I think the current generation of AI-based art/design tools like DALL-E are going to have their use, but more as a supplemental support for actual human designe

I think the current generation of AI-based art/design tools like DALL-E are going to have their use, but more as a supplemental support for actual human designers and artists. It would really depend on the application, but for design especially I think the big limitation being run into is that the tools can only be trained on content that is currently available. All of these generated designs look extremely generic down to the Corporate Memphis/Alegria type artwork which is trendy right now. Outside of unintended behavior there isn't much of a way to push the envelope or generate a "new" style without the eye of a human. As tools like this gain more widespread use, it will also become increasingly important to discriminate heavily on what content makes it into the datasets used for these tools. AI-based tools trained on AI-generated content sounds like a recipe for stagnation.