Robot Art: Who Owns Our AI-Designed Future?

In the film, Bicentennial Man, Andrew, an artificial intelligence (AI) android played by Robin Williams, is welcomed into the home of the Martin family. Though originally programmed as a live-in butler of sorts, Andrew becomes a clockmaker, fashioning elegant wooden timekeepers. When the ownership of Andrew’s unique clockmaking ability comes into question, the audience is presented with a moral conundrum. Should the clocks and any proceeds from their sale belong to Andrew — an artificial intelligence? Or do the clocks belong to Richard Martin, Andrew’s owner; or for that matter, NorthAm Robotics, which created Andrew in the first place?

For the viewer, it’s not a difficult choice to make. Despite his shiny metal exterior, Andrew is presented as possessing the complexity and agency of a typical human and therefore deserves the same ownership rights that any human being takes for granted. Andrew is eventually granted unofficial ownership rights by Richard, but it raises an interesting question when it comes to patents. Could the uniqueness of Andrew’s clock schematic also be legally patentable? Assuming Andrew’s clock schematic is subject-matter eligible, new, useful, and non-obvious (requirements for patenting), under current US law, neither Andrew, his owner, nor NorthAm would benefit from the ability to patent Andrew’s unique clockmaking approach. Patents can only be awarded to a human inventor. And since Andrew, an AI, developed his unique clock schematic, no one can be awarded a patent. In the AI-sympathetic future painted in Bicentennial Man, it’s easy to see this as unjust, but looking at the AIs of today, should patents continue to be only available to humans? In this paper, I examine some arguments on both sides of the issue, but ultimately favor that patents should be awarded for AI inventions.

Why the AIs of Today Argue For AI Creation Patents

Of course, today our artificial intelligence looks and thinks nothing like Andrew — lacking a complex personality that would make it easier for us to deem a set of algorithms human. But as more complex AIs autonomously develop potentially patentable processes, who stands to benefit from patenting those inventions? The contemporary AI certainly doesn’t possess any use for a patent, whereas the people or organizations that created that AI certainly would.

Another lens to examine AI-generated invention is to look at the AIs already walking among us — or better put, driving among us. Companies like Uber and Lyft have been testing AI driverless car programs over the last few years. While these AIs aren’t likely designed to sprout intuitive driving processes, they serve as a great example of liability when something goes wrong. Who is responsible when an AI driver makes a mistake and injures or kills someone? If Uber is liable for a negative expression of an AI, should they also benefit from a positive one such as an invention?

Though the one recorded driverless vehicle fatality was deemed to be the result of human negligence, the actual liability was unclear, as blame seemed to be shared by the driver, Uber’s lack of safety protocols, and the National Highway Traffic Safety Administration’s allowance of AI testing. Yet, how often in America do we hold a lack of regulations directly liable for negligence? Not too often. Many would argue that if an Uber AI caused an accident or a death, then Uber (and perhaps the AI developer) should bear some responsibility for negligence. If that is the case, it may seem appropriate that an AI owner should also bear the fruit of an AI’s innovation and be able to own the patent of its creation’s creation.

Let’s go down another rabbit hole and consider the copyright ownership of animal art. In 2014, nature photographer David Slater facilitated a macaque selfie at a nature reserve in Indonesia. Though Slater owned and set up the equipment, the photo was snapped by the monkey. While Slater provided the setup for the macaque’s big photo moment, because the monkey was the one to take the picture, Slater was unable to claim copyright. The United States Copyright Office deemed that because the photograph was the work of a non-human it cannot be copyrighted and therefore should be in the public domain. Slater argued that, “You could look at it like this, the monkey was my assistant. And, therefore, I was the artist behind the image and I allowed my assistant to press the button.” Unfortunately, the US Copyright Office didn’t agree that Slater’s monkey was merely a tool in creating his photograph. Score one for the monkeys, I suppose. Granted, not too many monkeys are auctioning off photographs for personal gain. But even if copyright could be issued to that monkey, it would be unable to grant usage rights to anyone, nullifying the usefulness of the product. What does putting the artistic work of a monkey into the public domain say about how the US values the work of a non-human species? It suggests that the artwork is a natural phenomenon, and one obviously cannot, and should not be able to claim intellectual property of the actions of nature.

Now imagine if some particularly intelligent — but not sentient — monkey had invented a new kind of camera. This, of course, would not have been possible without the help of a mad scientist who had trained this monkey and provided it equipment to experiment. Then one night while the scientist was sleeping, eureka! Amazing monkey photo technology! Had a human developed this system, it would be patentable, but because it was the monkey’s creation it cannot be. But the mad scientist might argue that without his training and equipment the monkey would never have been able to invent the new camera. Surely this isn’t a naturally occurring phenomenon and (ethics aside) the patent ought to go to the mad scientist, right?

Applying this framework to an invention of artificial intelligence, it becomes a more coherent argument. The key difference is that the AI is not a naturally occurring phenomenon or an animal — it’s human-created. Surely then, it seems more appropriate that the invention should be patentable by the AI’s owner.

Looking again at liability, if an AI were to create something it didn’t intend to create, say, a horrible virus, would the owner be liable for that creation? What if the AI created something it wasn’t programmed to create? Could that still be considered an intentional act by the owner? Surely, if an AI accidentally created a dangerous supervirus that infected someone, the AI’s owner would be liable for negligence. It would then seem appropriate to grant the benefit of a patent to the owner as well, even in the case of accidental invention.

Another argument in favor of granting AI creation patents to AI owners is that failing to do so would stifle innovation. At the moment, developing sophisticated AI requires considerable resources. If an organization couldn’t stand to benefit from the creations of an AI then they may not be willing to invest and spark innovation.

