This is true. I think most people are mostly using AI at work to fix bugs in existing codebases. A smaller group of people are benchmarking AI by giving it ideas for apps that no one needs and seeing if it can get close. The smallest group of people is actually designing new software and asking the AI to iterate on it.