Inside Singapore’s AI bootcamp to retrain 35,000 bankers

SINGAPORE - Mr Kelvin Chiang knew the five agentic artificial intelligence (AI) models built by his team could do in 10 minutes what used to take a private banker an entire day. With that in place, he went to show Singapore’s banking regulator the safeguards would sufficiently control the risks.

Before rolling out the tool that drafts documents for relationship managers at the private wealth arm of OCBC Bank, he took his team of data scientists less than 1.6km across the city’s downtown finance district to the Monetary Authority of Singapore (MAS). Mr Chiang highlighted the steps the bank would take if things went wrong with their blueprints and how staff would react if the system hallucinated.

After the presentation and discussion with representatives at MAS, he left feeling that engineers at the regulator embraced it, said Mr Chiang, 52, the head of financial crime compliance analytics at Bank of Singapore. “They were quite happy.”

Singapore’s AI efforts underscore a unique balancing act: The Government is working hand in glove with the country’s largest banks to teach new technology skills. The unspoken intent is to limit the large-scale job losses seen at some financial firms in the United States and Europe. At stake is the future of thousands of bank workers at a time when giants of the industry, such as Goldman Sachs, have told staff to expect more job cuts as they take advantage of AI.

Singapore’s Minister for National Development Chee Hong Tat, who is also deputy chairman of the MAS, has said the three large banks – DBS Bank, OCBC and UOB – are retraining all 35,000 of their domestic staff over the next one to two years.

“The Government is doing something about it because they realise that this capability and this change is infusing potentially a lot of fear,” said Ms Violet Chung, a senior partner at McKinsey & Co. “Given what we’ve seen in other markets like the US where (there are) much more aggressive job cuts and reductions, the Government is essentially aware that we need to do something as a country for these leading companies.”

For David, 39, who manages money for a bank’s wealthy clients, smarter technology is helping him work faster than at any time in his 16-year career, but that is raising the expectations of his bosses and has left him feeling unsettled. He would typically spend around an hour preparing a single customer order form – now, it takes him 10 to 12 minutes. That means more time in front of clients, he said. He declined to give his full name discussing private details about his work.

Banks will be able to increase the number of customers that each of their relationship managers can cover, say from 50 up to 60 or 70, according to Mr Mohan Jayaraman, senior partner at Bain & Co.

“The argument is that if you are able to spend more time with customers, you get better coverage,” he said. “You are increasing the ambit of people covered with your proposition, and at the same time giving the relationship manager more ability to earn on a larger pool.”

Singapore’s National Jobs Council, which sprang out of the country’s employment needs during the Covid-19 pandemic, now also works with the not-for-profit Institute of Banking and Finance (IBF) as a key partner to foster job growth for the finance industry.

In some cases, as roles change, the institute works with and pays for finance firms to map out how staff can take on redesigned roles or pivot to new career paths. In one scenario, it offers banks up to 90 per cent salary support to reskill mid-career staff or new hires, according to its website.

MAS said it works closely with IBF, financial institutions and unions to prepare the sector’s workforce for AI adoption.  

OCBC and UOB have said they are not making AI-related job cuts. DBS has said it will not lay off permanent staff, but expects to reduce about 4,000 temporary and contract roles over the next three years as vendor contracts end. A Bloomberg Intelligence analysis forecasts DBS will make bigger AI-derived cost savings than its peers and help boost pre-tax profit by up to $1.6 billion, or about 17 per cent, in the next few years.

Staff at DBS now use an internal AI assistant that handles more than one million prompts a month. It has also built role-specific tools, including one for customer service officers that has reduced call handling time by up to 20 per cent. UOB has given all its employees access to Microsoft Copilot and deployed more than 300 AI use cases across the bank. 

Watching from the inside, Ms Vania Lim, 22, spent last summer as an AI intern in the global payments team at HSBC Holdings. She is optimistic about new roles emerging in digital assets and improving use cases for wallets, but is worried that some of the skills for junior hires are not so unique any more.

“Everyone is so well-versed with AI nowadays,” she says. “It’s no longer so much of a competitive advantage.”

Older staff like Woon Leng, in her 60s, says the training to keep up with new technology feels like one more thing piled on top of an already demanding job of managing a branch. 

She has worked at a local bank for more than 40 years and now uses ChatGPT at her branch to help answer customer questions about products. Staff also have to take online courses on AI after work as it is too busy during the day, and try to pass a test afterwards, she added. She declined to give her full name discussing her employer.

Many banks are choosing not to fill roles when staff move on either within the firm or outside it, said Associate Professor Walter Theseira, an economist at the Singapore University of Social Sciences.

“Of course, companies don’t like to do layoffs and redundancies, especially if they are major firms that the Singapore economy and Government watch very closely,” he said. “But firms can pursue a headcount reduction simply by not filling roles when you have natural attrition.”

At OCBC, there were fewer than a handful of people in the AI lab when Mr Kenneth Zhu joined in 2018. Eight years later, the 36-year-old executive director of data science and AI oversees a lab that sits within a data office of more than 100 staff. Some 400 models make six million decisions every day, from flagging suspicious transactions to scoring credit risk and filtering out false positives in anti-money laundering systems that once chewed up compliance officers’ time.

While it is too early to assess if the Singapore approach will pay off for banks and their staff, Dr Jochen Wirtz, vice-dean of MBA programmes and professor of marketing at the National University of Singapore, said the technology’s potential remains huge and cautioned against firms holding back.

“My hunch is that we haven’t really scratched the surface yet,” he said. “Banks would be completely stupid now to load up on employees who they will then have to let go again in three or four years. You’re much better off freezing now, trying to retrain whoever you can.”

At UOB, the bank is busy building AI fluency across the organisation as well as training staff, such as via an internal Better U Pivot programme.

“Inaction is not really an option for us or anybody,” said Mr Alvin Eng, its head of enterprise AI. “It’s a slow path to irrelevance.” BLOOMBERG