Comment by saberience
5 days ago
But it does matter to you, that's precisely my point, we have already established that you don't want to be deceived.
If someone gives you a book by "JRR Tolkien" and it's written by an AI, you're going to be pissed off. If your "girlfriend" sends you a note and it's written by an AI, you're going to be pissed off.
If someone labels a book by "Jake Donaldson from New York" and it's ChatGPT, you're going to be pissed off.
People don't like to be deceived or pay for something and get something different from the thing they thought they were buying.
I don't want to buy something that an AI wrote, the only time I want to read AI writing, is if I directly ask it a question and get an answer.
The problem you are citing isn't "created by AI", the problem you are citing is "misattribution writ large".
If we had the capacity to legally enforce that things created by AI be labeled as such, then why stop at that terrible proxy for the real problem instead of just using that same legal infrastructure to outlaw all misattribution full stop?
I don't want to be legally required to put a bell on every single technological output that might in some circumstances make crimes easier to commit than doing it longhand was, and I am old enough to recall making the same case in the late nineties when folks were clutching their pearls over Photoshop's ability to enable counterfeit nudes.