As AI adoption rises, job fears grow across the US

AUSTIN, Texas – It did not take long for the myth to crumble.

A year ago, artificial intelligence (AI) was dismissed as too dim-witted for the front lines of finance. But Mr David Yin, a partner at Informed Ventures, has now backed an outfit that runs bookkeeping, tax compliance and financial analysis for start-ups and small firms, powered by both generative agents and licensed professionals.

Two young coders helm the start-up that provides end-to-end accounting at a fraction of the fees traditional firms charge for transaction sorting, bank reconciliation, book closing and analysis.

“As an investor, I have been shocked at how quickly companies are progressing in underlying technology,” said Mr Yin, a 35-year-old Singaporean who has worked in the San Francisco Bay Area for around five years.

“A year ago, for instance, it was said that AI was too dumb to do finance, but over the last six months, people have started using AI assistants to do just that. I’m not saying they are perfect, but they have started becoming very powerful.” 

Mr Yin’s Menlo Park, California-based venture capital fund has poured around US$300 million (S$384 million) into early healthcare and fintech bets.

The accounting start-up employs certified public accountants to vet the bots’ output, keeping the essential human oversight on taxes and audits. 

Still, Mr Yin spots the chill. Firms are not hiring – they are pruning. 

“A lot of us underestimate how powerful AI is and how good the models could become in the next three to five years. Most of the companies I’m talking to are trying not to hire people. They are trying to cut people, if anything.”

A ripple of unease is spreading across the US economy as the prospect of AI-related job losses conjures up dark scenarios like offices without young people because entry-level jobs are no longer open. And, maybe, even mass unemployment if the trend accelerates.

Over a million jobs were lost in the US in 2025, the highest number since 2020, when the Covid-19 pandemic caused economic disruptions worldwide. 

AI was cited in at least 54,694 of those layoffs, according to outplacement firm Challenger, Gray & Christmas. The numbers could be higher.

As AI began to be more widely adopted across the economy, several influential corporate figures spoke in 2025 of its game-changing capabilities.

AI will do nearly everything within a decade – making, moving and growing things – while humans choose to take on some jobs like playing professional baseball, Microsoft co-founder Bill Gates said in an interview in March 2025.

People will be free to choose to work only if they feel like it, said Tesla and SpaceX founder Elon Musk.

“In less than 20 years, maybe even as little as 10 or 15 years, the advancements in AI and robotics will bring us to the point where working is optional,” Mr Musk said in a November 2025 podcast.

Mr Dario Amodei, the chief executive of Anthropic, whose AI can almost code like a human, predicted that half of all entry-level white-collar jobs would vanish within five years.

The job of a CEO would not be immune from being assumed by AI either.

“What a CEO does is maybe one of the easier things… for an AI to do,” Google CEO Sundar Pichai said in a November interview.

Mr Sam Altman, the CEO of OpenAI, which took the world by storm with its large language models like the GPT series, said: “Shame on me if OpenAI is not the first big company run by an AI CEO.”

A prominent dissenter was Mr Jensen Huang, the CEO of Nvidia, which designs the advanced chips that power AI.

Asked if AI could replace him, he shot down the idea. “Absolutely not,” he said during an October 2024 event, suggesting that AI could do 20 per cent to 50 per cent of someone’s job a thousand times better, but could not do all of it.

That may change.

At the Massachusetts Institute of Technology, researchers built a digital “twin” of the US labour market and counted jobs which AI could perform at a cost that is either competitive or cheaper than human labour.

Their study, released in November 2025, found that one in nine jobs – across finance, healthcare, logistics, human resources and professional services – could already be done by AI. That is around 151 million workers, representing more than US$1 trillion in pay.

Stanford University’s Dr Erik Brynjolfsson and his co-authors of a report looking at labour market shifts following the widespread adoption of generative AI used the term “canaries in the coal mine” to describe what is happening to early-career workers, aged 22 to 25, in occupations highly exposed to AI, such as working in software development and as customer service representatives. 

Their professional demise would serve as an early warning of impending danger, just like the death of the birds in the expression alerted workers to evacuate the mine before toxic gases killed them, too.

Young workers saw a 13 per cent dip in employment from late 2022 to mid-2025, the Stanford team documented in their November 2025 research report, and concluded that a structural transformation of the labour market was well under way.

Mr Yin said he had observed that companies are trying not to hire newbies.

“With AI handling basic work, who needs an intern anymore? Who needs a copywriter? Who needs a researcher? So that makes it very difficult for young people to find their first job,” he said.

“It’s a big change – it used to be almost a social contract that a company would invest in young people, even though it’s only after two or three years of apprenticeship and training that they start to add value,” he added.

The scale of the change and its impact on jobs are hotly debated. 

Experts at the Special Competitive Studies Project (SCSP) warn against sweeping predictions like the imminent loss of 100 million jobs, nearly half the US workforce.

The non-partisan US think-tank and private foundation focused on technology and security was founded by former Google CEO Eric Schmidt in 2021.

“I would be very cautious about any predictions because we don’t know the full scope of adoption or the rate of change. What’s also not really accounted for is how jobs can change in place versus automation away from jobs,” said Mr Martijn Rasser, SCSP’s senior director for economy.

“For example, you’ll still need accountants, but what accountants do will change,” he said.

Dr Ryan Carpenter, a special adviser focusing on workforce issues at SCSP, said some uncertainties have to be faced. 

“It’s like what we saw with the internet boom 25 years ago. Did certain jobs go away? Yes, that’s every time there’s broad technological adoption. So, we won’t have nearly as many coders 20 years from now as we do today. But that does not mean these people will be unemployed for the rest of their lives; it’s a matter of reskilling.

“The bigger challenge is that the timeline for this transformation will probably be shorter than what we experienced, for example, in the industrial revolution. That’s probably what’s causing more anxiety,” Dr Carpenter added.

But there would be a net jobs gain over time, he said.

“Where I see a particular opportunity is in manufacturing – because it’s a bipartisan priority for America to make things again.

“We don’t have the workforce to be able to execute that now, so that will require new people moving into the manufacturing space.

“For widespread adoption of robotics, we’ll need technicians able to work alongside these robots, to reprogram them, maintain them.” 

Lawmakers, often criticised for being slow in regulating transformative tech, are trying to get a handle on the challenge. 

A new Bill introduced in the US Senate in October 2025 is seeking to compel companies to report to the government every quarter how many jobs were displaced due to AI, how many people were hired for AI jobs and how many retrained. 

Mr Yin said the government could also lean in to ensure that the education system produces AI natives.

The Trump administration’s AI Action Plan has made a start by roping in tech majors like Google and IBM to develop and support AI literacy programmes in schools and teacher training. 

The silver lining in these times, Mr Yin said, is that it is much easier to start companies now, using resources like cloud computing. These deliver on-demand computing services – like servers, storage and databases – without the need to own physical infrastructure.

AI could enable “solopreneurs” and make it possible to service niche markets, he said.

“In the past, it didn’t really make sense to build a tech app just for the Singapore market. But now, I’ve been telling some Singaporean founders that it’s possible to create a lot more Singapore-only opportunities. It’s not always jobs, but there are more opportunities,” he said.

A one-man shop, for instance, could create a mini app to order hawker food. 

“You could charge a thousand people $5 a month, let’s say, and you’d be making $5,000 monthly.”