For years, the economics profession has been the designated skeptic in the AI hype cycle. While technologists promised transformation and markets bid up chipmakers to stratospheric valuations, most economists shrugged, pointing to productivity data that stubbornly refused to move. That consensus is now cracking.
A new Financial Times survey of 94 economists — spanning central banks, academia, and major financial institutions — reveals a striking shift in expectations. The majority now believe artificial intelligence will deliver measurable GDP gains across both advanced and emerging economies by 2027, with benefits reaching far beyond the obvious winners in the United States and China.
The productivity puzzle, reconsidered
The skepticism was always grounded in history. Previous technological revolutions — electrification, computing, the internet — took decades to show up in aggregate statistics. Robert Solow's famous quip that "you can see the computer age everywhere but in the productivity statistics" haunted every AI bull case. But the surveyed economists now argue that generative AI's deployment speed is categorically different. Unlike previous general-purpose technologies that required massive capital expenditure and workforce retraining before yielding returns, large language models can be integrated into existing workflows with minimal infrastructure investment.
The median forecast suggests advanced economies could see an additional 0.4 to 0.8 percentage points of annual GDP growth attributable to AI adoption by late 2027. For context, that would represent the most significant productivity acceleration since the late 1990s dot-com boom — and this time, the gains are expected to be more evenly distributed across sectors.
Beyond the obvious winners
Perhaps more surprising than the bullishness itself is where economists expect the gains to materialize. The survey respondents identified healthcare administration, legal services, and government bureaucracies as likely to see the largest efficiency improvements — sectors not typically associated with technological dynamism. Several respondents noted that AI's ability to process and summarize large document sets could unlock productivity gains in precisely the knowledge-work segments that have been most resistant to automation.
Emerging markets, long assumed to be AI laggards due to infrastructure constraints, are now seen as potential outperformers. Economists pointed to India, Indonesia, and Vietnam as economies where AI could leapfrog traditional development bottlenecks, particularly in financial services and education delivery. The reasoning: these markets have younger, more digitally native workforces and less legacy infrastructure to replace.
The caveats remain substantial
Not everyone is convinced the optimism is warranted. A vocal minority of respondents — roughly one in five — warned that the survey may be capturing peak enthusiasm rather than durable conviction. They pointed to the gap between AI capabilities demonstrated in controlled settings and real-world deployment challenges, including data quality issues, regulatory uncertainty, and organizational resistance to change.
Several economists also flagged distributional concerns. Even if aggregate GDP rises, the gains may accrue disproportionately to capital owners and highly skilled workers, potentially exacerbating inequality trends that have defined the past two decades. The survey found little consensus on whether AI would prove net positive or negative for median household incomes.
Our take
The profession's shift matters less for its predictive accuracy — economists are notoriously poor forecasters — than for what it signals about institutional sentiment. When central bankers and IMF researchers start building AI productivity gains into their baseline models, policy frameworks follow. Interest rate paths, fiscal projections, and regulatory postures all bend toward the new consensus. Whether the optimism proves justified or not, the economics establishment has now placed its bet. The rest of us will find out if they were right.
