Human anxiety triggered by AI: what every leader should understand

Every AI conference keynote promises the same thing, efficiency, growth, a smarter way of working. Sit in the audience, though, and you'll notice something the slides don't capture. Not everyone in the room is excited. Some are anxious, and most of them aren't saying so out loud.

That gap, between how AI gets talked about and how people actually feel about it, is what Geeks co-founder Somayeh set out to explore on a recent episode of the Innovation Room podcast, with Philip Karahasan, a psychotherapist, psychoanalyst, and founder of Therapy in London. What follows is a look at what that conversation uncovered, and what it means for anyone leading a team through AI change. It's a conversation Geeks hosts specifically because these harder, less comfortable questions rarely get airtime in a typical AI strategy meeting, even though they tend to matter just as much as the technical ones.

Philip's explanation starts somewhere unexpected: dopamine. Dopamine is the hormone that drives action, the thing that gets you off the sofa and into the kitchen when you're hungry. Act on it, and the tension resolves. Don't, or can't, and the same hormone builds into something else entirely, anxiety.

Applied to AI, the mechanism is the same. There's a drive to understand what's coming and do something about it, and for most people right now, there's nowhere for that drive to go. Nobody can tell them exactly how AI will change their role, their industry, or their income, so the energy has nowhere to land. Philip compares it to the nerves before a race or an exam, the same drive that would normally propel you toward the finish line, but with nowhere obvious to run in the meantime. AI creates a very similar holding pattern for a lot of people right now. Everyone can feel that something is coming. Almost nobody has a clear, actionable next step to release that tension.

“It's the unknown right now that is the main cause of that anxiety that I can see.”

— Philip Karahasan, psychotherapist

That distinction matters for anyone trying to address it. AI anxiety isn't really about AI. It's about uncertainty, and uncertainty responds very differently to reassurance than a specific, well-defined fear does.

It's a fair question, and Philip's answer isn't a simple yes or no. Some worry is a rational response to genuine uncertainty. What makes AI anxiety different from other kinds of workplace change anxiety is what it actually threatens.

Humans have adapted to disruptive technology before, cars replacing horses, computers replacing typists. What's different about AI, in Philip's view, is that it's the first technology that replicates the thing people have always used to define themselves: intelligence. That's not just a job security question anymore. It edges into identity, into what makes a person useful or valuable, which is why the fear of AI often runs deeper than the fear that came before it.

Philip points to an old comparison to make the contrast clear: horses pulling carts gave way to cars, and the market adapted rather than collapsed. AI, in his view, doesn't fit that pattern as neatly. A machine beating a chess grandmaster was an early sign of something less comfortable than a market shift, evidence that a machine could outperform a human at a task built entirely on human intelligence.

This isn't a fringe reaction either. Gallup's 2026 State of the Global Workplace report found that 18% of US employees believe their job will be eliminated within five years because of AI, a figure that rises to 23% inside organisations that have already deployed AI tools. Artificial intelligence fears of this kind are now a majority workplace experience, not an overreaction from a nervous few.

Somayeh shared a moment that stuck with her from a client conversation. Mid-discussion about AI strategy, the CEO of a large organisation seemed to drift out of the conversation entirely. Asked what was wrong, he admitted what he'd actually been thinking about. Are we going to be replaced by AI? Not his team. Him.

It's easy to assume anxiety about AI sits with junior staff worried about their jobs. Philip and Somayeh's experience suggests otherwise. Leaders carry a version of the same identity question, just with an added layer, they're also expected to have the answers for everyone else. A leader working through their own AI anxiety while trying to project confidence to a whole organisation is effectively carrying two jobs at once, and rarely admitting to either. That kind of moment rarely makes it into a boardroom slide deck, but Philip sees it often in his own practice, senior people expected to project total confidence about AI while privately working through the same identity questions as everyone below them.

Anxiety doesn't usually announce itself directly. It tends to show up sideways, in behaviour that looks like something else entirely.

• Withdrawal. People who go quiet in conversations about AI, rather than engaging with it, are often managing anxiety rather than genuine indifference.
• Parroting. Repeating headlines and secondhand stories about AI, rather than voicing an actual opinion, is often a way of avoiding the harder, more personal question underneath.
• Rivalry. Treating AI, or a colleague using it well, as competition rather than a tool mirrors the same territorial response Philip has seen when a highly capable new hire joins a team.
• Dismissiveness. Unprompted complaints about AI's flaws can be less about the technology itself and more about needing to feel it still isn't a serious threat.

None of these signs are dramatic on their own. That's exactly why they're easy to miss in a busy team, and why naming them explicitly is often the first useful step a leader can take.

Philip's advice for leaders comes down to conversation, but not just any conversation.

The instinct to reassure a worried team, telling them it'll be fine, or that AI is only there to help, doesn't actually reduce anxiety. It tends to shut the conversation down instead. What works better, in his experience, is asking questions rather than supplying answers. What do you think the next step should be? What's actually worrying you about this? Questions like these surface what people are truly thinking, rather than what they assume they should say, and leave them feeling heard rather than managed.

Bringing people together helps too, not to broadcast a message, but to let a group discover that their anxiety is shared. Anxiety left to sit alone tends to compound. Anxiety spoken aloud, and recognised in other people, tends to ease.

The other half of it is the story leaders tell about AI itself. Positioning it honestly, as a tool that supports the work rather than a replacement for the people doing it, matters more than most leaders assume, even in cases where parts of a job are genuinely changing.

None of this means treating AI as purely a threat to be managed. Philip is careful to note there's a genuine upside too, including for people who find connection easier through AI than they do face to face. The point isn't to erase the anxiety or pretend it doesn't exist. It's to make sure it isn't the only conversation happening.

This article covers the core of the conversation, but the full episode goes further into the psychology of it, including how people are already forming emotional attachments to AI chatbots, why a nine-year-old politely said goodbye to a chatbot before it was switched off, and what that might mean for how the next generation relates to both people and machines.