Building Reliable AI for Financial Predictions

Lunch and Learn with Professor Bradford Levy

Modern AI models show impressive results on financial forecasting tasks — but many of these results fail in real-world use because models can unintentionally rely on information that would not have been available at prediction time. This session explores how large language models can be applied responsibly to financial prediction, with a focus on understanding and preventing lookahead bias. In conversation with second year student, Santiago Etchepare, from the AI student group, Prof. Levy will discuss why this issue is especially important with today’s frontier AI systems and what it takes to build forecasting models that are truly forward-looking, reliable, and suitable for real deployment.

Know what’s Happening Next — before everyone else does.

Get App

Host or Publisher Center for Applied Artificial Intelligence at Chicago Booth