The Financial Times, in partnership with research firm Focaldata, surveyed thousands of workers in the US and UK to understand AI adoption patterns. The headline finding reveals that the highest-paid and most educated workers are using AI significantly more than lower-paid employees, creating a risk of widening existing productivity and earnings gaps across the workforce.

This pattern contrasts with experimental studies showing AI actually reduces inequality within individual firms and occupations. While call centre workers and management consultants see productivity gains distributed more equally, the broader economy shows uneven adoption. Workers whose jobs involve writing code report the most positive feelings about AI, while professional writers are more negative about the technology than any other professional group.

Only 14 per cent of workers have received formal AI training from employers, despite roughly a quarter of UK workers and a third of US workers using AI tools daily. The survey also found 38 per cent believe AI will destroy more jobs than create in their own industry, while 51 per cent hold that view for the broader economy. University of Pennsylvania professor Ioana Marinescu noted that well-paid AI users today may become "losers in the long run" given their vulnerability to eventual replacement.

Recent examples illustrate the risks. Elite law firm Sullivan & Cromwell recently told a US bankruptcy court that a major filing contained multiple AI "hallucinations," underscoring that even sophisticated users are not immune to the technology's flaws.