Michael Webster, principal engineer at CircleCI, talks to Rob about testing AI-enabled applications. In this episode, learn how to face the unique challenges posed by the probabilistic and non-deterministic nature of AI output, as well as the importance of subjective evaluation criteria.
Webster covers how model graded evals can be used to test AI applications, and the importance of caution in using this approach.
CircleCI gives AI/ML teams the tools they need to iterate quickly, deploy safely, and deliver value continuously. To learn more, visit: circleci.com/ai-ml/
Have someone you’d like to hear on the podcast? Reach out to us on Twitter/X at @CircleCI!
The Confident Commit