Artificial intelligence (AI) and intellectual property (IP) have been in the limelight for a while now and have caused quite a stir, especially since the number of inventions stemming from AI “teaming-up” with researchers rapidly increased in the last decade. The crux of the matter of this ascendent trend is whether AI can own IP rights, issue that has started to divide legal and tech experts.

The spark that lighted this debate is DABUS (Device for the Autonomous Bootstrapping of Unified Sentience), an AI system created by Dr. Stephen Thaler that can function independently and create products on its own, for which Dr. Thaler has sought patents in 17 jurisdictions. Up until till this point, there is nothing out of the ordinary with filing for a patent, but Thaler filed national and international patent applications naming DABUS as the inventor. The applications challenged the traditional IP practice and raised the question whether only humans can be named inventors in a patent filing.

Can AI be a patent owner?

Traditionally, patent law deems that a patent can be awarded to any person that invents a product or a process that brings about a new way of doing something or a new technical solution to a problem. DABUS definitely brought something new to the world, and from all the 17 patent applications, so far, only South Africa and Australia have issued patents naming DABUS and the inventor. Meanwhile, UK and US rejected the application, on the grounds that AI machine cannot be deemed an inventor, because the inventor must be an individual/natural person. To that extent, with the Court of Appeal in England and Wales deeming thus that only a person can be an inventor, Dr. Thaler acknowledges that since he is not the inventor, he has no claims over the patent.

AI-created work and ownership

So, one of the questions addressed by IP specialists is whether AI can be a creator and whether it is capable to own what it creates. A line of argument would be that these AI technologies are autonomous prior to being “trained” and programmed to associate certain data with outputs. Of course, AI is groundbreaking and is revolutionizing technology as we knew it, but is it really that bright?

For example, an AI system, once “trained” can associate words with descriptions such as their grammatical indicator, be it a verb, noun or adjective and can further learn how to create sentences. It gets interesting especially since after the programmer writes the code through which it is given to the system the necessary tools to form sentences, the programmer does not really know how the AI system will coordinate the data and what sort of – comprehensible – sentences it will create. Then, once created, who really own these sentences?

To delve further into the ownership of the AI-created work, a sensible question to firstly ask is who is the author of that creation? Following a logical streamline, AI generated ownership of the work should lie with the person(s) that created the given algorithm of the program that produced the work – if we look at AI as involving machines/systems that act outside the human control. Conventionally, the author is considered to be the person who creates the work – of course, unless the person is an employee and according to the agreement with the employer, the copyright falls under the latter`s ownership.

Another issue is that AI is not considered a legal entity like companies or corporations, that have legal personality (meaning that the entity is formed and administered under corporate law). And, in the event that the company might commit fraud or other types of misconduct, the competent courts will hold accountable the individuals behind the company`s structure, however, if we follow the same reasoning line in the case of AI, then the software developer or the user accountable for the given infringement cannot really be hold accountable for a non-person.

At the beginning of December 2021, Google`s London-based researchers from the AI powerhouse DeepMind alongside a handful of mathematicians announced that AI has been used for the first time to prove new mathematical theorems showing that AI can be applied to tasks as abstract as pure mathematics or as challenging as protein generation. Some would go and say that AI would be nothing without the human contribution, AI being nothing more than a mere tool used by tech developers and scientists to solve specific problems, based on algorithms subject to the programmer`s know-how. But is that so? Another DeepMind`s AI, AlphaGo Zero, using a neural network, goes on and proves the opposite, teaching itself the complex game of Go becoming the first computer program that defeated a professional human player.

Conclusion

Even though DABUS had some wins, it seems that the whole inventorship debate regarding AI-generated work will continue to raise questions, divide opinions and push for the harmonisation of the legal framework with the technological developments. Meanwhile, UK has already started a public consultation on how the copyright and patent systems should deal and accommodate AI and the extent to which IP protection could be extended to the creations/inventions of AI systems. In the meanwhile, the European Union, US and other organizations such as WIPO (World Intellectual Property Organization) are also performing studies to find the right solutions for the IP disputes AI-related.  And, for that purpose, the European Commission alongside other interest groups started to take more and more into account the option of creating an ex novo legal framework, in the wake of the Fourth Industrial Revolution.