I suspect we are getting less from AI than we could because we are not bold enough with what we ask it to do.
We bring the internal brakes we developed during our careers: this takes too long, that needs a whole team, I need to learn another tool first. Those judgments may have been reasonable when we learned them. They deserve another look now.
Even knowing a good solution can get in the way. In a study of expert chess players, a familiar solution kept drawing their attention while they were trying to find a better one. That study was about chess. I think the habit is worth watching for in our work with AI too.
Try starting each task anew, from first principles. What do you want to achieve? Which constraints are real? Which steps are there because you have always done them?
For example, suppose you need to compare two ways of presenting some data. You might normally choose one before building anything because implementing both feels too expensive. With AI, try asking for two small working versions and compare them. Check whether the cost you remember still applies.
The approach you memorized may still work. It may also stop you from looking for something better as the tools change. Keep it available, and leave room to test another way.
Think like a child: curious enough to ask why, and willing to try something before deciding it cannot work. Then use your experience to check the result.
Not knowing what to not ask is not a bad thing.