With enhanced customer service and mounting cost pressures, financial services institutions are looking at Artificial Intelligence (AI) and Machine Learning (ML) to improve cost and operational efficiencies while mitigating business risks.
According to the consulting firm KPMG that issued a research paper on “Modern Risk Management for AI Models” accentuating adaptable key pillars of risk management that aid a bank’s framework. The report elaborates on how Banks should identify an appropriate risk appetite while establishing a variety of such technology-based models for bespoke assignments.
“With model induced decision making, widespread differences in the approach have been taken by banks on AI risks around bias, interpretability and other challenges. Hence, banks should conduct enterprise wise training programs to educate all stakeholders including the senior management on key aspects of AI/ML such that they can gauge the risks better,” said Rajosik Banerjee, Partner and Head, Financial Risk Management, KPMG in India.
“It is imperative that a cross functional governance framework must be established with clear definitions of roles and accountabilities. There are key elements which need to be specifically tested during the model’s life cycle, including e.g. during design, implementation, operation and validation. AI and ML models will entail risks and challenges that both banks and financial institutions do not yet sufficiently consider in the existing Model Risk Management (MRM) approaches. Also increasing regulatory requirements means, they are now having to adapt and expand their existing MRM approaches in specific areas,” said Banerjee.
Banks would need to develop the skill set inhouse or bring in external experts. External subject matter experts can help them benchmark themselves with peer banks around risk management, controls and governance framework enhancements. This should help validate techniques of AI/ML models.
To sum up, over the next few years, the regulatory scrutiny around AI/ML models is expected to grow as banks increasingly start adding more AI/ML models into their inventories, according to KMPG.
As of 2021, there are detailed proposals from the EU as well as Federal Trade Commissions (FTC) for stricter AI regulations.
It remains to be seen how this area evolves and regulations shape up over the next few years, but international guidance or standards in this area will be helpful in setting the minimum benchmark for MRM practices across jurisdictions.