6%
of companies see a meaningful impact on their EBIT from AI.
The problem
Eighty per cent of people who use AI at work report that it has improved their productivity, but only six per cent of companies see a meaningful impact (>5%) on their EBIT.1
ConclusionPersonal gains are not translating into business outcomes.
<4%
of workers can write an effective automation.2
A key reason
The use of AI is shallow. The most common use of AI at work is as a replacement for search, and more than seventy per cent of workers use it for one-off tasks. Most people have access to agents at work but only ten per cent understand what an agent is.2
ConclusionPeople are not using AI for the highest-impact work.
58%
of companies now offer AI training, up from a quarter a year ago.3
And yet
Companies have taken action. The share offering AI training has more than doubled in a year, but this has not produced the proficiency or the gains they expected: sixty-two per cent of knowledge workers have had some AI training, while only five and a half per cent are proficient.2
ConclusionThe training on offer today is not working.

