Broad adoption of artificial intelligence could lift Latin America’s GDP by 5.1% over a decade and wages by 4.2%, the Inter-American Development Bank (IDB) said in a new report.
The figures come from the bank’s flagship study, From Digitalization to Artificial Intelligence: Turning Promises into Productivity, the 2026 edition of its Development in the Americas series. The IDB makes the gains conditional on stronger institutions, digital skills, data and infrastructure, and says AI is not “plug-and-play”: firms and governments have to redesign processes, retrain staff and change how they are managed before it pays off. For a region whose productivity growth has been weak for six decades, the report reads as a reform checklist.
The report lands in a region that multilateral lenders describe as stuck. The UN Economic Commission for Latin America and the Caribbean (ECLAC) in August confirmed a regional growth forecast of 2.2% for 2026, down from 2.4% in 2025, and said the slowdown showed the region remained trapped in a pattern of low growth. Last year, ECLAC calculated that another year of growth around that rate would leave average expansion at just 1.6% for 2017-2026, a historically weak performance. “To escape the trap of low capacity for growth, the region needs more ambitious productive development policies,” ECLAC executive secretary José Manuel Salazar-Xirinachs said in December.
In a scenario of high adoption and a large impact on productivity, the region’s total factor productivity could rise by 2.3%, the IDB report says. When the bank previewed its findings in September, it put the GDP gain at only 0.3% if adoption stays limited and the productivity effects small. Wages depend on whether workers can move into the jobs that grow: they could rise by 2.3%-5.3% over a decade if they can, and fall by 13.5%-20.9% if they cannot.
The region starts well behind the rich world: some 35% of its adults said they had used generative AI in 2025-2026, against 43% in European OECD countries and 58% in the US. Only 26% had recently dealt with their government online, against 70% across the OECD, and 27% had recently bought something online, against 66%. Of the digitally engaged firms the IDB surveyed, 63% use generative AI, but 82% of those rely on off-the-shelf tools or light adaptations rather than systems built into their operations.
The World Bank reached a similar conclusion in its October outlook for the region, which also forecast 2.2% growth this year. It named AI as a possible new source of productivity but warned that access to the technology would not lift output on its own. The main constraint, the bank said, is “absorptive capacity”, meaning how well governments, companies and workers can put new technologies to practical use. Building it will require better human capital, digital infrastructure, usable data and improved management. OECD simulations cited in the report suggest AI could raise real income per person in Colombia and Mexico by only 0.3%-2.0% over ten years. Medium-sized and large Mexican firms have moved further ahead on adoption than companies in comparable countries, the bank found, but still lag well behind the US.
The gap to close is wide. Labour productivity per hour in Latin America stands at just 38% of the OECD average, a figure set out as ministers met at an OECD regional summit in Montevideo in September. AI readiness is also uneven. The Latin American Artificial Intelligence Index, compiled by Chile’s National Center for Artificial Intelligence with support from ECLAC, ranked Chile first among 19 countries in 2025, ahead of Brazil and Uruguay. It warned that gaps in infrastructure, governance and specialised talent threaten to widen the digital divide, and noted that Brazil alone holds more than 90% of the region’s high-performance computing capacity.
“Governments are going to play a fundamental role because they have to promote AI adoption when the digitalisation agenda is not yet complete,” IDB lead economist Julián Cristia told El País. He gave tax as an example. “Today 90% of people in the region can file their taxes online,” he said. That saves time, he argued, but it is an administrative improvement rather than the kind of organisational change that lifts productivity.
“The big change is for governments to use pre-filled returns, built from public and private data sources, so that the taxpayer only has to accept the return,” Cristia said. He also admitted the region lags badly on human capital, which shows in AI adoption rates that track education levels.
“Artificial intelligence can become a powerful accelerator of development, but its benefits will not come automatically,” said IDB Group president Ilan Goldfajn. The bank’s Partnership for AI Adoption brings governments from the region together with Google, Anthropic, Microsoft and Nvidia, and its September meeting in New York drew leaders from 12 countries, including Brazil, Chile, Colombia and Uruguay.
Uruguay opened a national AI centre in September with $10mn from an IDB loan, and an IDB study estimated that Colombia’s data centre build-out could add between $6.2bn and $42.1bn to GDP in 2024-2030. Chile’s state infrastructure company is sounding out power suppliers for an AI training campus in Antofagasta that could grow from 50MW to 1GW. The report cites early results elsewhere: an AI-based app cut users’ depression by 20% in Mexico, an electronic single window raised Costa Rica’s export values by 46%, and a digital health platform in São Paulo serves 3.4mn users and has delivered 2mn teleconsultations.
To manage the risks, the IDB recommends targeted training at firms with growth potential and more income support for displaced workers, education systems that build adaptability, more public and private spending on cybersecurity, and rules, including on data privacy, that weigh the costs and benefits to households carefully. Its three priorities are to strengthen the enabling conditions, to move governments beyond isolated pilot projects while keeping human oversight, and to help firms adopt AI through better access to finance, technical assistance and training.
.jpg)
No comments:
Post a Comment