Artificial
is learning copywriting still worth it when AI output is already client-passable?
Been using a few AI tools to help with copy for a small side project and it's raised a question I can't quite shake. The output is genuinely decent now. Not great, but decent enough that clients who aren't paying close attention probably wouldn't notice the difference. The thing is, I've been spending real time learning copywriting. Reading books, studying good ads, practicing hooks. And part of me wonders if that investment still makes sense the way it did two or three years ago. The counte...
Read the full discussion on Reddit
This article was aggregated from Reddit. Click to join the conversation.
View on Reddit