The quality of an AI solution is frequently thought of in terms of its predictive performance. If we can get good, generalizable, results on validation and test datasets, then we have satisfied the requirements for deploying an AI system, right? Not necessarily. Recently, the bar for AI solutions.

Read more >