A study by National Bureau of Economic Research surveyed nearly 6,000 corporate executives in the US, UK, Germany, and Australia. More than 80% of them said AI has not noticeably changed jobs or productivity in their companies so far.
The responses came from CEOs, CFOs, and senior managers across businesses of different sizes. Today, about 69% of companies already use some kind of AI. Around 75% expect to use it within the next three years. The most common uses include writing text with large language models, creating images, and processing data with machine learning.
However, the real-world impact remains small. Over 90% of managers said AI did not change employment levels in the last three years. About 89% also saw no improvement in productivity, measured as sales per employee.
Despite this, leaders still expect big changes ahead. Researchers estimate AI could eliminate around 1.75 million jobs across the four countries by 2028. They also expect productivity to rise slightly, about 1.4% over three years.
Interestingly, employees think differently. Workers believe AI will create more jobs than executives predict, though they expect only minor productivity gains.
A government test of Microsoft M365 Copilot in the UK also showed no overall productivity improvement. Some tasks became faster, others slower.
Microsoft executive Jared Spataro admitted measuring AI return on investment is difficult because knowledge work does not always translate into clear financial results. Meanwhile, Microsoft AI leader Mustafa Suleyman predicted many computer-based jobs could be automated within 12–18 months, according to Fortune.
At the same time, Lenovo reported 94% of enterprises expect positive returns from AI investments, even though evidence is still limited.
A new survey by Gartner found companies believe AI will reshape customer service. Routine tasks may be automated, while humans will handle complex or emotional interactions.
Around 80% of firms plan to move some workers into new roles, and 84% will train staff with new skills.
Overall, the data suggests AI’s benefits today are modest, far smaller than the massive spending by tech companies building these systems, even though expectations for the future remain high.
