AI is here, for better or worse. While we debate and protest over the use of copyrighted materials to train large language models, there’s a second question that hasn’t yet gotten much public airtime: who owns what AI produces? Hartley’s deep dive into copyright law, the options for who might own AI-created content, and how to ensure that the massive economic value AI already creates is equitably distributed is fascinating: both as an intellectual exercise, and as a primer for conversations we’re guaranteed to have in the not-very-distant future.
There is, finally, a sixth candidate for ownership of outputs: the AI itself. What would it mean to find that the system itself owned the patents and copyrights in its creations? Current law in most jurisdictions holds, explicitly or implicitly, that only humans can be authors or own intellectual property, and current AIs demonstrably fail a number of important tests for counting as any kind of legal agent. Among other things, they cannot accumulate and spend money; they cannot own property; they have no citizenship, no domicile, and no civic rights or duties. On the one hand, there doesn’t seem to be any meaningful way of punishing them; on the other, like the homo sacer of Roman law, they receive no protection in law against any conceivable punishment.
Yet the day when we have to legally recognize AIs as agents in themselves may be nearer than we imagine. Philosopher David Chalmers assigns a greater than 25 percent credence to the idea that generative AI systems might reasonably be called conscious by the year 2032. As Chalmers says, we tend to believe that the quality of consciousness is ethically significant; it is, at a minimum, wrong to destroy any conscious being for no purpose, and our ethical obligations to conscious beings are in many cases very much greater than this. But being recognized as possessing rights tends to be a necessary but not sufficient condition for being recognized as a potential property owner: a crested macaque has certain rights, including the right not to be treated cruelly, but cannot hold property. AIs may come to be recognized as moral and legal agents before they are recognized—if they ever are—as authors.
