I agree with Mike Masnick @mmasnick: copyright law is the wrong way to deal with the challenges posed by AI. Copyright law should properly only address the reproduction of content, such as writing. Not the ideas that underlie the content. Putting something in your own words has always been allowed, and always should be allowed whether the agent is human or AI.
So long as AI does that, and safeguards are taken to make sure original wording is not merely spit back out verbatim, the fact that AI is trained on copyright material is not itself a violation in the same way our own bio-brains have been trained on copyright material.
That being said, there is the idea of compensation. Shouldn't the owners of AI systems compensate the content creators? To which I would answer, yes, absolutely. And I would add that all of us, all members of the human race, have been collectively training AI without any compensation whatsoever. And that's unfair.
Which is one of the important reasons why I support UBI.
