Abstract

This research provides representative international data regarding firm-level artificial intelligence (AI) adoption, based on surveys of nearly 6,000 chief financial officers, chief executive officers, and other senior executives across the US, UK, Germany, and Australia. Four primary facts are identified. First, around 70 per cent of firms actively use AI, with adoption more prevalent in younger, more productive organisations. Second, while over two-thirds of top executives regularly use AI, their average usage is just 1.5 hours per week, and one-quarter report no personal AI use. Third, firms report minimal impact from AI over the past three years, with over 80 per cent stating there has been no effect on either employment or productivity. Fourth, businesses forecast sizable impacts over the next three years, predicting that AI will boost productivity by 1.4 per cent, increase output by 0.8 per cent, and reduce employment by 0.7 per cent. Notably, surveys of employees suggest a 0.5 per cent increase in employment, highlighting a significant gap in expectations.

Introduction

There has been a 20-fold surge in media focus on AI impacts since 2020. Despite a large quantity of data from sources such as job postings and Census data, existing research often faces challenges related to sample size, representativeness, and the nature of responses. High-quality, representative international data on firm-level AI use as reported by senior executives has remained scarce. To address this gap, four research teams from the Federal Reserve Bank of Atlanta, Bank of England, Deutsche Bundesbank, and Macquarie University fielded parallel surveys between November 2025 and January 2026. The research aimed to collect consistent data across four advanced economies to understand the broader impact of AI and guide future policy and research.

Data

The research utilises data from four main sources: the Survey of Business Uncertainty (SBU) in the US, the Decision Maker Panel (DMP) in the UK, the Bundesbank Online Panel of Firms (BOP-F) in Germany, and the Business Outlook Scenarios Survey (BOSS) in Australia. These surveys targeted senior financial decision-makers, with the majority of respondents being CEOs, CFOs, or in senior management positions. In the UK and Germany, for example, over 90 per cent of respondents held these senior positions. The sponsorship of central banks and academic institutions facilitated the recruitment of high-level management for participation.

Data Validation

The quality of the firm survey panels was validated through two primary exercises. First, a comparison of ten years of survey data on output and employment against national aggregate statistics demonstrated that the panels accurately track US and UK GDP and private sector employment growth. Second, an evaluation of sales and employment forecasting accuracy showed that executive predictions line up tightly with actual realisations occurring over the following year. This track record suggests that the survey panels are well-equipped to predict future business conditions and the potential impacts of AI.

Main Results

Current adoption of AI technologies is widespread, with 69 per cent of businesses across the four countries currently using some AI technology. Adoption is highest in the US (78 per cent), followed by the UK (71 per cent), Germany (65 per cent), and Australia (59 per cent). The most commonly cited uses include text generation using large language models (41 per cent), visual content creation (30 per cent), and data processing using machine learning (30 per cent). Adoption is not concentrated in a single use case, though LLMs are the most popular. Heterogeneity is evident, as larger, more productive, and higher-paying firms are more likely to adopt AI, while older firms and those with older directors show lower adoption rates. In the UK, adoption is highest in the finance, insurance, professional, and scientific sectors.

Regarding employment, over 90 per cent of managers report no impact from AI over the past three years. However, looking ahead to the next three years, executives anticipate that AI will reduce employment by 0.7 per cent, which could lead to approximately 1.75 million fewer jobs at existing firms by 2028. Approximately two-thirds of this reduction is expected to come from hiring fewer new employees rather than increased exits. In contrast, AI is predicted to boost productivity by an average of 1.4 per cent over the next three years. The largest positive impacts are expected in the US (2.3 per cent) and the UK (1.9 per cent), specifically within the information, communications, administrative, and support sectors.

Estimated AI Impacts By Employees

Surveys of individual employees via the Survey of Working Arrangements and Attitudes (SWAA) reveal a contrast in expectations. Employees are more optimistic than executives regarding future employment, predicting a 0.5 per cent increase as a result of AI. However, they are less optimistic about productivity gains, forecasting a 0.9 per cent increase compared to the 1.4 per cent predicted by firms. Employee reports on personal AI usage are similar to those of executives, averaging approximately 1.8 hours per week.

Conclusion

The findings indicate that while AI adoption is widespread and growing, its impact on employment and productivity has been modest thus far. However, senior executives anticipate significant transformations in the coming years, including sizable productivity gains and a net reduction in jobs. The contrast between executive and employee expectations highlights a critical gap in perceptions. Overall, the study provides a methodology for monitoring AI adoption and its macroeconomic impacts across multiple countries through consistent survey design.