Despite the global surge in artificial intelligence adoption across industries, a new study by the National Bureau of Economic Research (NBER) reveals that AI’s real-world impact on employment and productivity has been surprisingly limited. According to the research, over 90 per cent of companies surveyed reported no change in employment or labour productivity from using AI over the past three years, challenging widespread expectations about the technology’s transformative power.
The Study: What Companies Are Saying
Researchers surveyed nearly 6,000 senior executives in major economies including the United States, the United Kingdom, Germany and Australia. The findings paint a picture of cautious, low-intensity AI use — with about two-thirds of executives reporting they use AI, but on average only about 1.5 hours per week. Around 25 per cent of firms said they don’t use AI at all in their operations.
Despite significant investments in AI systems, very few companies have seen measurable effects on employment levels or productivity, measured as sales per employee. Over the three-year study period, over 90 per cent reported no employment change, while nearly 89 per cent said AI did not affect productivity.
Current AI Usage Patterns
AI use within these firms tends to be low and focused on basic tasks rather than deep automation. Common applications reported include text generation using large language models, visual content creation, and data processing with machine learning tools. However, these applications have yet to translate into major workforce shifts or output increases for most organisations.
Executives Still Optimistic
Even though immediate impacts have been minimal, corporate leaders remain optimistic about future gains from AI. On average, executives expect a 1.4 per cent increase in productivity and an 0.8 per cent increase in output over the next three years as AI adoption deepens. Some companies also predict a slight reduction in jobs due to automation — although individual employees surveyed believe AI might actually increase employment modestly.
The Productivity Paradox
Economists liken these findings to the so-called productivity paradox from earlier technological revolutions — where transformative technologies like computers initially failed to show up in productivity statistics, only to drive gains later on. This suggests that while AI holds promise, its impact may take longer to materialise at scale.
Summary
A National Bureau of Economic Research study finds that over 90 per cent of firms reported no impact on employment or productivity from AI over the past three years. Most companies use AI sparingly, but executives remain optimistic about future gains in productivity and output. The results reflect a productivity paradox, with significant technology adoption yet limited measurable economic effects so far.
