This feels right because fewer parts of AI as it's used in apps is going to fit the agile CI/CD model of quality control. It's back to manual functional testing

This feels right because fewer parts of AI as it's used in apps is going to fit the agile CI/CD model of quality control. It's back to manual functional testing for a lot of it. Also, the way that issues get reported and described will probably benefit from AI specific expertise.