As I have written in other places, I believe that artificially intelligent entities may one day be granted legal rights in the United States. [Buy the book The Law of Artificial Intelligence and Smart Machines.]
But 2020 is way too early to consider such a thing.
Last week the U.S. Patent and Trademark Office made the same decision, rejecting a patent that named a machine called DABUS as the inventor. The Office wrote that interpreting patent inventors “to encompass machines would contradict the plain reading of the patent statutes that refer to persons and individuals.”
Of course this is right. Patent offices in the EU and the UK have already made the same decision. The statute says “person”, and a machine is not a person, even if the machine can perform some of the same functions as people can – like proposing practical solutions to longstanding problems.
Machines can also move faster that people, throw a ball harder, play chess better, and factor multi-digit numbers faster than people. These skills do not transform the machines into people. Koko the gorilla learned a working vocabulary of over 1000 hand signs and used the signals to converse with the people around her, but was never legally considered a person. Performance of a particular task is not the heart of the issue.
Some people will not agree with us on this matter, saying that if the machine actually proposes a solution that no human raised, then the machine has invented the technology. Reading the terms broadly, this is correct, but it doesn’t entitle the machine to be granted a patent.
I would suggest that, contrary to most people’s understanding of the matter, patent protection is much less about invention and much more about economics and the capability to exploit the technology. The point of a patent is not to recognize invention for the glorification of the inventor. The point is to allow an entity to develop and commercialize certain technology and for the rest of us to learn about the technology. Identifying inventorship is only the methodology we use to figure out who, if anybody, gains these valuable economic rights.
Article 1 of the U.S. Constitution provides Congress with the right “To promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries.” From the beginning of its existence in law, the U.S. patent is a creature of business and economics. As an article in the Social Research Quarterly published by Johns Hopkins University Press laid out the basic economic reasoning behind Federal patent protection:
A patent is a contract between the inventor and the public, by which the inventor, in consideration that the exclusive use of his invention is secured to him for a limited period of time, confers upon the public the knowledge of the invention during that period and an unrestricted right to use it after that period has expired . . . The specification is the in statement in which the terms of these mutual considerations and promises are declared, and on its completeness and accuracy depends the validity and the value of the contract itself (Robinson, 1890: 70) Referred to as the “patent bargain,” this arrangement is typically represented as a quid pro quo in which the inventor discloses in order to receive patent protection from the state (Fessenden, 1810: 49).
Such a monopoly is a boon to the state because the world learns about a new technology and someone is given incentive to bring that technology to market quickly in order to exploit the government-granted monopoly for the longest period of time.
No machine is going to be pushing to make the most of its technology. General AI simply does not exist. No machine can currently create an invention and then bring it to market, negotiating with buyers and licensees.
Further, the monopoly granted by the USPTO includes not only the right to exploit the invention but also the right to keep others from exploiting the invention. This means that a patent holder, to take full advantage of the opportunity, must make sophisticated decisions about which competitors to sue and who to sell licenses to. A patent holder needs more than complex decision-making, but the desire to take full advantage of the opportunity granted within the context of a changing marketplace.
I believe there will be a day when a machine will have both general societal coping skills and the drive to support itself in the modern economy, but until that day arrives, machines should not be classified as inventors.
Under current practice, not many inventors actually take their inventions to market. Most inventors are working on behalf of corporations and assign their patent rights to the company, which exploits the property on behalf of the shareholders. This practice may explain why a patent applicant asked for the USPTO to rule that machines may be classified as inventors. If a computer can be an inventor, then companies could simply push their AI onto inventing binges and build large portfolios of owned tech without needing to deal with (or pay) pesky people. I do not believe this invention model should be encouraged by the Federal government’s patent system.
Inventorship, as imagined by this country’s founders and as built into our patent laws, is complex and inextricably intertwined with economics. A machine may actually demonstrate a new and valuable system, but until that machine can show the drive to protect and/or exploit that technology, we cannot treat the machine as an inventor under the law. The motivations that are deeply ingrained in the human psyche are also necessary for economic decision making. Without those motivations, we should not be expected to officially classify an electronic problem solver as a person.