Stress, safety, and the worst possible time to get comfortable
The qualities an AI future demands are the ones we've spent five years removing
In 1998, thirty thousand Americans were asked two questions: how much stress have you experienced this year? And do you believe stress is harmful to your health?
Eight years later, the researchers checked who had died.
High stress increased the risk of dying by 43 percent. But only for people who believed stress was harming them. People who reported high stress but didn’t see it as harmful had the lowest death risk in the entire study - lower even than people who reported very little stress at all.
The researchers estimated that over those eight years, 182,000 Americans may have died prematurely. Not from stress. From believing stress was bad for them.
That story comes from Kelly McGonigal’s book The Upside of Stress. It came back to me after reading something Christine Armstrong wrote the other day.
Christine’s old boss used to tell her: “never be fragile.” Great advice, she said, but rather out of fashion nowadays, quoting people avoiding personal questions at work for fear of causing offence. And colleagues who don’t know each other well enough to trust each other, so they say as little as possible.
She’s right, of course.
We’ve spent the last five years optimising work for individual comfort - remote, async, fewer awkward conversations, less proximity - and somewhere along the way we’ve confused ‘safety’ with avoidance.
The intention was good. But the result is that we’ve quietly built organisations full of people who aren’t used to being uncomfortable anymore.
The worst possible timing
AI demands exactly the qualities we’ve been letting atrophy.
On an individual level, this could be the acknowledgement that some of the skills you’ve spent years developing may not be as useful any more - and the willingness to pick up a tool you don’t understand and fail with it for an hour to learn something new.
At a team level, it’s most likely something like having the confidence to say “this process we’ve been running for three years is a waste of time” and deal with the silence that follows. Or the patience and courage to try something, show it to someone, and hear that it doesn’t work.
All of it requires a tolerance for friction that we’ve spent five years systematically removing.
Nassim Taleb has a word for systems that get stronger under stress: antifragile.
Fragile breaks. Robust survives. Antifragile improves. Most companies aren’t any of the three. They’re just...comfortable.
The team that never disagrees, the leader who never hears bad news, the organisation where everything runs smoothly right up until the moment it doesn’t. That’s comfortable.
An antifragile team has difficult conversations and comes out more aligned, not less.
Start something uncomfortable
McGonigal’s point is simple: stress isn’t the enemy. The belief that stress is the enemy is the enemy.
If you’ve been putting off getting to grips with AI, the discomfort of not knowing what you’re doing isn’t a sign you’re not ready. It’s what getting ready feels like. Start before you have a plan.
If there’s a conversation you’ve been avoiding - about your career, about how the team works, about something that’s broken - the willingness to raise it is the thing. You don’t need the answer before you walk in.
And if you can see problems in how your company operates but you’ve been keeping quiet because it’s easier - the ability to say what needs to change, and bring people with you while you do it, is one of the few skills that becomes more valuable the more AI can do.
Machines are very good at optimising what exists. They are useless at telling you it shouldn’t exist.
Thanks for reading.
Ollie
Ollie on Work is a weekly newsletter about what I’m learning from building with AI, advising leadership teams, and trying to bridge the gap between what technology can do and how businesses actually work. If someone forwarded this to you, you can subscribe here.