FT Poll: 54 Percent of 94 Economists Expect AI Gains to Spread Across the G20 in the Next Decade

The economic argument over artificial intelligence has mostly been about two countries: whoever builds the best models wins everything. A Financial Times survey of 94 economists across Asia, Europe and the United States suggests most of the profession disagrees. Fifty-four percent expect many or nearly all G20 countries to see significant AI-driven economic and productivity gains over the next decade — benefits reaching far beyond the US and China, where frontier systems are built.

The reasoning is that economies do not need to invent frontier AI to profit from it. Gains arrive through business adoption, imports of AI-enabled goods and services, knowledge spillovers and falling prices — the same diffusion that eventually carried the benefits of electricity and computers far from the countries that invented them. Jumana Saleheen, chief economist at Vanguard, argued that economies able to adopt AI as a plug-and-play tool can share in the gains, albeit with a lag, while Martin Beck of WPI Strategy said the range of channels points to positive income growth across the G20 even if the dividend differs by country.

The scale question draws on World Bank research published in June, which estimated that the productivity gains from AI could surpass those of electrification. That matters because global productivity growth has been decelerating for two decades — from an average of 1.4 percent a year in the 2000s to 1.1 percent in the 2010s, and a projected 0.8 percent this decade absent a major technological shift. If AI delivers at anything like the upper range, it reverses the slide.

But the second number in the poll is the sharper one: thirty-two respondents — roughly a third — flagged African countries as the ones likely to fall behind, with Nigeria, Somalia and Ethiopia among those mentioned. The vulnerabilities are structural: weaker power and internet infrastructure, thinner capital markets, and economies that depend on outsourcing exactly the kind of routine cognitive work — customer service, back-office administration — that AI absorbs first.

On jobs, the surveyed economists split from the doom-loop narrative. Administrative, customer-service and junior professional roles are most exposed to disruption, most agreed, but few expected a large or lasting rise in unemployment — new roles tend to emerge, and AI may offset demographic drag in economies with shrinking working-age populations. Dame Diane Coyle of the University of Cambridge said the technology will change the character of jobs but is unlikely to trigger a jobs apocalypse.

The dissenting notes deserve attention too. Several economists warned the gains could concentrate among shareholders of advanced companies in large economies, shifting wealth from workers to capital, and the Bank for International Settlements has cautioned that the AI-driven investment and stock-market boom could reverse, dragging growth down with it. The survey lands in the same week Anthropic pledged 5 percent of its shares to a public benefit fund — an acknowledgment, from inside the industry, that who captures the value of AI is an open question. Ninety-four economists have now put numbers on that debate: a majority sees the gains spreading, and a bloc of one in three is preparing for a world where they do not.