1. The potential of AI to increase efficiency is obvious. So why are companies struggling with organizing how to deal with generative AI? Why are firms reluctant to simply let everyone experiment with it? Why does one top executive say "we don't want it to become 200 experiments in the company"? What is wrong with that? 2. On the other hand, what is the downside of centralizing experimentation with generative AI? Why is it nonetheless tempting to many companies? 3. What are some various ways of finding a happy medium between centralizing and decentralizing experimentation with tools like ChatGPT? What has VMware done, for instance? 4. Now think about this university! If you were consultants to the president and provost, how would you advise them to manage the use of ChatGPT? In particular, how would advise them to encourage innovation in teaching based on experimentation with generative AI