Don’t Lose the Skin in the Game
If I hear the word 'Boomerang' one more time in reference to Ai firing and hiring, I think I might actually explode.
Where am I hearing it? It’s being used to describe the employee who gets laid off in the name of Ai, and then sometimes within weeks, often within six months, gets quietly hired back to do the job that, it turns out, the machine couldn’t. A Careerminds study of 600 HR leaders found that around two-thirds of companies who cut roles because of Ai have already started rehiring the people they let go. More than a third have brought back over half of them. And here’s the detail that should stop any executive cold: nearly a third found that the cost of rehiring exceeded what they’d saved by cutting in the first place.
So as a piece of arithmetic, much of this has been a failure. The savings were a mirage. The work didn’t vanish; it shifted, got messier, and demanded exactly the kind of human judgement nobody had bothered to account for.
But I don’t want to talk about the arithmetic, because the arithmetic, however badly it’s been done, is the least of it.
The thing the spreadsheet can’t see
When a company makes an Ai-driven layoff, it performs a particular kind of conceptual sleight of hand. It takes a person, with a career, a craft, a history of decisions made well, a sense of being good at something, and converts them into a line item. A cost. A variable in an equation that, as we now know, most of the people running it didn’t fully understand.
And then, months later, when the equation breaks, it converts them back.
We talk about this as though it were a logistics problem. A planning error. An over-correction to be quietly walked back. But sit with what it actually is for the person on the other end. You were told, in the clearest possible institutional language, that your contribution could be automated away. That what you did, the thing you’d spent years getting good at, was a rounding error a piece of software could absorb, and then you were told, just as clearly if rather more sheepishly, that actually, no, they needed you after all.
You can rehire a role. You cannot so easily rehire someone’s belief that their work mattered. That damage doesn’t show up in the cost model, because the cost model was never built to see it. It lands on self-esteem, on identity, on the quiet internal story a person tells themselves about whether they’re any good. It lands on mental health. It follows people home.
We have built a habit of making decisions about human beings using instruments that are constitutionally incapable of registering the human being, and then we act surprised when the consequences are human.
This is a design problem
I’ve spent my career arguing that design is, at its heart, the discipline of seeing the person behind the data point. Not the persona. Not the segment. Not the user-as-conversion-event. The actual person, with their context and their constraints and their dignity intact.
That’s not a soft skill. It’s the whole job. Good design refuses the abstraction that lets you optimise a system while quietly degrading the lives inside it. It insists on asking who is this for, and what does this do to them before asking how fast or how cheap.
Which is why the current moment is, for designers, both an enormous opportunity and a profound responsibility. We are watching organisations make decisions of exactly the kind design exists to interrogate: high-stakes, human-affecting decisions, made fast, made in fear, made on the basis of a capability nobody yet understands. The temptation is to treat Ai adoption as a pure efficiency exercise, a problem of cost, throughput and headcount. And efficiency, unexamined, has a way of becoming a euphemism for we stopped seeing the people.
The designerly instinct (sit with the ambiguity, ask the question before reaching for the solution, understand the problem properly before deciding how to solve it) is precisely the corrective this needs, not because designers are nobler than anyone else, but because we are trained to notice the cost that doesn’t appear on the invoice.
To the people making the cuts
If you’re a leader weighing one of these decisions, I’d ask you to hold two things at once.
The first is simple self-interest, and the data backs it: the cuts are frequently not saving what you think they’re saving. The institutional knowledge that walks out the door, the morale that quietly collapses, the customer trust that erodes, the contractors you hire to plug the gaps, the rehiring you’ll likely be doing in six months at a premium. None of that is in the headline saving. The boomerang isn’t a feel-good HR trend. It’s the sound of a miscalculation coming back.
The second is harder, and matters more. Every name on that list is a person whose sense of their own worth you are about to put your thumb on. You may need to make hard decisions; sometimes leaders genuinely do, but there is a vast difference between a hard decision made with clear eyes and full respect for the people it affects, and a panicked one made by treating those people as figures to be optimised away.
The first can be survived, by everyone. The second leaves marks.
Get under the skin, but don’t lose it
I keep coming back to a question a thoughtful designer asked recently: as this new capability arrives, what are you doing to get under the skin of it?
It’s the right question. We should be relentlessly curious about what these tools can and can’t do, where they fit, where they fall short, where the genuine partnership lies. I’m all for getting under the skin of the capability.
I just don’t think we can afford to lose the skin in the process. The people are not the inefficiency. They were never the inefficiency. They are the thing the whole enterprise exists to serve, and the thing it cannot, in the end, run without.
Don’t do things better, do better things, starting with how we treat the people we were so quick to reduce to numbers.
I explore the human cost of Generative Ai further in my new book, Drawn to Extinction, which digs into the mental health implications of Ai when it’s deployed as a tool for replacing people rather than working alongside them.