforce is worried about AI, and roughly a third say they regularly feel overwhelmed by it. [8] If part of someone suspects the tool is here to replace them, no amount of cheerleading from the top will shift them. You have to address concerns in a way that builds trust and buy-in. Out in the open, not in a reassuring sentence buried on page nine of a policy PDF.

5. Nobody can see it working

We copy what we see the people around us doing. With no real champions, the actual colleagues rather than the IT department, and no specific local wins to point at, there's no proof the effort pays off, so a tired team sensibly keeps its energy for something else. The answer is social proof. Stand up an AI Champions network and let them share success stories and embed grassroots enthusiasm. One real result from someone two desks over beats any number of slides promising "productivity gains".

Market the change, don't mandate it

The cure for AI fatigue isn't more AI. It's treating your rollout as what it actually is: an internal marketing campaign, aimed at tired, sceptical, busy humans who need winning over, not telling. It's the same principle behind Marketing for Learning®. Build it and they will come is a lovely idea for baseball films and a terrible one for software. They don't come. You have to bring them.

In practice that means running a people-centred adoption programme instead of firing off announcements. Doing the work to prepare and motivate employees for AI roll-outs before the tool ever lands. Building an AI Champions network so colleagues can share success stories and embed grassroots enthusiasm. Handling worries in a way that builds trust and buy-in. Get it right and you drive organisation-wide AI uptake without grinding anyone further into the ground.

When you don't drop it, it works. At Capgemini, we ran an AI and digital-skills programme like a proper launch. Messaging tailored by audience. A real value proposition. Communication that kept going instead of landing once and vanishing. Engagement hit 82% across a workforce of more than 130,000. [9] Another client went from 5% to 23% platform adoption in 30 days, with not a scrap of new software. The fatigue didn't lift because we added something. It lifted because, for once, somebody marketed what they already had.

Stop mistaking exhaustion for resistance

Your people aren't the problem. The fatigue is real and it's measurable, and it's also fixable. Just not with another mandate, another tool, or another calendar invite to a training session nobody asked for.

If you want the practical version of all this, the guide How to Increase AI Adoption in Your Organisation walks through it step by step. And if you'd rather just talk through where your own rollout went quiet, get in touch. Sorting that out is the bit we love.