As widely reported in the media, judges in the UK Supreme Court have ruled that AI cannot be granted a patent. The same decision was handed down on the other side of the Atlantic by the US Supreme Court. Harvard Law professor Lawrence Lessig and other academics supported the plaintiff in a brief in the US, saying that the Federal Circuit's decision “jeopardizes billions (of dollars) in current and future investments, threatens U.S. competitiveness and reaches a result at odds with the plain language of the Patent Act.” I was also reminded of the old fracas about whether monkeys can own copyrights to their photos, where the courts eventually ruled that monkeys do not have a standing in the court of law and therefore cannot sue for rights. While it may look like we are taking a very anthropomorphic perspective, I thought the implications went deeper than simple anthropomorphism.
I was intrigued about what this all could mean today and how it could shape the future. The more I thought about it, it became clear (at least to me) that there were more questions than answers. These questions and thoughts eventually coagulated around the ideas below:
- Data—Data are the fuel for AI. In the case of patents, there must be some place where the data, which are the basis of the AI, originated from, not to mention the patent as well. In the case of a patent and rights and benefits accruing from such patents, do the sources of data have rights? Will human data subjects benefit from the patent? Even if it is assumed that synthetic data is used, the metadata for this synthetic data originated from humans—what happens then? If AI has the right to the patent from human or synthetic data, do secondary or even tertiary rights accrue to humans’ data sources?
- Rights—Rights are going to be slipperiest of the slopes when it comes to AI. In some parallel universe, if we assume for a moment that this patent was granted, I think it would open Pandora’s box. While I am no expert, in legal circles, all precedents are watched mawkishly. In this case, granting a patent to AI would be tantamount to recognizing a right of that AI. However loosely this right is interpreted, it would at least prop open the door for other such rights to accrue, which could lead to its own rabbit hole and consequences.
- The future—Utopia or dystopia? Finally, the sci-fi enthusiast in me thinks this could be beginning of those dystopias that grip us in so many fictional settings—think Ready Player One or something like that. Once rights accrue, I am sure benefits will accrue as well, which leads us to the paper clip conundrum that Nick Bostrom propounded as early as in 2003. While the discussion is ongoing about how this rise of AI will end and whether we will end up with a benevolent or aggressive artificial general intelligence (AGI) in the long term, in the short term, the potential for concentration of power in the hands of a few is very plausible. Mustafa Suleyman in his book, The Coming Wave, argues for early containment irrespective of the form this containment takes, which could be one way of looking at the problem. For now, all we have are opportunities and pathways to explore.
At a more mundane level, the implications of the above ruling could be far-reaching. For example, social media is rife with “avatars,” or other forms that represent humans or even operate as AI, that rake in the money. If patents are granted to AI, and if this can be interpreted as rights being allotted to AI, then can this be extended to other forms of AI, leading to AI having rights? Which begs the additional question: will AI have a right to life or its digital equivalent? Will this be a slippery slope with no end in sight? Can we get to a situation where we are either denying AI its (their?) rights or even denying the people who are behind these AI some hereto unknown, unthought of or unperceived rights?
While it may look like we have more questions than answers, from the time I read Warren Berger’s A More Beautiful Question, I have come to believe that questions are the way to start. Questions are the only way we will get to answers, and in the case of AI, the more questions we ask, the better we will be able to deal with the future. It is going to be us and them, and not us versus them.
