AI in Policy with Ellie Graeden from Luminos Law

Data Integration

Ellie Graeden from Luminos Law highlights the critical challenges in data integration for AI applications. She emphasizes the need for patient-level, population-level, and facilities-level data to address complex issues like hospital power outages and patient evacuations 1. Graeden explains that understanding and communicating the iterative nature of data models is crucial for effective integration.

The biggest challenges are in Ize, I really find the biggest challenge is to be at the data level, really deeply understanding what your data are, where they come from, what training data you're using, whether you believe those data in modeling systems and understanding and communicating to others the degree to which these are iterative processes.

--- Ellie Graeden

She also shares examples of large-scale knowledge graphs and databases that integrate diverse data sources, such as wildlife disease data and human epidemiological modeling, to answer complex questions 2.

   

Project Management

Managing AI projects requires a clear definition of requirements and the right analytical tools, according to Graeden. She points out that many projects fail because they start with a tool in mind rather than a clear objective 3. Graeden also discusses the importance of governance and rapid decision-making in data-driven environments.

The tendency tends to be, well, I either really want this big fancy tool that I want to be able to say I'm using an AI model. I guess we'll figure out what we're going to do with it and what the product is after the fact.

--- Ellie Graeden

She shares her experiences working with organizations like the Army Corps of Engineers, emphasizing the need for effective translation of data into actionable decisions 4.