Singapore’s Radical Bet: Training 100,000 Workers to Be ‘AI Bilingual’ by 2029

When Singapore makes such an announcement, it usually exudes a certain level of confidence. Not theatrical, not boisterous. Just a quiet assurance that the budget is already in motion, the partners are already lined up, and the planning has already been completed.

This week, during the ministerial budget debate, Minister Josephine Teo presented a figure that seems almost insignificant until you consider it: 100,000 workers will have received artificial intelligence training by 2029. That’s not a pilot in a nation of about six million people. It’s a generational wager.

Teo has begun referring to the National AI Impact Program’s concept as “bilingual,” despite the fact that it has nothing to do with Mandarin or Malay. The ability of an accountant, lawyer, or even a receptionist to continue doing what they have always done while an AI sits next to them and completes the tedious portion is what she refers to as bilingualism between human expertise and machine fluency. She stated, “Not every citizen needs to be an AI engineer,” which is the kind of statement that goes unnoticed but has greater significance than the surrounding statistics.

Over 24,300 mid-career Singaporeans have already been placed in technical roles thanks to the TechSkills Accelerator program, also referred to locally as TeSA. The program has been in operation since 2016. The audience is different now. For the first time, professionals who have spent decades rejecting the notion that software could replace their workflows are the target of TeSA instead of engineers. First, accountants. Next, attorneys. Next, HR.

The pilots, which are planned for the first half of 2026 and are being co-developed with the Singapore Academy of Law and the Institute of Singapore Chartered Accountants, indicate that the government has completed the laborious task of gaining support from professional associations prior to going public.

Teo cited KPMG senior manager Geraldine Lau, who developed an AI agent in late 2024 to retrieve regulatory announcements from the Singapore Exchange, to support her argument. It cut the time needed for audit risk assessment in half. Lau doesn’t write code for production. When combined with a functional AI tool, her keen intuition about which data points are truly important in an audit proved to be the key. It’s the kind of little, almost unremarkable story that most likely does a better job of explaining the bilingual concept than any policy paper.

As this develops, there’s a feeling that Singapore is attempting to change the course of events before they happen here by reading the world’s fear of AI displacement. People should be trained early. Provide instruction on data governance, responsible use, and the murky ethical edges. To avoid making it seem abstract, make it sector-specific. And use the organizations and agencies that people already have faith in. Depending on how quickly technology advances underneath them, 100,000 by 2029 may prove to be ambitious or modest. Whether AI fluency acquired in 2026 will resemble AI fluency in 2029 is still up in the air.

Naturally, similar promises have been made by other governments, and the majority of them have not lived up to expectations. The operational details that are already available, such as the named partners, named sectors, and named timelines, give this one a slightly different texture. Singapore has a tendency to place wagers that are obvious in retrospect but appear unremarkable on the day they are announced. That pattern might apply to this one. Or it may not. In any case, the experiment is now underway.