I think its valuable for mathematicians to think through how AI changes things for them. I think it's a bit premature to know exactly the impact AI will have. Math is one field that I feel will just naturally sort itself out without trying to predict or prescribe how it should work, it's a bit like software development, you adopt it in and see where it takes you. The presentation then sort of tries to argue for human understanding because AI might optimize for the wrong thing. That may well be what we need for the immediate future, but it's hard to know how things will play out. It might be such that it will become more important to people that AI understands something. ie, Does AI, with access to all current medical knowledge, think this cancer drug will be effective and safe? or has only humans said it's ok? At the moment we are in the chaos of change and its going to take a while for things to settle.