In recent months, public concern about the development of artificial intelligence appears to be growing. According to a recent survey conducted by the Pew Research Center in 37 countries, 37% of respondents said they were more concerned than enthusiastic about the growing use of AI in daily life, compared with just 13% who expressed more enthusiasm. There are marked differences between countries. Public opinion is generally most concerned in mature economies such as Australia, the United States, and Canada, where AI adoption is more widespread and skilled jobs are more at risk. Several European countries follow closely behind. Few emerging and developing countries appear concerned at this stage.
Unsurprisingly, specific concerns about job loss appear particularly strong, even though AI can be perceived both as a driver of growth and a threat to certain jobs, as illustrated by the case of South Korea: only 18% of respondents say they are concerned about AI, but 76% anticipate a decline in the number of jobs. This paradox could be explained by a concrete perception of the benefits of AI, as the country directly benefits from its global boom thanks to its semiconductor industry.
The extent of the productivity gains associated with AI adoption remains uncertain, which explains these heightened concerns. The most recent estimates, however, suggest a greater impact than earlier projections, driven by faster-than-expected adoption. For example, the OECD recently revised its projections: while in 2024 it estimated that AI would contribute 0.4 to 0.9% per year to labor productivity in the United States, it now forecasts gains of 0.45 to 1.35 percentage points. At this stage, the effects of AI are most evident in investment: in the United States, investment in information-processing equipment and software contributed on average about 0.85 percentage points to GDP growth in 2025 (or 40% of that growth) and nearly one percentage point in the first half of 2026.
For governments, the effects of AI on public finances are also uncertain and will vary from country to country. Indeed, while it may have a positive effect on GDP growth, AI can also raise interest rates through its effects on the demand for financing, inflation, and growth expectations. It therefore fuels hopes for new fiscal leeway but also raises the need for supporting measures. A sustained increase in productivity would broaden the tax base and reduce the debt burden. However, concerns about employment and the distribution of value added may necessitate spending on training and social protection. Furthermore, the risks identified by developers require assessment and oversight capabilities. Governments may therefore need to finance the transition to AI before reaping its benefits.
