updated on 25 February 2020
Question
What is AI’s role in the future of inventions?On the 4 December 2019, the UK Intellectual Property Office (IPO) made a landmark decision: artificial intelligence cannot be listed as the inventor of a patent. This article considers the impact of this decision on the future of AI as an inventor.
Key case study
The facts leading to the decision concerned an AI machine called DABUS (Device for the Autonomous Bootstrapping of Unified Sentience), created and owned by Dr Stephen Thaler. DABUS invented two new products: a drinks container and a flashing device to attract attention. DABUS generates novel ideas by running lots of cycles and combinations of input data, and then considering the validity of the new idea by comparing it to pre-existing knowledge. To enable this generation of novel ideas, Thaler spent several months ‘training up’ DABUS before the system could generate the ideas itself.
The key legal points that are now a hot topic for AI can be found in the Patents Act 1977 (the Act), namely:
Thaler argued that a ‘person’ under s13 should not be limited to meaning a ‘human’. For example, a company is considered in law to be a person. In any event under s7(3), DABUS most accurately fulfils the definition of the inventor. To write otherwise, Thaler argued, would not be honest and would mislead the public. In fact, Thaler said that there was a moral right to identify DABUS as the inventor; to do otherwise would undermine the purpose of naming the inventor in the first place.
What does the case mean for future AI creations?
Interestingly, the UK IPO explicitly accepted that DABUS created the inventions outlined in the applications. However, the hearing officer determined that sections 7 and 13 were intended to refer to the same person, namely a human person and not an AI machine. It is settled law that an inventor cannot be a corporate body and there have not been any judgements or rules to the contrary, so the analogy is that an AI machine such as DABUS also cannot be considered as the inventor.
In the short term, this will mean that AIs cannot be listed as inventors on patent applications. The UK IPO has even updated its guidelines for patent examiners to explicitly states that “an ‘AI Inventor’ is not acceptable [as an inventor] as this does not identify ‘a person’”. (Interestingly, this update came before the DABUS decision). But the comments from the hearing officer (considered below) suggest that in the long term, the attitude towards AIs as inventors will begin to change.
Why is this an issue?
The default position in patent law is that the inventor will be the owner of an invention, unless another person or entity has a legal right to the invention. This might arise where the inventor discovered the invention in the course of their employment, or where the inventor assigns rights in the invention to a person or company with the funds and resources to protect and exploit the invention. The owner of an invention can apply for a patent, which means they can exercise a monopoly right over the technology, preventing third parties from using any competing technology which falls within the claims of the patent. This is a valuable right as it means the patent holder has a monopoly and competitive advantage, and can take action against others who copy the technology or infringe the patent. The patent becomes an asset which can be sold, purchased or licensed.
However, DABUS is not a sentient human; it cannot determine when a different company has used its novel idea or pursue court action against an infringer. If an AI cannot recognise when its rights have been infringed, or take action against any infringement, then what value does the patent actually have?
Does the law need to adapt?
The hearing officer in the DABUS case concluded by saying that the present legal “system does not cater for such inventions and it was never anticipated that it would, but times have changed and technology has moved on. It is right that this is debated more widely and that any changes to the law be considered in the context of such a debate, and not shoehorned arbitrarily into existing legislation”.
In fact, in response to DABUS’s patent applications in the US, the US equivalent of the IPO (the US Patent and Trademark Office) published a notice calling for public and industry opinions on AI and IP law (see here).
Additionally with increasing discussion around AI in patent law, the World Intellectual Property Organisation has also launched its own public consultation on AI and IP policy more generally (see here). The consultation includes questions around whether an AI machine should be capable of being listed as an inventor of a patent, whether the law should even allow for machine-generated inventions to be patentable, and how other IP concepts such as copyright should interact with AI.
How can the law adapt?
There are various ways that the impasse between an AI inventor and rights enforcement could be resolved. One way would be to allow joint inventorship between the AI and the human creator of the machine, with the ownership of the patent attributable to the human co-inventor. However, this option creates a conceptual difficulty as, technically, the human co-inventor did not create anything at all; the AI did. Instead, the co-inventor acts as a token natural person as a work-around to the current legal provisions.
An alternative and more long-term solution could be to amend the wording used in s7(2) of the Patents Act, which details in what circumstances a patent may be passed to someone other than the inventor; namely either by law or agreement, or by being a descendant to the inventor or person who has acquired the patent by law or agreement.
Under the existing wording, there is no adequate way to justify Thaler as the owner of the patent: he is not the inventor, nor has he been transferred the ownership, nor is he a successor in title. Adding a fourth subsection to s7(2) would make provision for the creator of an AI system to be recognised as owner of the patent where the inventor is an AI system. A similar existing provision can already be found in UK copyright law, where computer-generated work can be attributed to the human creator of the computer system (s9(3) Copyright, Designs and Patents Act 1988).
Conclusion
As AI continues to rapidly develop, the issues surrounding its interactions with IP law will become ever more prevalent. As you have seen, there have already been calls on an international scale to start the discussion. The result is that existing legal principles will have to be reconsidered and revised in light of the emerging technology, leaving lawyers at the forefront of change.
Victoria Potts is a first-year trainee solicitor in Shoosmiths’ Birmingham office, currently sitting in the intellectual property team.