By, Kunal Bhardwaj and Rachit Taparia

 

What is AI?

AI is loosely defined as any technology that undertakes to accomplish an activity which would normally require human intelligence. It is capable of-

(i) comprehending unstructured data

(ii) producing computer reasoning

(iii) self-learning

Hence, it produces unexpected and unconventional results which are independent from the programmer’s instructions.

It’s well-settled that AI related inventions are patentable. However, what happens when AI creates an invention of its own, without any human intervention or a human engagement which is insufficient to recognize a human inventor. This question remains largely unanswered in patent legislations across the world. The reason for this legislative gap may be due to the non-consideration of such a futuristic technological possibility at the time of drafting the patent statute.

Who is an inventor?

US courts have held that an “inventor” is the party which conceives an invention, and only he reserves the right to obtain a patent on this invention. In Goldstein v California,[1] US Supreme Court held that authors and inventors should be given a broad interpretation. It should be construed to mean an ‘originator’, to whom anything owes its origin.

In Eastern District of Texas’s Oasis Research, LLC v. Carbonite,[2] the US Federal Court held that a person who merely suggests an idea but not the means of accomplishing it is not an inventor. In TS Holdings, Inc. v. Schwab[3] it was further clarified that merely financing or initiating the process of invention does not qualify the person to be a patent-holder.

A person is required to feed general information about the relevant field and define the  parameters to the AI for initiation of an invention.[4] A direct application of the abovementioned precedents suggests that a person who is merely financing, owning AI or feeding the data and initiating the inventive process is insufficient to qualify himself as an inventor for the purposes of patent law.

Can an AI be an inventor?

On July 29, 2019, Stephen Thaler filed the world’s first patent application before USPTO for an AI created invention. He utilized his AI system named DABUS for creation of a new type of beverage containers by feeding general information about the field in DABUS’ neural network. In his application, Thaler named DABUS as the inventor and himself as the assignee of the AI’s invention.

The USPTO rejected this application because of the mandate of US patent statute which only recognizes inventions created by natural person or group of natural persons. USPTO also reasoned that the threshold for inventorship is the conception of the idea and the scope of the statute is very narrow to accommodate conception by an AI.

Thaler also filed patent applications before UKIPO and EPO. Both the applications were rejected[5] in the same fashion as USPTO. They reasoned that an inventor can only be a natural person and not an AI or machine. On 28 October 2019, the UKIPO updated its patent manual, whereby it clarified that an AI is not acceptable as an inventor for the purposes of seeking patent.

At present, AI cannot not be considered as inventors in most jurisdictions including US, Japan, Korea, China, UK, Switzerland.[6] Patent legislations in all these countries require conception from a natural person for claiming inventorship.

Ownership of inventions by AI machines

Generally, the creator of the invention gets the ownership right over a patent. However, the ownership of these rights may pass from an individual to another individual or an employer by the way of an assignment or through law. From this discussion the issue arises whether an AI can be a patent holder for its inventions or the patent right automatically transfers to the person owning the AI.

UKIPO answered this issue in DABUS case,[7] where it held that AI does not have a personhood status in law and lacks both moral and legal rights which therefore prevents it from holding any ownership rights like a natural person. The underlying object behind granting patent is to incentivize innovation, however, an AI or machine does not require any  such incentive. Thus, it cannot hold any patent for both legal as well as practical reasons.

In the same breath, the Director General of WIPO also rejected the notion of ownership rights for AI. He further reasoned that there is no efficacy in granting a to an AI since sooner or later, a natural person is involved in the creation.

Why patent protection for inventions by AI machines might be beneficial?

  1. Most experts suggest that AI will be at the crest of innovation within the next 20 years. Patent protection would further boost the AI created inventions to step into the future- inventions although may be created by AI machines, the patents for the same are assigned to the creator of the AI machine itself, since currently the law does not recognize AI machines as capable of holding patents as ‘inventors’. Hence in this sense, it boosts innovation and the creators of such AI machines have an incentive to employ such machines to create new inventions.
  1. From a utilitarian perspective, patent protection ensures the greatest good of the greatest numbers. It ensures that the society benefit from the creation of socially valuable products by AI in a cost-effective manner and the subsequent disclosure of the technical know-how. It will balance the right of individual against the society by incentivizing the creator of the AI to develop, own and use AI for creation of novel inventions.
  1. No patent protection would force the owner of the AI to file applications identifying himself as the creator of the invention which is morally wrong as it degrades the value of human inventorship and will also mislead the public.

 

Policy Considerations and Need for coherent Guidelines

Late Prof. (Dr.) Shamnad Basheer had highlighted the ability of combinational creativity of AI machines,[8] that is, the ability to come up with new innovations by combining existing prior arts in infinite ways. This ability of AI machines can also be used to easily surpass any existing patent literature and come up with innumerable novel innovations.

Hence, there is an alarming need of policy considerations for dealing with AI-generated inventions. Although at present AI generated inventions form a small chunk of innovations in economic terms, they may become a significant part of research and development in the future due to its exponential growth.

UKIPO also observed in the DABUS case that there is a void in the current patent system regarding AI generated inventions and hence, invited and encouraged for a wider debate on this topic for finding creative solutions to accommodate such inventions.

In the absence of any clear guidelines regarding the inventorship criteria and ownership of patents, the general welfare of the society is bound to suffer.

The patent offices of Japan, Korea, China, US and Europe held its first meeting on New Emerging Technologies and AI on 20 March 2020 in Berlin. One can only hope that the matter of handling applications for AI-generated inventions will be high on the task force’s agenda.

[1] Donald Goldstein v. State of California, 412 U.S. 546 (1973).

[2] Oasis Research, LLC v. Carbonite, Inc., Case No. 4:10-CV-004358 (E.D. Tex. Jan. 8, 2015)

[3] TS Holdings, Inc. v. Schwab, No. 09-CV-13632, 2011 WL 13205959 (E.D. Mich. Dec. 16, 2011).

[4] Watson cooks up computational creativity, Florian Pinel, https://perma.cc/GGV7-NHT4.

[5] BL O/741/19, UK Intellectual Property Office, https://www.ipo.gov.uk/p-challenge-decision-results/o74119.pdf.

[6] Artificial Intelligence and Inventions, HFG Law and Intellectual Property, http://www.hfgip.com/news/artificial-intelligence-and-inventions.

[7] supra note 5.

[8] Obsessing on Newness: The Death of Patents?, Shamnad Basheer, SpicyIP, https://spicyip.com/2018/01/obsessing-on-newness-the-death-of-patents.html.

Image source: Photo by Hitesh Choudhary on Unsplash