
AI sounding the death knell for copyright?
The tenor of my comment is directed neither toward discussing the validity of the concept of copyright, nor to its expression in law. Those topics are irrelevant to whether copyright is sustainable in practice.
Globally, so-called 'intellectual property' (IP) is awarded its status through international conventions. In particular, details of law protecting copyright vary among nations, but in essence support the same general aims. Nations particularly dependent upon copyright generated income actively promote convergence of national laws to a common standard, and through trade agreements demand evidence that copyright is enforced rigorously; in this context the USA is the major player.
Before supposed 'AI' stormed onto the scene, copyright holders were engaged in vigorous defence against widespread circumvention of their 'rights'. Commercial misappropriation of 'works' is tackled through law enforcement. However, unlawful 'competition' is facilitated by the substantial number of people willing to opt for the cheaper offerings. Likely of much greater importance is illicit distribution within the ethos of 'sharing'. The matter is bedevilled by difficulties for rights holders' in establishing plausible monetary losses. However, application in the USA of statutory damages staves off demands for universally agreed accounting of losses.
Copyright enforcement suffered a serious, perhaps ultimately fatal, blow when the 'digital era' arose. Disobedience to copyright law is rampant. It seems impossible to stem either through appeal to peoples' supposed 'better nature' or by enforcement mechanisms based on technology and civil/criminal law. Current efforts at enforcement have the look of fierce rearguard action.
'AI' has thrown a further spanner in the works. Setting aside much hyped claims for AI's capabilities, there remains the fact that 'large language models' enabling anyone so-minded to interact with AIs are acting as repositories of information. These AIs are not merely an analogue of books sitting on shelves. They offer services akin to those from a skilled librarian who also is a particular subject specialist. The oft reported nonsense output sometimes emanating from these computational resources can, in part, be attributed to imperfections of the underlying technology and to indiscriminate use of 'training' materials.
Take-up of these resources by ordinary people is occurring remarkably fast. This type of AI seems set to become a major contributor to education, to aspects of the work of various professionals offering services, and in the context of academic research. State legislatures, the national Houses of Congress, and courts in the USA may be able to hobble AIs or to turn them into cash cows for holders of rights. It's likely 'Western' nations, and others highly dependent upon trade with the USA, will step into line.
Looking globally, the future for copyright enforcement in AI is far less rosy. Nobody other than their own countrymen have a hope of preventing a free-for-all feeding of mankind's accumulated knowledge and broader culture into AIs. This will occur in universities, some other public institutions, and in commerce, with little chance of it being stemmed. Moreover, some of this will be shared globally on the Internet.
Copyright is a legal construct. One which many believe enforces a natural right to property ownership. Others think differently, else they are unwilling to engage in metaphysical argument. Regardless of that, if one nation goes against the grain, the copyright construct collapses globally. I posit that an initiative supporting open dissemination of information shall arise from BRICS nations when they flex economic muscle.
The foregoing leaves the interesting question of what IP-dependent economies could do in preparation for the inevitable, so that a seemingly untenable mode of business is replaced by other means fostering cultural and material prosperity. Perhaps, differing perspectives on the meaning of 'property' will become casus belli.