Speaking at the 106th annual conference of the Indian Economic Association in Delhi last week, Rao said that while GenAI has transformative potential, the potential impact is on productivity, employment and income distribution. Although its proponents expect broad economic and social benefits, there are skeptics. Concerns such as rising unemployment and challenges in reallocating resources during transition.
Rao outlined specific concerns and expectations for financial institutions implementing AI in their processes and decision-making. These include dependency, resilience issues, data privacy, cyber security, consumer protection and maintaining financial stability. AI gets biases and errors from its training data, so regular fair audits are important, he said. He highlighted the challenge of AI models being black boxes, difficult to decode for auditing and monitoring reviews.
„AI could present some new challenges to governance, especially when technology facilitates autonomous decision-making and limits or eliminates human judgment and oversight,” Rao said. With the aim of balancing innovation and responsible use of technology, he outlined ten aspects for financial institutions to implement AI models. These include fairness, transparency, accuracy, consistency, data privacy, interpretability, accountability, resilience, monitoring, updates and human oversight.
He stressed that combining these aspects would build public trust and ensure responsible implementation of AI, and called for a supportive and globally collaborative regulatory framework within that framework to guide the implementation of AI.