Can You Trust Your Large Language Model?

Explore the critical question of model trustworthiness in this conference talk by Jodie Burchell at NDC London 2025. Delve into the challenges of verifying machine learning algorithm outputs, particularly as "black box" models and Large Language Models (LLMs) become increasingly prevalent in everyday applications. Learn about the various ways models can produce misleading results and discover practical methods to evaluate model performance accurately despite these complexities. Gain valuable insights into how to navigate the confusion surrounding model outputs to determine their true reliability, an essential skill as AI systems continue to handle more complex and consequential tasks in our digital landscape.