The Artificial Intelligence (‘AI’) machine called the Device for the Autonomous Bootstrapping of Unified Sentience (‘DABUS’, for short), recently became the world’s first non-human inventor for a granted patent in South Africa after patent offices in the United Kingdom (UKIPO), Europe (EPO) and the United States of America (USPTO) refused to grant the application. The patent relates to a fractal-inspired food container created by AI under the support of Dr Stephen Thaler. AI is increasingly being used to find new chemical compounds and repurpose drugs. The use of computers and machines to produce various patentable inventions in fields such as chemistry and biotechnology is not new. However, should intellectual property laws be updated to protect such AI generated discoveries?
This subject has prompted an extreme debate on the future of patent law and policy in the past and has recently continued to build momentum after the latest grant of the South African patent naming DABUS as an inventor on the granted application. While there is general agreement that AI generated inventions are unsuited with the concept of human inventorship, it remains uncertain as to what extent concerns regarding ‘non-human’ inventiveness can be justified. For years Ryan Abbott, a professor at the University of Surrey, has led a battle for patent offices around the globe to recognise AI as inventors. However, is the grant of this South African patent application indeed a moment of triumph for Prof Abbott, Dr Thaler and the likes? Why have the EPO, the UKIPO and the USPTO refused to grant this application?
Unlike the EPO, the UKIPO and the USPTO, the South African patents office does not subject patent applications to any form of substantive examination. Put differently, South African patent applications are not investigated for their novelty or inventive merit. The South African patents office only formally examines patent applications in a check box sort of evaluation and if all formalities are met, the patent application will simply be granted. Any detailed examination of such patent application is reserved for subsequent challenges in court.
It is therefore easy to recognise that the prosecution processes the DABUS patent application went through in South Africa cannot be compared to the more stringent prosecution processes it was subjected to at any other patent office in which it had been rejected. Nonetheless, the grant of this patent application in South Africa has sparked interest and reignited the policy debate surrounding the topic.
Patent law was traditionally developed with human inventors in mind. The acceptance of AI inventors raise a wide range of novel issues which the current patent framework may not be able to adopt simply. For example, what is meant by AI generated inventions or how do we define computer autonomy during the inventive process? Notably, when raising these wide range of novel issues and addressing how patent law needs to be modified in the introduction of AI generated inventions, policymakers neither provide a working technical definition of such inventions, nor do they define how they differ from AI aided inventions, nor review the technological state of the art. A more detailed inquiry into computer/machine generated inventions as well as an inquiry into general patent law and practice globally will need to occur in order to decide whether or not patent offices should grant such machine generated patent applications.
According to general patent law and practice, a patent is granted as part of a statutory monopoly right which provides protection for an invention. This is often referred to as a quid pro quo. In essence, the right to prevent others from exploiting the patentee’s invention for a limited time period (generally 20 years – as is the case in South Africa) is granted to a patentee in exchange for his/her full disclosure of the invention so that others may exploit it once the patent has lapsed or expired. According to the South African Patents Act 57 of 1978, a patent may be granted for any new invention which involves an inventive step and which is capable of being used or applied in trade, industry or agriculture.
An invention will be considered new or novel if the invention does not form part of the state of art immediately before the priority date of the invention. Broadly defined, the state of art includes all matter made available to the public (in South Africa or elsewhere) by written or oral description, by use, or in any other way. Machine generated inventions seem to lack novelty where the AI algorithm does not have any variability as to its outputs and where it specifically relies on data sets previously used before the patent application. However, should an algorithm rely on randomness or other variability it would most likely satisfy the requirement of novelty. Even though the requirement of novelty may be met by a machine generated invention there remains the danger that such AI systems may make it substantially more tricky for an applicant to establish novelty as AI could dramatically expand the prior art universe. The growth of prior art references will also further burden the general patent examination process jeopardising efficiency, sustainability and level of scrutiny currently in place in countries practising substantive examination.
An invention will be considered to have an inventive step if it is not obvious to a person skilled in the art, having regard to the state of the art immediately before the priority date of the invention. The determination of obviousness goes beyond the depth of this article, however, simply put a patent application cannot be granted for an invention which is within the reach of the average of an average expert in the field in which the patent relates. With regards to AI generated inventions, the AI’s inventive output would in many cases satisfy the inventive step and non-obviousness threshold. Be that as it may, an amendment to the existing definition of a person skilled in the art to include ‘a person equipped with AI’ may need to be evaluated in cases of AI generated inventions.
Another important requirement to consider is that an invention needs to be adequately described and/or illustrated and exemplified in a manner that allows the invention to be performed by a person skilled in the art of such invention. It is often far too difficult to describe how the algorithm of an AI generated invention works and operates. This makes patent law’s disclosure requirement under attack by AI which can negatively affect the patent system as a whole.
As for inventorship, it is assumed that patent law suggests that the inventor of an invention must be a natural person. This argument has been supported by lawmakers in Europe whom have continuously linked the concept to the contribution to the inventive concept (for example: in the United Kingdom, patent laws state that an inventor refers to the actual devisor of the invention). The issue with accepting AI and machine’s as inventors, as in the case of the DABUS patent granted in South Africa, is that a number of possible contenders for ownership arise. These include the AI’s owner, the person who created the AI system otherwise known as the programmer of the AI software, the data set provider and/or the person who first recognises the significance of the result.
One might argue that the South African patents office did not make a mistake in granting the DABUS patent with AI mentioned as the inventor. The reason for this is that numerous policies in South Africa have suggested that that South Africa seeks to increase innovation to aid in solving the country’s socio-economic issues by economic growth and employment. However, as pointed out above, there is still a need to evaluate patent laws and policies prior to allowing AI to be mentioned as an inventor.
Unfortunately, the lack of substantive examination in South Africa has made it hard to prove whether the South African patents office in fact intended to grant the DABUS application and we will have to wait for a party to challenge the granted patent in court in order to receive a proper judgment on the matter.
Written by Claire Gibson
This article is for general information purposes and should not be used or relied on as legal or other professional advice. No liability can be accepted for any errors or omissions nor for any loss or damage arising from reliance upon any information herein. Always contact us At DKVG Attorneys for specific and detailed advice.