Two facts about artificial intelligence and work are both true right now, and they point in opposite directions. The first is that the hard numbers remain stubbornly calm. A recent analysis by the Budget Lab at Yale found no connection between the rapid adoption of AI and changes in employment rates, and a paper from the US Bureau of Economic Analysis reached a similar conclusion, associating AI use with stronger economic growth rather than job losses. Stranger still, workers in the occupations most exposed to AI are seeing the fastest pay rises: Indeed data cited this week shows advertised salaries in the most-exposed US roles up 46 percent since 2021, against 25 percent for the least-exposed. If a machine is coming for these jobs, the market has not priced it in.
The second fact is that almost nobody on the ground believes the first one. "Everybody's probably going to get fired, just wait," one American told Fox News Digital when shown the reassuring figures. That "just wait" captures the mood precisely. Three-quarters of surveyed US manufacturing and utilities workers see AI as a threat to their jobs. In a King's College London study, more than half of Britons expect AI to cause widespread unemployment, and one in five thinks it will do so at a rate that triggers civil unrest.
Some of that anxiety is now showing up in specific corners of the data, if not the headline totals. In Britain, the Office for National Statistics counted 101,000 fewer payrolled employees in the year to July, with a provisional August estimate of a 145,000 annual fall. Government-backed research using LinkedIn data found UK hiring down 14 percent year-on-year, with the steepest drops in exactly the roles where AI has become visible: entry-level postings for accountants fell 29 percent, graphic designers 28 percent, software engineers 27 percent. In Singapore, retrenchments hit 4,620 in the second quarter, the highest since late 2020. In China, a Beijing programmer named Fei Zhaojun was among 160 workers laid off after his employer began testing whether AI could do the coding, as the share of Chinese industrial firms using AI leapt from under 10 percent to roughly half in a single year.
How can both pictures hold at once? Partly because aggregate employment is a slow, noisy measure that can stay flat while pain concentrates in particular occupations, age groups, and regions. Older workers and new graduates are absorbing more of the strain than the averages suggest. And partly because the fear itself has effects that no jobs report captures. Carl-Benedikt Frey, the Oxford economist who studied the anti-Uber protests of the last decade, argues the sequence is the reverse of what most people expect: "civil unrest will come long before mass unemployment." A general-purpose technology that shaves even 5 or 10 percent off incomes across many industries, he notes, can generate plenty of discontent without a single mass layoff event.
The political class is starting to treat that possibility as a planning problem rather than a talking point. Darren Jones, until July a senior UK cabinet minister, warned this week that AI could open "cracks in the very foundations" of the British economy and provoke a "populist backlash" if displacement outruns job creation, straining tax receipts and welfare at the same time. Meanwhile the workplace itself is quietly changing shape: California's legislature has passed a bill to curb "robo bosses" from firing workers, Meta faces a lawsuit alleging AI-assisted systems were used to rank employees for layoffs, and one Brookings scholar has floated the idea that companies will need a "robot relations" department to handle grievances against algorithmic managers.
There is a more hopeful reading, and it deserves airtime. Among the AI leaders studied by Boston Consulting Group, 59 percent said they use the technology mainly to expand what each employee can deliver, while only 10 percent use it to "do the same with less," and US manufacturing still carries roughly half a million unfilled openings. AI as a capacity multiplier, not a headcount cull, is a genuine pattern in the firms that use it best. The honest conclusion is that the reassuring data and the rising dread are not really in conflict. The averages describe a transition that has not yet arrived at scale; the anxiety describes a bet about where it is heading. For now, the safest thing to say is that the story is early, uneven, and moving faster than the statistics that measure it.