First the basics. 69% of companies say they are currently using AI, with text and image generation via LLMs being the most common use case. On average, AI tools are used 1.5 hours per week, with three in ten companies reporting they do not use AI tools on a regular basis, and only 7% reporting AI tools used more than five hours per week.
However, AI adoption is increasing rapidly. In the first wave of the survey, from February to April 2025, 61% of UK companies reported using some AI tools. By January 2026, that had increased to 71%.
Where it gets interesting is to look at the past and the expected productivity gains and possible employment losses. Let’s first look at productivity: companies hardly report any productivity gains over the past three years. On average, productivity has increased by about 0.3%, which is largely in line with the rather pessimistic expectations from Daron Acemogluwho thinks that productivity gains will be around 0.1% per year over the next ten years.
However, companies still have high expectations for the future and expect productivity gains to increase by a factor of five to 1.4% over the next three years, driven mainly by US and British companies, which expect productivity gains to increase by a factor of eight to ten.
A separate survey of US workers shows that productivity gains have increased by 0.5% over the past three years and are expected to double to 0.9%.
Productivity gains through AI
Source: Yotzov et al. (2026)
However, when it comes to employment, employers and employees are beginning to fundamentally disagree. US workers generally expect AI to generate new jobs, while employers report no impact on employment in the past three years but expect a significant decline in employment in the future. Once again, American and British companies appear to be the most optimistic about the potential of AI. US companies expect employment in their businesses to fall by 1.2% over the next three years, while UK companies expect employment to fall by 1.4%.
Impact of AI on employment
Source: Yotzov et al. (2026)
I think this research works like a Rorschach test on AI. If you are optimistic about the possibilities of AI, you can look at these charts and say that AI adoption is increasing rapidly, and the power of AI tools is growing so quickly that the productivity gains will be great. I suspect Fed Chairman Kevin Warsh would make that argument to justify even more rate cuts.
AI skeptics, on the other hand, might argue that after three years, when AI could have had the biggest impact on productivity since it starts from scratch and low-hanging fruit abounds, productivity gains are still only 0.3%. Moreover, use cases seem to be limited mainly to writing texts and creating images (some would say that enterprise AI is sloppy), with productivity gains coming to an end quite quickly in the future. So it seems a bit like AI is more of a promise than a reality and that productivity gains are always just out of reach, but never here.
Which side do you fall on?
#High #expectations