Three reports landed in the same news cycle, from three places that do not usually rhyme. The International Monetary Fund warned in its latest Article IV review that more than 40 percent of jobs in Ireland are exposed to AI, the highest share in the EU. A new Robert Walters survey found that only 15 percent of workers in Singapore feel their job is safe over the next two years. And RTE in Dublin published a sober analyst piece titled simply Big fears in Big Tech over AI job losses. None of the three documents alone would force a rethink. Together they describe a labour market that has stopped treating AI as a future story.
The IMF figure for Ireland is the headline number, but the texture matters. Ireland's exposure is high precisely because the country has spent two decades building a service economy heavy in professional, financial and tech work, the categories where large language models bite first. The Irish Times notes the IMF was careful: exposure does not mean replacement. A job exposed to AI can mean an accountant who becomes more productive rather than redundant. Still, the language was unusually direct for a Fund report. Ireland, it argues, is more exposed to AI economic risks than most peer countries and should plan accordingly.
The Singapore survey reaches the same conclusion from the worker's side. Only fifteen in every hundred Singaporean employees told Robert Walters they felt secure in their role through 2027. The Business Times read this as evidence that government messaging about AI as a productivity dividend has not landed; the population it is supposed to reassure is unconvinced. A separate piece in The Independent Singapore captured the academic counter-current. An NTU researcher publicly questioned whether the country's headlong rush into AI adoption is widening rather than closing inequality, given that the workers most exposed are also the ones with least bargaining power to negotiate the transition.
RTE's analysis adds a third texture, this one from inside the industry. Walking through the recent wave of layoffs at Meta, Google and Microsoft, the piece argues that even Big Tech, which has spent two years promising AI would lift everyone, is now openly using it to justify headcount reductions. An ex-Meta employee profiled in IBTimes UK described turning down a rehire offer because, as he put it, "axe everybody, rehire where there's pain" had become a stated management style. Whatever the actual share of cuts attributable to AI rather than to ordinary cost discipline, the rhetorical permission to blame the technology is now fully established.
What is striking about these reports as a set is that they do not split into the usual optimist-versus-pessimist columns. A study highlighted in The Brighter Side argues AI will radically transform jobs rather than wholesale eliminate them, and a piece in Nagpur Today catalogues which skills are becoming indispensable as automation spreads. Both fit comfortably alongside the IMF warning. The transformation story and the exposure story are the same story told at different time horizons. In year one, productivity climbs and a few roles vanish. In year three, the shape of an industry has changed and the workers who did not retrain are looking at a much smaller hiring funnel.
The hard policy question, which none of these documents fully answers, is what to do about distribution. If forty percent of Irish jobs change in five years, and only fifteen percent of Singaporean workers expect to be untouched, the gains and the losses fall on different people. Productivity dividends accrue to firms and to capital. Adjustment costs, retraining, lost income, fall on workers and on national exchequers. Tax systems built around steady employment income do not redistribute compute returns gracefully. Pope Leo's encyclical, released the same week, makes essentially this point in moral language. The IMF makes it in fiscal terms. The Robert Walters survey makes it as a measure of public mood.
For now, the official line in most capitals remains that AI will create more jobs than it destroys. That may turn out to be true. What changed this week is that the data and the rhetoric are no longer giving anyone permission to wait and see.