(An earlier version of this article by Arthur Boff was published in CIPA Journal, February 2016; this revised version, with thanks to Jim Boff for further suggestions, first published 28th August 2019.)
The recent attempt to achieve recognition for the Dabus AI system as an inventor is a confrontation many have seen coming. It is preceded by years of speculation about "computational creativity" and the question of whether a computer, suitably prepared to undertake investigations, could be seen as an inventor – see, for instance, Ryan Abbott's discussion of the question in Hal the Innovator.
In considering this question, it may be worth considering that inventive behaviour on the part of non-humans has been observed in the wild. Our primate cousins exhibit widely documented examples of tool use, as do other animals; certain species of crows and ravens are famous for it.
An interesting thought experiment is suggested by the famous "monkey selfie" case; that case related to copyright, but there are overlapping principles, in that the law holds that just as copyright ownership arises from the work of an identifiable author and their status (for instance, whether it was work done to hire), so too does patent law hold that a patent must be based on the work of an identifiable inventor, with ownership of the patent following on from that.
Let's suppose a naturalist in the wild is observing a chimpanzee. The chimpanzee devises a tool which, whilst simple, displays clearly novel and inventive features. Supposing the naturalist is well-informed about intellectual property, and decides to submit a patent application for the tool the chimp has devised.
Who should be named as the inventor? Naming the chimpanzee as the inventor would have the benefit of being true, but is not necessarily supported by the law, particularly in the wake of the monkey selfie case ruling against animal ownership of copyright. Naming the naturalist as the inventor would be incorrect. Naming nobody as the inventor is not, in general, an option; it used to be in the case of "patents of importation", but these have largely disappeared from the legal sphere.
For that matter, does the naturalist even have the right to make the application in the first place? On the one hand, the naturalist didn't come up with the concept themselves. On the other hand, we wouldn't have the benefit of the invention at all if the naturalist hadn't reported it in some fashion, and they surely deserve some reward for that. On one foot (which on a chimpanzee can be just as dextrous as a hand), the naturalist was making observations as a scientist, and we don't give patents for scientific discoveries. On the other foot, we give patents for inventive applications of scientific discoveries all the time (although since the Alice case confusion in the US over patentable subject matter has made this more difficult).
If the reader is sceptical about the possibility of a chimp coming up with a simple tool that humans have not come up with, let's replace the chimp with a New Caledonian crow, who devises tools for its own use. The physiological differences between crows and primates are obvious to human and crow alike, and tool-making crows will have put far more thought than many humans into how a tool could be adapted for use by beak and talon. It is entirely conceivable that a crow could invent a tool with features novel and inventive simply because of the unique challenges of the technical field the crow is working in.
Or if the idea of inventive animals feels like too much of a fantasy, let's take a left turn into science fiction. Let's imagine some aliens irresponsibly crash their spaceship in your back garden. First on the scene, you loot the starship and take the aliens' faster-than-light hyperdrive, dismantle it, discover the principles behind its operation, and file for a patent for it; you are aided in this task by the aliens' instruction manual, which bizarrely enough they happened to translate into English, so you undertake no particular creative effort here.
Who is the rightful owner of the patent in this case? The aliens have no particular status in terrestrial law; if an animal cannot own IP, and a computer can't, that suggests a general principle that only humans can own IP, after all. (Thomas Jerome Newton, the alien played by David Bowie in The Man Who Fell To Earth, only acquires his patents by masquerading as a human being – and it's not clear whether he's the genuine inventor there or merely cribbed some notes from one of his species' textbooks. Arguably, his trouble with the government arises from him deceiving the US Patent Office...) At the same time, it's hardly true to say that you have invented anything in this case, because you have undertaken no inventive activity in your own right.
These examples are whimsical, but there's a point to be made here. If a human being doesn't deserve a patent simply for observing and transcribing some unexpected inventive activity on the part of an animal, or on the part of an alien from another planet, why should a human being get a patent from observing and transcribing some unexpected inventive activity on the part of an AI system? This is a question which will only become more acute as AI systems with greater autonomy are developed.
More broadly, the concept of inventorship and identifying the inventor is given extreme importance at the early stages of the patent application process. Once the inventor has been identified, in more or less all cases their identity largely becomes irrelevant when it comes to questions of the validity of a patent; it is in questions of ownership that the identity and status of the inventor is of most importance.
Consequently, one can imagine an invention which is entirely novel and inventive, and by that measure is deserving of patent protection, but at the same time is incapable of receiving protection simply because no specific human or set of humans can be positively identified as inventing it. The current notion that the inventor of an invention arrived at by a computer is the person who made arrangements to set the computer to work on the problem in the first place is at best a stopgap measure, because it cannot account for inventions which arise as an unexpected consequence of an AI's activity – inventions which nobody specifically asked the AI to devise, but which the AI devised in the course of their work anyway.
Reaching a self-consistent and future-proof answer to the question of who owns the rights to an invention generated by a non-human entity may require consideration of the questions:
- Does inventorship require more than mere discovery?
- If inventorship requires more than mere discovery, is a person who first observes an invention made by a non-human entity an inventor?
- If the owner of an artificial intelligence is deemed the owner of an invention made by the artificial intelligence, is the owner of an animal deemed the owner of an invention made by the animal?
- Is there a material difference between inventions made by owned entities and by free entities?
- When assessing inventive step does one have to consider from the perspective only of unaided humans, humans using cybernetic assistance, or modified humans having cybernetic implants/interfaces? (Supposing a wonderful dementia cure is developed which entails providing artificial intelligence-based assistance to a patient's cognitive processes; would the patient then become unable to be an inventor, because it wouldn't be clear whether any specific idea they devised originated in their AI-based cognitive implant or in their biological grey matter?)
- Is the concept of inventorship outdated, save as a factor in determining relative rights in inventions?
Of course, if an animal or computer can be an inventor, could they also constitute members of the public for the purposes of prior art disclosure? Non-confidential disclosure to a crow may not seem like much of a risk to novelty, but some corvids can mimic and repeat human speech; computers may be imbued with a sort of creativity, but can they be taught about confidentiality? Ryan Abbot talks about "Hal the Innovator", but those who have watched 2001: A Space Odyssey also know that Hal is a lip-reader.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.