It may be smart, but it’s not that clever. Artificial intelligence is nothing without human input. The algorithms that drive AI rely on the expertise of programmers and it’s still no more than a tool – albeit a powerful one – that scientists and engineers can use to solve problems.
Yet this is not to say that AI isn’t the fastest-growing deep technology in the world, with the potential to transform people’s lives and boost nations’ economies. Facilitating AI innovation has even become a priority for the UK government, as laid out in the National AI Strategy it published in September.
The UK Intellectual Property Office (IPO) recognises the importance of AI to such an extent that it has just started an open consultation seeking views on matters such as whether AI-generated inventions should be protected and, if so, how. The current legal system offers no protection, because it recognises only humans as owners of intellectual property.
The IPO says: “We need to ensure that the law keeps up and is appropriate to incentivising creativity and innovation.”
So, in cases where an AI system creates something new, surely the people who programmed that system should be acknowledged for the innovation and its contribution to the public good? Or should AI alone be credited as the inventor of ideas that forge a new future?
This is the crux of a debate that’s raging in the tech and legal professions: can AI own intellectual property? And, if it can’t, could innovation be stifled?
Test cases based on patent applications for two hypothetical AI inventions have become the focus of these arguments. Dr Stephen Thaler, a US scientist who has created an AI system called Device for the Autonomous Bootstrapping of Unified Sentience (Dabus), has been putting the laws of numerous countries to the test with the support of Ryan Abbott, professor of law and health sciences at the University of Surrey. They are advocating not for an AI system to be granted its own patents, but for its owners to be granted a patent on any invention it may generate.
Thaler’s patent applications have credited Dabus as the sole inventor of “an improved beverage container and a ‘neural flame’ device used in search-and-rescue missions”. The applications have been rejected in both the UK and the US on the basis that only a person, and not a machine, can own patent rights. Leave to appeal has been sought in the UK Supreme Court.
The IPO has welcomed the judgment and its clarification of UK law as it stands. It says that it doesn’t envisage awarding such rights to an AI system, but it does foresee granting them to “the people and firms that use, and invest in, AI technology”.
Judges in both Australia and South Africa, on the other hand, have ruled that AI can be listed as the rights owner in patent applications.
How should we think about these judgments and what are the consequences? Opinions among leading patent attorneys are divided.
“In the ‘real world’ beyond this case, I expect we’ll see a practice developing whereby patent applicants with a genuine interest in protecting AI-generated inventions will avoid naming the AI as an inventor,” says Pete Sadler, partner at patent attorney practice Reddie & Grose.
But then the courts may have to decide the link between the people credited with devising the inventive concept and the role played by their technology.
Dr Alexander Korenberg is a partner with patent attorney practice Kilburn & Strode. A former computational neuroscientist, he points out that the current debate is based on a hypothetical premise, because it has yet to be established that a machine can actually create inventions autonomously.
“Anyone who’s au fait with the technology knows that this would require something like artificial general intelligence [human-like AI], which is thought to be at least a few decades away,” he says. “It’s not even certain that this will ever be a possibility.”
The argument that Korenberg and others propound is that AI is still no more than a tool without sentience, lacking the ability to think independently and spontaneously invent something.
But let’s assume that AI systems will come up with new things in such a way that makes it impossible to identify the human intellectual input in the final invention, if this is not already happening. In such a case, to ensure the continued development of this valuable economic engine, British IP law would need to be updated to deter inventors from moving to a more sympathetic jurisdiction abroad.
Robert Jehan, partner and patent attorney at Williams Powell, whose clients include Thaler and Abbott with their Dabus applications, believes that there will no longer be a problem with patenting AI-generated inventions in the UK, because the government now seems more inclined to help.
“There appears to be a general willingness to accommodate the patenting of inventions created by AI systems. The potential rewards for doing so and the losses for failing are simply too great to ignore,” he says. “Any country that refuses to protect AI-generated inventions, whether these are created solely by the technology or in conjunction with one or more natural persons, will stand to lose vast amounts of investment in their AI industries. We’ve come to a situation where governments are actively seeking to help.”
Indeed, the outcomes of Westminster’s National AI Strategy could directly affect how AI inventors and rights-holders are viewed in the UK.
If it does become necessary to reform patent law, if only to cover future advances in AI, could precedents in copyright law offer guidance? Richard Johnson, a patent attorney and partner at Mewburn Ellis, notes that the Chartered Institute of Patent Attorneys has been encouraging experts to seek a solution down this route.
“Copyright law already has provisions recognising that a computer could be involved in the creation of a work that would attract copyright protection,” he says. “These provisions give ownership of that copyright to the people who caused the computer to operate to create the relevant work.”