• NBER study of executives finds widespread AI adoption.
• Measurable productivity gains emerging, stable employment during early deployments.
• 1.4% productivity gains by 2028 expected, workforce change driven by slower hiring.

If there is one question enterprise technology leaders are asking, it is where AI return on investment is appearing in the data. Boards and investors want evidence. A large-scale representative study now provides a baseline. For the moment, the aggregate effects are modest, consistent with an early phase of new tech deployments.

A new working paper from the National Bureau of Economic Research, drawing on surveys of almost 6,000 CFOs, CEOs and senior executives in the US, UK, Germany and Australia, found that while many firms report no large measurable impact from AI on employment, stability during adoption suggests firms are integrating AI into existing workflows without disruption.

The study, titled Firm Data on AI and produced by research teams from the Federal Reserve Bank of Atlanta, the Bank of England, the Deutsche Bundesbank, and Macquarie University, did not use paid online panels and was not vendor-sponsored.

Executives were recruited by telephone, verified in their roles, and not paid for their responses. A McKinsey survey taken around the same time estimated 88% of organisations used AI in at least one function.

The variations reflect differences in sampling and definitions, and the NBER approach focused on the most senior decision-makers and linked responses to national macroeconomic data, making its forward-looking findings most relevant for executives planning their long-term strategies.

AI ROI is emerging as deployment deepens

Across the four countries, 69% of firms are already using some form of AI, led by text generation by large language models at 41%, data processing via machine learning at 28%, and visual content creation at 29%.

Adoption has accelerated. In the UK Decision Maker Panel, the share of businesses using at least one AI technology rose from 61% in early 2025 to 71% by year end.

Measured productivity gains over the past three years average 0.29% across the four countries but the detailed figures show much higher impact on outputs in the UK and US. Nicholas Bloom, one of the study’s authors, observed that previous waves of automation took years to register fully in internationally-averaged data.

What executives expect and how it informs planning

Over the next three years, executives expect AI to raise productivity by 1.4% and output by 0.8%. Employment is projected to decline by 0.7%, which translates to around 1.75 million fewer roles at existing firms across all four countries by 2028. Much of that adjustment is expected to occur through slower hiring rather than sudden job losses.

Expectations and opinions vary from country to country. US executives project a 2.25% productivity increase and a 1.19% employment reduction, while German and Australian executives report smaller shifts. By sector, information and communications, and administrative functions show the strongest expected productivity gains. Wholesale, retail, accommodation, and F&B show the largest expected employment adjustments. For firms in these sectors, workforce planning and re-skilling will be central to capturing the value of any AI deployment.

The expectation gap inside organisations

The study also compares executive responses with those of employees. Using the Survey of Working Arrangements and Attitudes, researchers asked US workers the same forward-looking questions. Employees expect AI to increase employment at their firms by 0.5% over the next three years, and they also anticipate a 0.92% productivity gain, below the executive forecast.

This divergence reflects different experiences. Employees usually encounter AI as a tool that improves task performance. Controlled studies by Brynjolfsson, Li and Raymond report productivity gains of around 14% in customer support roles using generative AI, for example. Other research shows gains in legal drafting and software development. Yet higher up the organisation, executives consider a broader range of issues affecting the organisation as a whole, such as cost structures and competitive pressure. Alignment between the two perspectives will influence how effectively AI tools are deployed, reiterating the need for training of employees.

What the AI ROI gap requires

The study does not dismiss task-level improvements, noting that larger and more productive firms are more likely to use AI and therefore to expect returns. Smaller firms may need to plan more carefully to realise similar gains. Concentration of adoption among well-resourced firms may then effectively increase adoption and help permeate best practices with the technology to the SMB/SME sector.

Planning for cumulative impact

The findings suggest that AI impact is developing gradually, and while aggregate data does not yet show large shifts, adoption is broad and expectations are positive. Realising the available gains depends on integration, training, and to some extent, working processes being redesigned. Firms that align workforce expectations with boardroom strategic objectives are likely to capture value more quickly and with greater effect.

See also: Unit 42: Identity gaps and AI speed increase enterprise risks

Want to experience the full spectrum of enterprise technology innovation? Join TechEx in Amsterdam, California, and London. Covering AI, Big Data, Cyber Security, IoT, Digital Transformation, Intelligent Automation, Edge Computing, and Data Centres, TechEx brings together global leaders to share real-world use cases and in-depth insights. Click here for more information.

TechHQ is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.