AI's productivity promise remains elusive for 90% executives, study finds

AI adoption is spreading rapidly across businesses, but most executives say the technology has yet to deliver meaningful productivity gains

  Artificial intelligence is spreading rapidly across businesses, but the productivity gains promised by the technology are proving harder to find. About 90 per cent of executives surveyed across four major economies said AI had not improved productivity at their firms over the past three years, raising questions about whether widespread adoption is translating into meaningful business gains.

 

The finding comes from the Atlanta Federal Reserve study based on surveys of senior executives in the US, the UK, Germany and Australia. Researchers found that 89 per cent of firms reported no impact from AI on labour productivity, while the average productivity gain reported by businesses was only about 0.29 per cent over the past three years.

 

The results come as companies continue to invest heavily in AI and executives remain considerably more optimistic about what the technology could deliver in the future.

 

AI adoption is rising, but productivity has barely moved

 

For the study, researchers surveyed nearly 6,000 senior executives using comparable questions across the four countries between November 2025 and January 2026. The researchers found that 69 per cent of businesses were already using at least one AI technology.

 

Text generation using large language models was the most common application, followed by visual content creation and machine-learning-based data processing.

 

AI adoption was highest among US firms at 78 per cent, followed by 71 per cent in the UK, 65 per cent in Germany and 59 per cent in Australia. Larger, more productive and higher-paying firms were also more likely to use AI.

 

However, adoption alone has not produced a dramatic change in productivity. The study found that 89 per cent of businesses reported no impact on labour productivity over the previous three years. When the reported effects were converted into an average estimate, AI was found to have lifted productivity by only around 0.29 per cent.

 

The findings also suggest that AI is not yet being used intensively by many senior decision-makers. More than two-thirds of executives said they personally use AI during a typical working week, but average usage was only around 1.5 hours a week.

 

Executives expect a bigger AI boost ahead

 

Executives are considerably more optimistic when asked about the next three years.

 

Businesses surveyed expect AI to increase productivity by an average of 1.4 per cent over the period. They also expect output to rise by 0.8 per cent, while employment at their own firms is forecast to fall by 0.7 per cent.

 

The expected productivity gains vary significantly by country and industry. US executives anticipate the biggest boost at around 2.3 per cent, followed by the UK at 1.9 per cent, while Australia and Germany are both at about 0.9 per cent.

 

Information and communications firms and businesses in administrative and support services expect some of the largest gains. By contrast, sectors such as accommodation and food services, construction and recreational services expect smaller improvements.

 

The employment outlook is more concerning. Across the four countries, executives expect AI to reduce employment at their firms by around 0.7 per cent over the next three years, which the researchers estimate could amount to about 1.75 million fewer jobs by 2028.

 

However, the researchers caution that this does not represent the overall employment effect of AI because it excludes jobs created by new businesses and wider economic effects from higher incomes, lower prices and new products.

 

The employee perspective

 

In a separate study, Tracking Artificial Intelligence Sentiment in the US Labor Market, the researchers analysed millions of employee reviews, corporate financial reports, and AI investment and layoff announcements involving US public companies.

 

The study found that employees tend to view AI more negatively than their employers do, with job security emerging as the biggest concern. Workers also pointed to inadequate training, limited opportunities to develop new skills and weak corporate AI leadership.

 

The research found that AI-related layoff announcements were followed by a sharp deterioration in employee sentiment towards the technology, while management discussions about AI remained consistently optimistic. However, that optimism had no significant relationship with productivity outcomes.

 

“In fact, these job cuts damage employee sentiment toward AI – which is one of the strongest predictors of firm productivity when AI is used,” said Mark Ma, professor of business administration at the University of Pittsburgh and one of the study's authors.

 

Turning AI adoption into gains

 

For businesses, the findings suggest that turning AI adoption into productivity gains may require more than deploying new tools or reducing headcount. Training employees, communicating how AI will be used and building confidence around the technology could be important as companies seek to realise the benefits they expect from AI.