Banks have made cautious but deliberate moves in adopting generative AI within their credit businesses, with McKinsey and the International Association of Credit Portfolio Managers surveying 44 institutions in late 2024. These included megabanks, super-regionals, and core regionals. While 52% of respondents said gen AI adoption was a priority backed by investment and hiring, only a few have achieved full-scale deployment. Institutions have made the most progress in utilizing AI for summarizing data in early warning systems and credit decision-making. However, many banks are still in early phases, with just 12% of North American banks having deployed any use case. The report noted that institutions focusing too early on return on investment were more likely to abandon the technology. Meanwhile, agentic AI, which integrates decision-making capabilities across operations, is beginning to show promise, especially in underwriting and customer engagement.
Despite growing interest, most banks remain cautious due to risks such as data security breaches, model inaccuracies, and uncertain financial outcomes. Model validation challenges, limited historical data, and complex stakeholder involvement continue to slow adoption. The report stated that more than two in five institutions had slowed the development of use cases due to disappointing outcomes. Still, some banks are laying strong foundations, focusing on modular architecture, talent hiring, and test-and-learn pilots. The report concludes that banks that align internal support and apply AI to broader workflows are beginning to see measurable improvements, particularly in productivity and customer experience.
