Best Practices for AI Data Modeling

Produced by Steve Nathans-Kelly

Data is the soul of any algorithm, so investing time and resources can reap benefits in the future. Morgan Stanley AVP Supreet Kaur discussed best practices for AI data modeling during Data Summit 2023.

“It's important to understand that nothing will be possible without the data,” Kaur said. “So earlier it was all about availability of data. Now companies are also thinking about do we have enough data?”

It’s key to create a data quality framework that can determine what data is more important when sorting through a variety of information.

“Data governance models are also needed to have responsible AI,” Kaur said.

Before moving forward, organizations need to ask themselves, “do you have the right technology, right people, the right skillset to be able to deal with this extremely valuable customer data?”

Choosing a suitable model will determine whether AI is a good fit for the data project. There are variable measures to determine how the AI model is doing compared to legacy systems.

“Because if you launch something, it should do better than the other systems. So there has to be that benchmarking and bias versus variance,” Kaur said.

Save the Date for Data Summit 2024—May 8-9, 2024—Hyatt Regency|Boston, MA!