Anticipating AI Failures
Daniel and Patrick discuss the importance of anticipating AI failures and the simple steps that can be taken to test the sensitivity and robustness of models. They also highlight the need to track which models operate on which user data to avoid potential issues. Patrick emphasizes the responsibility of data scientists to study AI incidents and learn from them, comparing it to the aviation industry's approach to studying airplane crashes.In this clip
From this podcast
Practical AI
When AI goes wrong
Related Questions