RBI deputy governor flags concerns surrounding financial institutions using AI

M. Rajeshwar Rao, Deputy Governor of the Reserve Bank of India

The Reserve Bank of India (RBI) Deputy Governor M Rajeshwar Rao has flagged concerns surrounding financial institutions using artificial intelligence.

Rao said that these concerns are design-specific risks such as biases and robustness issues, others are more traditional and user-specific such as data privacy, cybersecurity, consumer protection and preserving financial stability.

Rao delivered a speech at the 106th Annual Conference of Indian Economic Association in Delhi, on December 22.

Many banks have introduced chatbots on their websites and apps that are used mainly for basic communication and handling customer queries. ICICI Bank’s chatbot, called iPal, guides visitors on the website to service options. HDFC Bank’s Eva chatbot assists website visitors with complaints and other services.

Rao said AI is as good as a the data on which is has been trained and thus inherits the issues, biases and errors in its training data. Human beings accumulate such training data over a lifetime of exposures, experiences, evidences, and upbringing. That makes us capable of coming to different conclusion based on the same data set.

"Further, humans can collaborate, combine and brainstorm to reach at an optimal solution. This is not to say that humans are free of biases, but we have embedded checks and balances in the institutional decision making framework to check and prevent them," Deputy Governor added.

On the governance front, Rao said AI may also pose some novel challenges, especially where the technology is used to facilitate autonomous decision-making and may limit or even potentially eliminate human judgement and oversight.

Some of the data and model issues such as prompt injection, hallucinations and toxic output can also have implications for governance frameworks, especially in financial institutions," Rao said in a speech.

"This may necessitate that regulators and the management have a re-look at the frameworks for consumer protection, cybersecurity and data privacy," Rao added.

Some suggestions have been made to overcome governance issues including the option of ‘putting a human in the loop’ to help build trust in the AI driven systems.

The AI models are inherently complex and opaque requiring extra caution to ensure accountability. For this reason, financial institutions may find it difficult to explain an adverse or biased decision outcome from an AI model to customer or supervisors, Rao said during speech.