Gen AI: Mitigating the Risks to Succeed

Whether one believes harnessing Gen AI is a Promethean moment or simply the next chapter in one's AI journey, most would agree that the Gen AI genie is out of the bottle and can’t be ignored.?Unlike with other efficiency and/or growth strategy plays, the risks and returns of adopting Gen AI are relatively more challenging to estimate. Clearly the upside potential is material. The cost of the risks, however, is unclear. This is partly due to adverse outcomes that can be hard to identify (e.g., hallucinations and bias). Even if an FI does not implement any Gen AI, it is exposed to heightened external threats by users of Gen AI (e.g., more effective phishing and social engineering attacks).

With any fast-moving technology, FIs are challenged to identify and track sources of risks. As a result, adoption of Gen AI tools requires extensive collaboration spanning the front office to back office (e.g., leaders identifying use cases) to IT and data teams and functional areas, risk, compliance, human resources (e.g., training). The best place to start is to test to what degree current risk guidelines and mitigation protocols can combat the additional risk generated by the use of Gen AI and by exposure to others using it (e.g., fraud attempts). Next, an FI must develop and implement risk specific mitigation protocols and tools to fill the gaps.

We, at Celent, outline five strategies to mitigate risk as well as three foundational elements (figure below). For details, please see Celent report, Generative AI – Mitigating Risk to Realize Success in Banking.

Figure: Mitigating Gen AI Risks