AI governance failures threaten banks’ returns

95% of GenAI spend has no outcome as organisations remain in the early stages of adoption.

Banks are entering a new phase of artificial intelligence (AI) adoption where governance, rather than technology, will determine how successfully the industry scales AI.

Executives from ING, Maybank Singapore, Green Link Digital Bank, and former Prudential chief strategy and transformation officer Sherwin Siregar said at the Asian Banking & Finance and Insurance Asia Summit in Singapore on 1 July that AI has already proven its value in areas such as compliance, operations, and customer service.

The next challenge is ensuring institutions can deploy it responsibly whilst maintaining accountability, transparency, and regulatory compliance.

Siregar said organisations should stop viewing governance as something that slows AI adoption. He referred to a study from the Massachusetts Institute of Technology, saying "95% of GenAI spend has no business outcome."

He said organisations remain in the early stages of generative AI adoption and leaders should first understand how the technology works before expecting meaningful returns.

"AI gives us speed, but speed needs direction, and that direction has to come from humans and leaders. Only speed times direction gives you velocity. Otherwise, AI will amplify broken processes and mistakes,” he said.

He added that leaders need to become "AI fluent" themselves so they can decide what should be delegated to AI and what should remain under human judgment.

Anand Sachdev, country manager for Singapore and head of South & Southeast Asia at ING, said AI is already being used in anti-money laundering (AML), know-your-customer (KYC) processes, transaction monitoring, sustainability assessments, and financial markets.

However, he said institutions should first identify where AI creates value instead of using it to automate inefficient processes.

“Applying AI to inefficient or broken processes won’t fix them - it can actually make things worse. You risk significant spend without seeing the return on investment,” said Sachdev.

Sachdev added that banks have a responsibility to ensure AI models remain ethical, explainable, traceable, and free from bias because financial institutions operate in a highly regulated environment.

Mohammed Meraj Khan, group head of digital banking delivery & operations and tech transformation director at Maybank Singapore, agreed that many organisations are still using generative AI mainly as a productivity assistant.

He said larger business benefits are likely to come only after AI becomes embedded into enterprise workflows rather than being used as a standalone tool.

David Song, head of digital business unit at Green Link Digital Bank, said governance should continue after AI systems are deployed because models continue learning from new data.

"After launching, it's not like traditional software development, where you complete a checklist and deploy. You need continuous tracking and guardrails to monitor the AI solution, especially the models, because once real customer data flows in, they keep learning and the model may drift,” Song said.

The panel also discussed AI's impact on jobs, with Sachdev saying history suggests technology creates new roles over time. Instead of replacing workers, he said AI will require organisations to invest in retraining and reskilling employees so they can work alongside the technology.