We haven’t had many significant figures speaking so frankly AI’s job displacement. Yet last week, a heavy memo landed from Amazon CEO Andy Jassy, “We will need fewer people doing some of the jobs that are being done today.” After years of hype around generative AI, Amazon was no longer experimenting. Over a thousand AI projects were transitioning from pilot to production. This wasn’t a pilot programme. It was a pivot.
This memo shared publicly on 17 June 2025 revealed a deeper truth: job displacement isn’t coming. It’s already quietly happening. And not just in automation-friendly roles like warehousing or delivery. White-collar support functions, from marketing analysts to procurement assistants, are on the line. Not because they’re bad at their jobs, but because models can now do passable versions faster, cheaper, and at scale.
But if Amazon is cautious and corporate, another story from the other coast shows what it looks like when that caution is removed altogether.
In a New York office, a start-up called Mechanize is building software not to assist workers, but to replace them. Its founder, Tamay Besiroglu, doesn’t hedge. “We’re trying to automate the entire economy,” he told The New York Times in a feature from 11 June 2025. The company is backed by major venture capital and explicitly treats the $60 trillion global wage bill as its market opportunity.
There’s something almost theatrical about the clarity: no upskilling promises, no gentle transitions. Just a thesis that if AI can do it, it should, and that whoever builds it first, wins. Mechanize is designed to be the company that takes your job, and sells the agent back to your employer.
It’s easy to feel alarmed. But zoom out, and a more complex picture begins to emerge. Its one that reaches beyond job loss and into who gets to participate in the AI economy at all.
In June, the Alan Turing Institute and LEGO Foundation published a landmark report on generative AI use among children in the UK. They found that one in four children aged 8–12 are already using AI tools. But that number hides a deeper divide: 52% of children in private schools use GenAI, compared to just 18% in state schools.
That’s not just a usage gap. It’s a preview of the future workforce. If AI becomes foundational to how we think, work, and solve problems, then these early divides will calcify into systemic inequities. While one child grows up fluent in prompting and delegation, another may never learn the language of automation.
We’re watching three stories unfold in parallel: a corporate giant scaling back human roles, a start-up scaling up machine-only ambitions, and a school system accidentally scripting tomorrow’s winners and losers. These are not isolated events. They’re threads of the same story, a shift in how labour is valued, allocated, and displaced.
So what does this mean for the rest of us?
It means we’re at a point of inflection, not inevitability. It’s tempting to focus only on whether a job will be replaced. But the deeper question is how we distribute the capabilities, benefits, and burdens of AI. Who trains the new workforce? Who reaps the efficiency gains? Who builds the models, and who gets modelled?
Perhaps the hardest truth is this: AI displacement is not just a technological problem. It’s a policy problem. A cultural problem. An education problem. A justice problem. And no single actor, corporate or public, has the full answer.
But what we do now, in response to signals like these, will decide whether the future of work feels like a collapse, a transition, or a re-imagining.
As Jassy quietly closes one chapter of Amazon’s workforce, and Mechanize opens another with a bang, there’s a classroom somewhere where a child has just discovered what it means to ask an AI a question…and be heard.
The curve is already bending. The question is: in which direction?