But what are some arguments against AI creation patents?

Why the AIs of Tomorrow Argue Against AI Creation Patents

Let’s return to our friend Andrew from Bicentennial Man. While an automated, intelligent human-like AI like Andrew may not be our present, if they become our future will we be prepared? Maybe an Andrew-bot is too far off, but is the possibility of creating an animal-like AI that absurd? We’re stepping into philosophical theory here, but despite human-creation, there must be a threshold of sentience where an AI ceases to be “code” and starts to be a life on its own. Whatever that point is, there must be a system of laws in place that considers it a naturally occurring phenomenon and protects it from exploitation by preventing a creation from being patentable at all. Surely, in far off science fiction Earth, if a sentient AI possessed the ability to consent to use of its patent, and to directly benefit from its creation then it should be awarded its own patent. While this may not be an argument against AI creation patents in the present, it could be one in the future and must be considered. What precedent is being set if AI creation patents are allowed to be granted now without the ability to consent to an owner agreement?

What if an AI is designed to develop AI offspring that then goes on to invent. Just because you’re six degrees away from Kevin Bacon does not make you Kevin Bacon. That is to say that just because someone invents an AI there ought to be a certain point at which the inventor can’t claim it’s their creation. A noisy neighbor can’t claim ownership of another’s innovative invention made to block out neighbor noise, despite their “influence” on the project. Further, if patents were awarded in all cases to a human AI developer, and the successively more advanced AIs invented new ideas at an accelerated pace, a collection of AIs owned by one person could monopolize different sectors, making competition impossible.

For that matter, what if the AI is designed to be able to modify their code? Just as Andrew’s code sparked the ability to become a clockmaker, could NorthAm Robotics really claim they had a hand in its development? It could be argued that once Andrew left the factory floor and the artificial laboratory setting that it ceased to be “training” designed by NorthAm. Andrew’s day-to-day input of naturally occurring data could suggest that regardless of the unique abilities NorthAm programmed him with, his unique, naturally-occurring experience is what prompted his specific evolution to create clocks. Certainly, NorthAm could not claim ownership of Andrew’s unique place in the universe.

What happens when AIs become so complexly intelligent that the mere input a person gives them is just a command like, “invent a new type of hat that teaches me how to play guitar.” Should that novel hat be patentable by the person who developed the AIs code or the person who provided the specific input? Could that really be claimed by a human inventor? The specific input could not be claimed by the AI’s inventor and it does not seem to be in the spirit of the invention when its impetus is such an obvious, general suggestion. Keeping in mind that in order to be awarded a patent in the first place an invention must be non-obvious. Surely, that is a strong argument against the AI’s hat invention being patentable at all.

What Approach Makes Sense for Today?

Despite the valuable arguments for the rights of AIs with agency in the future, we have not crossed that threshold. The AIs of today are not the AIs of tomorrow, and until we get to that point, individuals and organizations should be able to fully benefit from the work and financial investment put into developing AIs.

While there could very well come a point at which AIs possess the ability to consent, they do not at this time. To hold the place for future consenting AIs by not listing them as inventors today may constrain the innovation necessary for creating them in the first place and certainly denies them the potential to benefit from their own creation in the future. It does seem appropriate today that AI owners should also own AI inventor listed patents with the caveat that such automatic ownership may one day need to be reviewed as AIs reach sentience.

Works Referenced

Chen, A. (2020, April 2). Can an AI be an inventor? Not yet. Retrieved from https://www.technologyreview.com/2020/01/08/102298/ai-inventor-patent-dabus-intellectual-property-uk-european-patent-office-law/

Free, R. (n.d.). Artificial Intelligence — Questions of ownership. Retrieved from https://cms.law/en/int/publication/artificial-intelligence-questions-of-ownership

Hawkins, A. J. (2019, November 20). The world’s first robot car death was the result of human error — and it can happen again. Retrieved from https://www.theverge.com/2019/11/20/20973971/uber-self-driving-car-crash-investigation-human-error-results

Kelion, L. (2019, August 1). AI system ‘should be recognised as inventor’. Retrieved from https://www.bbc.com/news/technology-49191645

Mendonca, S. (2014, August 7). British photographer in Wikipedia monkey selfie row. Retrieved from https://www.bbc.com/news/av/uk-28684353/british-photographer-in-wikipedia-monkey-selfie-row

Porter, J. (2020, April 29). US patent office rules that artificial intelligence cannot be a legal inventor. Retrieved from https://www.theverge.com/2020/4/29/21241251/artificial-intelligence-inventor-united-states-patent-trademark-office-intellectual-property

Porter, J. (2020, April 29). US patent office rules that artificial intelligence cannot be a legal inventor. Retrieved from https://www.theverge.com/2020/4/29/21241251/artificial-intelligence-inventor-united-states-patent-trademark-office-intellectual-property

Smith, H. (2019, December 9). Can monkey who took grinning self-portrait claim copyright? Retrieved from https://metro.co.uk/2011/07/14/can-monkey-who-took-grinning-self-portrait-claim-copyright-77773/

Stephens, K. (2019, December). Who Owns an AI-generated Invention? Retrieved from https://www.twobirds.com/en/news/articles/2019/global/who-owns-an-ai-generated-invention

US government: Monkey selfies ineligible for copyright. (2014, August 22). Retrieved from https://www.csmonitor.com/Technology/Tech-Culture/2014/0822/US-government-Monkey-selfies-ineligible-for-copyright

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Elliot Trotter

Elliot Trotter

Content Designer, UX Writer | Microsoft | Master of Communication in Digital Media