TL;DR
Half of the UK chief executives surveyed for EY-Parthenon’s latest CEO Outlook name AI-enabled tools among the biggest drivers of their productivity gains in the last 12 months, Consultancy.uk reported on 19 October. Yet only 14% say they have comprehensive visibility of what AI costs them and what it returns. The other 86% lack that full picture.
The UK numbers
EY polled 1,200 chief executives of large companies in 21 countries during September 2026, including 100 in the UK. Among the UK group, AI tools (50%) came ahead of redesigning and simplifying business processes (40%) and reskilling staff (38%) as a source of productivity improvement.
The visibility figure is the striking one. With so few able to see AI’s full costs and returns, Consultancy.uk says it remains possible that AI is not delivering the gains bosses credit it with.
EY’s wider argument is that the productivity debate is shifting from whether gains are possible to whether they can be measured with confidence. Benefits can show up inside individual firms before they are large enough to register in national statistics, and without good measurement, early gains can be cancelled out elsewhere by the cost of implementation, work done twice or fresh demands on staff.
What EY says leaders should do
Silvia Rindone, who leads EY-Parthenon in the UK and Ireland, said “the AI conversation has shifted from adoption to impact”, with UK chief executives now wanting clear evidence their spending delivers. Productivity alone, Rindone argued, will not decide the returns; firms also need fitting processes, skills and ways of operating, and leaders must line up “capital, technology and talent” to turn AI gains into lasting growth.
Looking forward
A sample of 100 UK bosses is small, and the survey measures what chief executives believe rather than audited results. In our view, the practical step for UK firms is unglamorous: tag AI spending in the accounts, set a baseline before each deployment and review it after six months. Without that, boards will keep approving AI budgets on conviction rather than evidence.