Evidence of AI-driven job losses remains limited, says Oxford Economics report

Claims that artificial intelligence is already driving large-scale job losses appear to be overstated, according to a new global research briefing from Oxford Economics [registration], which suggests that the impact of AI on labour markets so far has been uneven and modest. The report argues that while there is anecdotal evidence of job losses in sectors most exposed to automation, firms are not yet replacing workers with AI at a scale that would materially raise unemployment rates. Oxford Economics concludes that near-term fears of widespread AI-driven unemployment are not supported by current data.

Much of the recent concern has focused on rising graduate unemployment in the US and parts of Europe, particularly in professional and technical services, where AI tools are increasingly being used for tasks traditionally carried out by junior staff. However, the authors caution against drawing direct causal links. Graduate unemployment has historically risen more sharply than overall unemployment during economic slowdowns, and recent trends appear consistent with this pattern rather than indicating a structural shift driven by AI.

Cross-country comparisons reinforce this view. Economies with softer labour markets have generally seen larger increases in graduate unemployment, while countries such as Japan and South Korea, where labour conditions have remained tight, have not experienced similar rises. In the Eurozone, a growing supply of graduates has also played a role, with the proportion of young people holding university degrees increasing sharply in recent years.

The report also questions whether AI-related layoffs are as significant as headlines suggest. In the US, around 55,000 job losses were attributed to AI in the first eleven months of 2025, accounting for less than five percent of total reported layoffs. By contrast, job losses linked to broader economic conditions were more than four times higher. The authors suggest that some firms may be framing layoffs as AI-driven to present a more positive narrative to investors, rather than citing weaker demand or earlier over-hiring.

Oxford Economics points to productivity data as another reason for caution. If AI were already replacing labour at scale, productivity growth would be expected to accelerate. Instead, productivity growth across major advanced economies has remained weak and volatile, indicating that AI adoption is still largely experimental rather than transformational.

There is evidence of disruption in specific sectors where AI can be readily applied to routine tasks, but the report argues that this often reflects budget reallocations rather than direct substitution of workers. In some cases, firms have even reversed AI-related job cuts after finding that service quality suffered.

While the authors acknowledge that AI could lead to more significant labour market disruption in the future, they stress that this is a risk rather than an inevitability. For now, they conclude, the macroeconomic impact of AI on jobs remains limited, and changes to working practices are more likely to be evolutionary than revolutionary.