What’s been your (or your colleagues’) approach to building strong business cases for AI investments, including addressing potential risks and uncertainties?

We are pursuing a multi-pronged AI strategy focused on both immediate value and long-term scalability:

1. Targeted Enablement Across Core Platforms
We are proactively reviewing and enabling AI capabilities within our core platforms—EMR, ERP, non-clinical tools (primarily Microsoft), and ITIL systems. The cost-benefit ratio is typically favorable for EMR and ERP platforms. For other areas, we are conducting deeper evaluations before broader adoption.

2. Proofs of Concept in the Data Lake Environment
We are executing a series of AI proof-of-concepts (PoCs) within our Databricks lakehouse environment. These are selected based on preliminary pro forma analyses. Costs are tightly managed, and the success of even one PoC at scale can offset the investment in the others.

3. Portfolio Governance and Financial Discipline
Every AI initiative is supported by a pro forma and tracked for performance. This allows us to identify early initiatives with negative 3 or 5-year NPV projections and adjust accordingly.

We recognize that AI is a greenfield space. Some PoCs will not succeed—and that’s expected. What’s critical is that we learn, iterate, and mature through these experiences. We do not see avoiding experimentation as a viable strategy.

The key is doing what's right for the company quickly and efficiently. We hold weekly sessions, akin to a "Shark Tank," where the business can present vendors or ideas for evaluation. This process occurs before procurement to avoid wasting time on unsuitable AI products. It's a model that helps us quickly assess and respond to business needs, preventing departments from independently bringing in AI tools without oversight. This approach allows us to embrace AI wisely and effectively.

Our approach begins with understanding the current cost of business and ownership of applications from both IT and business perspectives. We define the problem we're trying to solve, identify success factors, and determine if AI is necessary. If the same results can be achieved without AI, there's no need to invest in it. We evaluate potential outcomes, costs, and what we might replace to avoid unnecessary expenses. It's crucial to explore the competitive market and not rely on a single vendor. We assess internal capacity, skill sets, and training needs, considering whether external resources are required. All these factors form the basis of a comprehensive business case, ensuring that AI is not seen as the only solution.

For us, building a business case for AI investments hinges on understanding the strategic importance of AI to our organization. At the end of last year, I was able to quantify a potential investment figure to present to leadership. This allowed us to discuss whether we wanted to proceed with AI at scale. While our organization isn't massive, we aim to be an AI thought leader, using AI in our academic missions and other areas. It's less about ROI and more about strategic use and managing costs realistically. Everyone acknowledges the risks involved, whether ethical or compliance-related, and we're committed to proceeding safely within the frameworks set by our institution. The landscape is constantly changing, and we must adapt accordingly, ensuring our fundamentals in security and execution are sound.