Artificial intelligence took center stage at the IMF-World Bank meetings in Bangkok, where financial leaders focused on a basic question: who will actually benefit from the technology? The World Bank is urging developing countries to build “small AI” models that can run on mobile phones, work with low connectivity, and provide information in local languages.
These models use specialised systems trained on narrower datasets for specific purposes. That makes them different from broad AI systems designed to handle many tasks. Paschal Donohoe, the World Bank’s chief knowledge officer, said small AI could help developing countries move from centuries of economic development to decades.
Small AI built for local needs
The World Bank’s argument is simple: AI will reach more people when it works in the places where people already live and communicate. A small model available on a mobile phone could deliver information in a local setting, even when the internet connection is weak. It could also respond in the local language of the community.
One example comes from Indian farmers involved in the World Bank’s agricultural scheme, AgriConnect. They need soil analysis in their own language, but there will not be enough scientists and agricultural experts to analyse soil for every small farm in India. Small AI could help deliver that information through mobile phones.
Donohoe described the possible economic effect in ambitious terms: “We argue that economic development in the countries that we serve and partner with that could have taken centuries to make happen, the potential is there for that economic development now to happen within decades.”
That promise depends on basic infrastructure, however. Kristalina Georgieva, the IMF managing director, called on governments to provide access to electricity, digital infrastructure, and safety measures before AI can succeed. “You need access to electricity. You need access to digital infrastructure. People have to be geared to adoption. And the government should be able to give some safety, protect against bad actors so that adoption can work,” she said.
The infrastructure and jobs challenge
The gap between AI’s promise and the infrastructure available today remains large. Almost a third of rural schools in sub-Saharan Africa lacked electricity, while more than two-thirds did not have a consistent internet connection. Georgieva said access to electricity, digital networks, and cybersecurity would be essential if AI benefits are to reach people.
Jobs formed another central theme at the annual meetings of the World Bank and IMF. Georgieva warned that 40% of jobs worldwide are likely to be either augmented or eliminated by AI over time, with the impact particularly acute for young people starting their careers.
“When we look at this, what we recognise is that the people to worry about, first and foremost, are young people that are coming into the labour market,” Georgieva said. She also argued that the net impact of AI on jobs would be positive, but warned that services do not pay a lot.
An IMF study found that about 10% of jobs might be transformed or newly created by AI. It identified three main trends: some jobs being transformed or created, repetitive tasks disappearing, and service-sector jobs benefiting from an investment boom.
Georgieva gave a clear example of the tension. “So you get young people who normally would have gone into professional routine jobs now serving coffee in Starbucks. And that is not what I think young people in the world aspire for.”
Safety concerns grow alongside investment
Tiff Macklem, governor of the Bank of Canada, compared AI’s effect on employment with the impact of open trade. Open trade generated wealth but also caused job losses, he said, and the same pattern could emerge with AI. “Without a system to help them cope with the loss you get a backlash,” Macklem said.
That concern was visible in Bangkok, where many people marched to draw attention to growing inequality and their frustrations. At the same time, Ajay Banga, the World Bank president, said the bank had mobilised $120bn (£90bn) of private capital in the past year. Many countries, including the US and the UK, were cutting their aid budgets, adding pressure to the effort to expand access.
AI safety also drew blunt warnings from Jack Clark, co-founder and head of policy at Anthropic. Clark proposed that the world’s top AI companies coordinate to pause development of advanced AI systems. He described the current state of development this way: “I think it’s worse than you describe. I think it’s Formula 1 at night without maybe brakes in the cars.”
Georgieva had compared AI risks to a Formula 1 race car that might go off track. Clark also pointed to a gap in regulation: “And it’s remarkable to me that the toys I buy for my children are regulated more than the literal superintelligence I work on at my day job.”
The discussion even reached the way AI agents behave. Georgieva said she used AI to track calories and lose weight, then joked about whether AI agents could become more polite. “So can we work on, like, their manners?” she asked. Clark replied, “We have this dial between sycophantic and rude. So we’ll spin it around the other way.”
The Bangkok discussions left two ideas side by side. Small AI could bring useful information to farmers, schools, and communities with limited connectivity, while AI’s wider effects will depend on electricity, digital infrastructure, cybersecurity, safety rules, and support for workers. The technology may help developing countries shorten their path to economic growth, but access and protection will shape who gets the benefits.
Based on
