The Importance of Explainability in Deep Learning

I imagine when you're talking to enterprises that are considering these types of approaches, there are concerns like explainability and even robustness. Explainability will always be a concern at some level when using deep learning. But typically, we don't operate with customers for whom explainability would be extremely strong. We haven't seen this with the type of problems we work on because it's really about logistics and it's about optimizing well-known constraints.