In August last year, a pair of unusual patent applications sought to credit artificial intelligence (AI) software as the inventor of two new innovations and caught the attention of international media. Whilst the decision by the European Patent Office (EPO) to refuse the application – after only 16 minutes of deliberation – was hardly surprising due to its non-compliance with the European Patent Convention (EPC), the case has bought a much bigger issue to light: whether patent laws are fit for the future.
In this particular case, researchers from the University of Surrey together with the US creators of artificial intelligence machine, Dabus AI, filed patents which designated the AI as the sole inventor. Ryan Abbott, Professor of Health Sciences at the university and one of the team working on this Artificial Inventor Project (AIP), wanted to explore what would happen if AI was legally recognised as an inventor on a patent. Professor Abbott noted: “If I write a Word document with Microsoft Word, that doesn’t make Microsoft Word an author, and if I use an Excel spreadsheet, that’s not making Excel an inventor.” However, he says that in some cases, software or an algorithm should be awarded inventor status.
The first invention AIP attempted to patent under the name Dabus AI was related to the use of fractals – never-ending patterns which form a geometric shape – in the walls of a food container. The second was related to the use of fractals in warning lights. Neither of these are areas of expertise for Stephen Thaler, CEO of Imagination Engines and creator of Dabus AI. Instead, these innovations were the result of the software being fed reams of data about general subjects. As Mr Thaler admits he would not have been able to come up with the same ideas on his own, the AIP team argues that it would be a fiction to credit him as the inventor, rather than Dabus AI.
It’s important to clarify that AIP is not suggesting that Dabus AI should be the owner of the patent, usually the owner is the company which employs the inventor. So in this case, the inventor would be Dabus AI and the owner would be Stephen Thaler. However, the European Patent Convention specifically requires that each patent application is designated an inventor, and that the inventor must be a natural person. The application must include the family name, first name and address of the inventor. Clearly, in this case, the requirements of the European Patent Convention were not satisfied, as DABUS is not a natural person with a family name and an address.
The applicants could have easily overcome this by naming a person as an inventor; the EPO doesn’t investigate the accuracy of the named inventor so it’s likely that the application would have been allowed to proceed to examination. However, the AIP team argued that it would be morally incorrect for a human to take the credit for an invention which they had not really developed. Additionally, they put forward the idea that properly identifying the actual inventor is a fundamental aspect of patent law, which should not be circumvented in order to fulfil Patent Office formal requirements.
Unsurprisingly, the EPO rejected these arguments during their deliberation, choosing to refuse both applications after only 16 minutes. There is little to be surprised about as the Receiving Section of the EPO was simply following the rules laid down for them by the EPC. Whilst it’s likely that the AIP team will appeal this decision, this also seems destined to fail; there is simply no wiggle room around current patent law. This means that Stephen Thaler will probably fail in his attempt to obtain a patent for the fractal food container and the use of fractals in warning lights. It is unlikely that there will be any negative commercial consequences flowing from this. In fact, they probably selected these two inventions as test cases since they had no real commercial value.
In the UK, where the team filed a similar patent application, the UK-IPO updated its formalities manual, on 28th October 2019 to address future situations of this nature. It now includes instructions that state: “An AI inventor is not acceptable as this does not identify ‘a person’ which is required by law. The consequence for failing to supply this information is that the application is taken to be withdrawn.”
This test case was designed to generate publicity by raising questions about machine inventors and whether it is a possibility that machines could develop inventions entirely on their own – and be recognised for it. In our experience, at the cutting-edge of AI, there is always some human involvement in the development of inventions that use AI. One thing is clear: systems which use machine-learning can, and do, outperform humans. However, they do not do this by copying high-performing people. Instead, they exploit the distinctive capabilities of new technologies such as massive data storage capacity and brute force processing.
Current technology uses AI as a tool that human inventors can use to explore new areas of knowledge. For that reason, it’s advisable to name real people as inventors when filing a patent application. These should be the people who actually developed the invention. In practice, those are the people who programmed the computer or designed its operational parameters.
Although the decision in this case is exactly as expected, it does raise challenging questions for the European patent system. In particular, legislators will need to consider whether naming an inventor or inventors can survive as part of the patent system as AI continues to develop. There are also questions about the nature of inventive step (i.e whether an innovation is sufficiently inventive to be patented) and whether it makes a difference if an invention was developed by a machine rather than a human. Hopefully cases like this will help to inspire a review of patent law, especially as it relates to computer-implemented inventions, so that we can have a law that is fit for the future, rather than the past.