You can view it (if viewing is offered for free, but meta famously used a bunch of PDFs where that is not true), you can’t use it to make a profit without a proper licensing agreement. Which is the essence of copyright law.
If I pay someone to create an image in a style of another artist, and they look at that artists work, have I stolen that artwork?
That is not the situation at hand. We are distinctly not hiring a human being, with a life interests and opinions separate from our own, and a history of art skills, to spend time studying a new piece of art to learn a new skillset, and then apply that skillset to create an original piece of art inspired by the existing art, and compensate them fairly for their efforts.
No, the situation instead is: taking one product without licensing it to build a new product, and then selling access to the new product.
For the two situations to be equivalent, you would have to convince me:
That the AI is conscious
That this new consciousness is being treated ethically and not forced into involuntary servitude.
That it has its own interests and opinions distinct from the corporation who created it
That it is being compensated fairly for its work fairly.
Not only is this not the case, the corporations making these AIs deliberately don’t want to make this the case, because then they can’t have what they want: all the inginuity of a human, with none of the obligation to treat them like a human.
No, we need to be clear about what this is: it’s a machine designed to generate an image stochastically based exclusively on a bunch of inputs. Which is a great invention, it’s just that when they did it, none of the inputs they used were properly licensed.
It’s definitely a difficult problem to know how to regulate, but the fact is we have not created consciousness, nor do they want to. So imo this is most akin to using an unlicensed copy of Photoshop to create an original piece of artwork, and then selling that artwork. If Adobe can show you did that, they will own everything you sold. But they can do that, because they can afford the lawyers, and a random deviantart user can’t.
You can view it (if viewing is offered for free, but meta famously used a bunch of PDFs where that is not true), you can’t use it to make a profit without a proper licensing agreement. Which is the essence of copyright law.
That is not the situation at hand. We are distinctly not hiring a human being, with a life interests and opinions separate from our own, and a history of art skills, to spend time studying a new piece of art to learn a new skillset, and then apply that skillset to create an original piece of art inspired by the existing art, and compensate them fairly for their efforts.
No, the situation instead is: taking one product without licensing it to build a new product, and then selling access to the new product.
For the two situations to be equivalent, you would have to convince me:
Not only is this not the case, the corporations making these AIs deliberately don’t want to make this the case, because then they can’t have what they want: all the inginuity of a human, with none of the obligation to treat them like a human.
No, we need to be clear about what this is: it’s a machine designed to generate an image stochastically based exclusively on a bunch of inputs. Which is a great invention, it’s just that when they did it, none of the inputs they used were properly licensed.
It’s definitely a difficult problem to know how to regulate, but the fact is we have not created consciousness, nor do they want to. So imo this is most akin to using an unlicensed copy of Photoshop to create an original piece of artwork, and then selling that artwork. If Adobe can show you did that, they will own everything you sold. But they can do that, because they can afford the lawyers, and a random deviantart user can’t.