Generative AI has rapidly moved from novelty to mainstream in financial services. To avoid missteps that could derail the progress, here are six key takeaways from a hectic year of implementation.
- Make AI a strategy centerpiece
Generative AI is a major investment that requires CEO-level focus for successful implementation and scaling. Top-down prioritisation energizes the organisation, removes bottlenecks, and ensures proper sponsorship and funding. Recent insights show that 60% of banking executives now see gen AI as a top strategic priority, driving broader adoption.
- A centrally led organization is the key to scaling
Financial institutions with centrally led generative AI organisations see the most success. A review of 16 major banks found that over 50% have adopted this approach, crucial for building AI infrastructure and making key decisions. Centralization simplifies risk management and regulatory compliance, but balancing it with core business needs is essential.
- Sequence the gen AI roll-out across domains
Few institutions have yet benefited substantially from scaling generative AI. This is expected to change in the coming months. Focusing on 1-2 key areas rather than launching multiple pilots appears more effective. Financial services must expand AI applications across domains, from customer service and coding assistance to back-office operations and regulatory compliance.
- Robust (and reusable) scaffolding is crucial
Launching and scaling pilots without a robust enterprise AI infrastructure can increase risks and model management bottlenecks. Financial institutions need an extensible ‘knowledge scaffolding’ for effective generative AI deployment. This includes innovating products, enhancing productivity, and adapting to market dynamics. Key considerations involve operational models, use case prioritization, data governance, and regulatory frameworks. Scaling requires a versatile tech stack supporting workflows, machine learning operations, and diverse language models. The challenge lies in balancing local optimization with organization-wide applicability, ensuring the adoption of suitable technology solutions.
- Treat data as a corporate asset
Data is invaluable and must be treated as a strategic corporate asset to maximize its value, crucial not only for generative AI but for any digital transformation. Many institutions still struggle because they haven’t elevated data to this level. Proper data governance can either enable progress or hinder it significantly. Challenges include the quality of unstructured data, security classification of new sources, and data permissioning, as highlighted in discussions with banking executives focused on data and analytics.
- AI is a people play
The success of generative AI hinges on effective end-user adoption and change management. This requires investment in change management, reskilling, and impact measurement. Cultural shifts are essential, especially for long-time employees adapting to new technologies. With fewer than a third of organizations using AI in multiple functions, rigorous change management is needed to avoid ‘pilot purgatory’ and capture AI’s full value.
These are exciting times as major players rapidly embrace generative AI. AI capabilities are evolving quickly, now automating complex processes like writing credit risk memos. Assumptions about use cases today may not hold true tomorrow. One year in, much about gen AI remains to be discovered. It’s crucial to maintain momentum, broaden use cases, and manage this powerful tool with care.
Click to read more: https://www.mckinsey.com/industries/financial-services/our-insights/banking-matters/one-year-in-lessons-learned-in-scaling-up-generative-ai-for-financial-services