There’s a difference between using a tool as a knowledgeable professional, and using a tool as a novice or a member of the general public: the professional can likely tell when the answer an AI gives is wrong.
AI is a tool that can be useful sometimes.
Controversial opinion: If there is no real child whose likeness is used to produce the image, then there is no real victim. The art that is created is of the same quality as a drawing from imagination.
The other problem with illegalizing content on the basis of whether it’s CSAM is the eternal question of whether an unreal likeness can be said to even have an age. How would you be able to prove with certainty that an AI-generated likeness was below the age of consent? You would think it would be obvious, but there is a huge gray area here. If the protection weren’t fairly strong, I 100% guarantee that conservative Bible-thumpers would take advantage of the opportunity to ban all ai pornographic content, and then move on to real porn after that.
In addition, Banning AI material we might be able to interpret as CSAM does nothing to protect real children. There is no empirical proof that any kind of porn has a real, independent effect on people’s sexual behavior.
Edit with addendum: In addition, civil rights are a matter of principle. It doesn’t matter whether the presence of AI generated material makes the job of policing more difficult. The Bill of Rights was created to try to make sure that we don’t become slaves. Just because something is highly offensive to my own tastes doesn’t mean that the civil rights of the person who offends me disappear.
I can stand AI as long as the code is reviewed line-by-line by a human. The only problem I have aside from that is that I wonder if code generated by an AI can validly be placed under an open source licence instead of being public domain. Didn’t a court rule that you couldn’t copyright generated code?