Banking executives have claimed AI can help both their corporations and their clients. AI tools can speed up know-your-customer checks, help improve customer service channels, or spread access to wealth management tools to those outside the high-net-worth-income bracket.Yet the most pressing need for AI, according to Sopnendu Mohanty, co-founder of the Global Finance and Technology Network (GFTN), is to help banks get loans to low-income populations and solve the financial inclusion challenge. GFTN is backed by the Monetary Authority of Singapore. It was previously known as Elevandi, which organized the annual Singapore Fintech Festival. Now, GFTN has expanded its remit to provide advisory services to emerging markets on how to best leverage technology in their finance sectors, while putting forward Singapore as a model. During a Wednesday conversation at the Fortune Brainstorm AI Singapore conference, Mohanty laid out why banks still struggle to expand their lending. “When you are borrowing money from a bank, the bank asks you for collateral. That’s the standard process of lending,” he said. Yet “the low-income segment has no collateral,” meaning the traditional model for financing can’t serve those people.“That model cannot [solve] the global need for credit,” Mohanty said, describing it as the “elephant in the room.”The World Bank reports that only about a quarter of people in low-and middle-income economies engaged in formal borrowing from a bank or credit card company last year. For Mohanty, banks are still focused on how to use AI to increase productivity—such as by allowing them to “fire hundreds of people.”“That’s not what we want AI to do,” he argued. Instead, AI can instead help create “credible, predictive, golden source behavioral data, which will replace the need for collateral.”Several Southeast Asian companies, like Grab, Sea and Goto, are expanding into financial services and serve underbanked customers, leveraging data collected from their main businesses like ride-hailing and e-commerce. Mohanty, who was the first fintech officer at the Monetary Authority of Singapore, the country’s central bank, also pointed to the importance of creating infrastructure to handle identity data. He specifically cited India’s Aadhaar system, which offered every Indian resident a unique ID number—for some, the first ID they might have ever had.Yet Mohanty also hoped that, within a decade, identity might be taken out of the hands of the government. “You don’t need a state to control and deliver that trusted ID,” he suggested, and instead rely on decentralized networks.But in the nearer term, Mohanty expressed a worry about how AI could threaten jobs. “My biggest priority now is going to be upskilling,” he said. “If we don’t upskill our people, we are heading to a massive economic disaster.”Earlier this year, Singapore launched a scheme to improve proficiency in AI across the entire workforce, including in sectors like retail and manufacturing. The country also plans to triple the number of “AI practitioners” to 15,000 over the next few years.On Tuesday, at Fortune Brainstorm AI Singapore, digital minister Josephine Teo said that the pool of “AI practitioners” will include professionals like lawyers and doctors, as well as those from the manufacturing sector. These practitioners “will become the early adopters of AI and then they show their peers how to make better use of it,” she said. This story was originally featured on Fortune.com