AI isn't coming for your job. It's coming for your busywork.

In May 2025, Anthropic's chief executive Dario Amodei predicted that AI could wipe out half of all entry-level white collar jobs and push unemployment to 20 percent. A month later, OpenAI's Sam Altman went further, forecasting the end of entire job categories. Companies began citing AI in layoff announcements. Students reconsidered their degrees. A year on, the Guardian ran a piece with a blunt headline: AI was supposed to destroy jobs, so where is the carnage.

It has not shown up, at least not in the way it was described. That is not us being optimistic on behalf of the technology industry, it is what a Stanford Institute for Economic Policy Research brief published in July 2026 found when its authors, including a former Commissioner of the US Bureau of Labor Statistics, went looking for it directly. Unemployment among the most AI-exposed workers has risen only marginally more than among the least exposed. Firms that adopted enterprise AI actually grew employment by 10 percent over the following two years. Job postings for software developers, one of the most exposed occupations there is, have been growing faster than average.

01What's actually changing for the people doing the work

Look closely at the evidence and a more human picture appears. AI is taking on the repetitive, mechanical parts of jobs far more than it is taking the jobs themselves. The clearest illustration sits in a well-known 2025 field experiment at a large customer support call centre, where a generative AI assistant lifted overall productivity by 15 percent. But that gain was concentrated almost entirely among newer staff, who saw a 30 percent jump in issues resolved per hour, as the AI helped them draw on knowledge they hadn't built up yet. The most experienced agents in the same centre, whose value was already in their judgement, saw no real change at all. The tool didn't touch what made them good at their jobs. It gave newer colleagues a faster start.

There's something worth naming here that the research doesn't quite capture but almost everyone who has ever loved their work will recognise. Most of us didn't spend years in school, and years more learning a craft, because we dreamed of formatting invoices, chasing approvals, or reformatting the same report for the fourth stakeholder that week. We did it because somewhere underneath all of that sat the part we actually wanted, solving the hard problem, building the thing, helping the person in front of us. Before AI, getting to that part meant wading through a lot of what we didn't sign up for just to earn the right to do what we did. That was simply the deal. Nobody questioned it because there wasn't an alternative.

There is one now. Strip away the busywork and what's left is closer to the pure form of the job, the part that actually pulled people toward the profession in the first place.

That last detail is worth pausing on. A company rehiring under a new title isn't just correcting a mistake, it's an admission that the work those people did was more valuable than the spreadsheet suggested. Harvard Business School research on how AI is reshaping job requirements found this pattern spreading widely, roles are being redefined around what people bring that AI still can't, which tends to be judgement, relationships, and situations nobody has written a playbook for yet.

02The one thing worth protecting on purpose

If there's a genuine risk in all this, it isn't the one that dominated the headlines last year. It's quieter, and it's about something harder to put a number on, our own creativity. Two separate 2026 studies found that while AI tools measurably improve the quality of writing and creative output for less experienced people, something is lost at the same time. Work produced with AI's help starts to converge, it becomes more similar to everyone else's than purely human work ever was. A related study following scientists found that those who adopted AI tools published more, but the range of topics the field explored actually narrowed, along with how much researchers were talking to and building on each other's ideas.

"AI can make almost anyone better at what they already do. It doesn't automatically make anyone more original, and if we're not careful, it can quietly make us all sound a little more alike."

This is the part of the AI conversation that deserves more of our attention, not less. Losing a job is a visible, dateable event, you notice it happen. Losing the habit of thinking something through yourself happens slowly, and by the time it's obvious, it's expensive to build back. If AI genuinely hands people a kind of superpower, and we believe it does, the responsibility that comes with any superpower is to keep pushing what we do with it, rather than letting the tool quietly become the ceiling instead of the floor.

03What this means for how people build a career now

The clearest throughline in all of this research is that what makes someone valuable at work is shifting away from a job title and toward what they can actually do, especially alongside AI. A large-scale Danish study following AI adoption at the worker and firm level found that AI is already reshaping how people spend their time on the job, well before it shows up in headcount, hours, or pay. The job survives. What someone actually spends their day doing inside it keeps evolving.

For the person living through that shift, this is mostly good news, if they get the chance to grow into it. Harvard Business School's research points to AI literacy, sound judgement, and the confidence to catch an AI's mistake as the capabilities now setting people apart, not raw output, not speed alone. That favours curiosity and a willingness to learn over any particular job title someone happened to start with. It rewards exactly the instincts that made good workers good in the first place, just applied to a new set of tools.

04Where Trusted Hub fits into this shift

This is the shift Trusted Hub is building for directly. Trusted ActionVerse exists to take the busywork off someone's plate, the data entry, the chasing, the reformatting, the parts of a job nobody dreamed of doing, and hand back the time and energy for the part they actually came for. Automate the tedious half of the work, and accelerate the half that made someone want the job in the first place. That's the whole idea behind TAV, not a faster version of the same grind, but a way back to the pure form of the job underneath it.

And critically, it's the person doing the job who builds that automation, not a vendor handing down a system from outside. Every agent a staff member designs themselves is also a small, real step of learning, they see what the tool is genuinely good at, where it still needs a person watching closely, and where their own judgement was the point all along.

We're not trying to take people out of their jobs. We're trying to give everyone inside an SME the chance to spend more of their working life doing the part they actually love, with the right tools in hand, rather than have that chance simply pass them by.