Banks must ensure that humans remain accountable for decisions made with the help of artificial intelligence, Reserve Bank governor Sanjay Malhotra said on Tuesday, warning lenders against treating algorithms or technology vendors as responsible for their actions.Speaking at the annual Fibac event in the financial capital, Malhotra said banks adopting AI must ensure that the technology supports human judgement rather than simply replacing it. He also asked lenders to build “meaningful human oversight” into AI systems from the design stage itself.“For a bank’s decision, the ultimate responsibility has to lie with the bank and not with the vendor or with the algorithm. We cannot say that it is the model who decided it. That can never be an acceptable answer, whether to the bank, to the customer, or to the regulator,” Malhotra said.“Meaningful human oversight, the ability to explain, to intervene, and where necessary to override, must remain a design principle and not an afterthought,” he added.
AI should augment human judgement
Malhotra said the RBI‘s approach is to see AI as a capability that banks should responsibly harness rather than a risk that needs to be contained. He urged lenders to embrace the technology instead of remaining on the sidelines.At the same time, he cautioned that greater reliance on AI could gradually weaken human judgement and accountability.“(AI must be) used to augment rather than merely to replace human judgement,” Malhotra said, amid wider concerns over the effect of AI on employment intensity across sectors of the economy.Indian lenders have increasingly adopted AI for functions such as credit underwriting and customer service. Some banks are also hoping that their dependence on human capital will decline over time.Malhotra, however, pointed to how AI could be used to make human employees more effective. A relationship manager equipped with an AI system that identifies the appropriate product and highlights the relevant risk flag, he said, could serve more customers efficiently.
Cyber security, bias among AI risks
The governor also highlighted risks that banks could face as they expand their use of AI. Cyber security is one of them, with Malhotra referring to a recent incident involving AI itself attacking the system.Biases within AI models are another concern. Malhotra warned that models could develop preferences for or against particular geographies, occupations or even communities.He also cautioned against “herding”, which could emerge if a small number of financial models or technology vendors end up supplying platforms across the system. A bias could then be perpetuated across the system, he said.Malhotra asked banks to approach data privacy as a responsibility that goes beyond merely meeting legal requirements. Lenders must also strengthen their management of AI-related risks while working with external partners.AI platforms, he said, should also be capable of explaining why they recommend a particular decision on a proposal.
RBI seeks inventory of AI models
Malhotra said the RBI should know which models are actually operating within banks and asked lenders to maintain a complete inventory of them.Banks should also have a board-approved AI governance policy that establishes clear accountability for outcomes, rather than focusing only on the procurement of technology, he said.Meaningful human oversight must be retained at every stage where an error by an AI system could cause material harm to a customer or raise concerns about financial stability, Malhotra said.
