Regional banks want purpose-built AI - Bangkok Post
Senior banking leaders across Asia-Pacific have made it clear that general-purpose artificial intelligence (AI) is insufficient for financial services, as three-quarters believe AI tools used in banking must be built specifically for financial services, not adapted from consumer applications.
A survey conducted by the global data analytics and credit-scoring company Fico found 77% of banking leaders say it is "extremely important" that AI tools used in banking be built specifically for financial services. Only 23% viewed such specialisation as merely "somewhat important".
"Financial institutions operate under regulatory scrutiny that most consumer technology companies don't face," said Dattu Kompella, managing director in Asia-Pacific for Fico. "Banks need AI models that are explainable, auditable and designed from the ground up for compliance. Off-the-shelf consumer AI doesn't meet that bar."
Banking leaders identified regulatory and governance concerns as the primary barriers to AI adoption, not technology or cost. Data privacy and security emerged as the top concern, cited by 42% of respondents.
A further 21% identified regulatory and compliance risk as their biggest worry, while 21% flagged model transparency and explainability as critical. Only 16% cited computing cost or infrastructure as a barrier.
"The barriers aren't about capability or price," said Mr Kompella.
"Banks are concerned about trust, auditability and the ability to explain decisions to regulators. That's exactly why purpose-built AI for financial services matters. It's designed with those constraints in mind from day one."
Banking leaders see immediate high-impact applications for generative AI.
More than 60% identified fraud and scam detection as the area where generative AI could most improve operations in the next 12 months. A further 21% cited underwriting decisions, 15% customer-facing digital banking services, and just 3% regulatory compliance monitoring.
This focus reflects both the urgency of the scam epidemic across the region and the potential for AI-powered behavioural analytics to detect patterns human analysts might miss, noted Fico.
Banking leaders expect purpose-built AI to become standard quickly. Nearly half (46%) expect task-specific AI models to be deployed in fraud, compliance and underwriting operations within 1-2 years, while 26% expect adoption within 3-5 years. Only 26% report having already deployed such models.
Across the region's leading banks, the priorities are clear. Beyond fraud prevention, 64% identified credit underwriting and lending decisions as the back-office process that would benefit most from purpose-built AI. A further 18% cited regulatory compliance monitoring, 10% customer due diligence, and 8% internal policy and business operations review.
The focus for purpose-built AI is moving from whether to adopt to how to execute a plan, said Mr Kompella.
Banks will need to invest in domain-specific models designed for financial services, supported by cleaner data infrastructure, faster integration with legacy systems, and clearer governance frameworks, he noted.
The poll was conducted in August during the Fico Banking Leaders Forum in Bali, with the respondents 51 senior executives and corporate banking leaders from Asia-Pacific institutions.
