Artificial Intelligence (AI) continues to be in the news in the Intellectual Property world.
In a recent development in the UK, a decision to refuse an AI patent application was heard at appeal by the UK High Court. Interestingly, the appellant was unrepresented at the High Court hearing and the judgement was remotely handed down on 22 January 2021.
The UK IPO originally refused the patent application on the basis that the application was excluded from patentability under section 1(2) of the UK patents act as being a computer program as such. Broadly speaking, the application relates to a structured array of data that is said to enable the evolution of AI.
In more detail, the specification explains that for AI to evolve without human intervention, the structure of the AI must be designed in a way that allows the AI to pass on who it is and/or what it knows with as much ease as possible using a “genome” that contains the genetic information of the AI. The specification also explains that for the AI genome to be used in machines, three components are required, namely:
- The genome itself – structured similar in nature to the structure of a human genome;
- An Artificial Intelligence Genome Organizer containing information about the genome; and
- An Artificial Intelligence Genome Controller which is a program used for the automation of genome activity.
Unfortunately, the specification does not provide much more specific detail beyond this, and there was a lack of specific examples as to how the AI evolves so that it can perform better than previous generations or versions of AI.
The High Court Decision
The Court followed the same legal test as the UK-IPO in determining whether an invention is excluded from patentability under section 1(2).
The (Aerotel/Macrossan) test is outlined here, and in summary the test has 4 steps which are to:
1. Construe the claim;
2. Identify the actual contribution;
3. Ask whether it falls solely within excluded subject matter; and
4. Check whether the actual or alleged contribution is technical in nature.
Having determined that the contribution of the claimed invention was a particular way of structuring and organising data that may allow for the production and evolution of future AI code, the judge went on to agree with the UK IPO that the contribution was not technical in nature.
The AT&T signposts
So why was this considered not to be technical in nature? The judge referred to the AT&T signposts which, if met, point to a technical contribution.
Firstly, judge noted that the specification does not describe in any detail how the way of structuring and organising data leads directly to the creation of physical systems outside a computer. Rather, it merely suggests that replicating the structure of computer code in the same way that the building blocks of biological life are organised should lead to the same results in computer systems as observed in nature. The judge commented that this is a theoretical effect and there is no direct external technical effect outside the computer, so that the first signpost was not met.
Secondly, the judge commented that the second signpost points to a technical effect when the contribution is made at the level of architecture of the computer or where the effect is produced irrespective of the data being processed or the applications being run. However, in the present case, the organisation and structuring of the claimed AI code did not operate at an architecture level, irrespective of the data being processed or the applications being run. Thus, the second signpost was not met.
Thirdly, the judge noted that the third signpost says that a technical effect can be found when the invention results in the computer being made to operate in a new way. However, the judge noted that in the present case, there was no evidence that a computer system was being made to operate in a new way, so that the third signpost was not met.
The fourth signpost was not considered to be relevant to the present case. However the fifth signpost, where a technical problem is solved by a technical solution, rather being circumvented by a non-technical workaround was considered. Unfortunately for the appellant, the judge commented that there was no evidence that the problem of making AI evolutionary (nor any other problem) had actually been solved using the system described in the specification.
It is also interesting to note that in the decision, the judge added the following comment in relation to “alleged technical contribution”.
- Mr Birss added the words “or alleged contribution” in his formulation of the second step. That will do at the application stage—where the Office must generally perforce accept what the inventor says is his contribution. It cannot actually be conclusive, however. If an inventor claims a computer when programmed with his new program, it will not assist him if he alleges wrongly that he has invented the computer itself, even if he specifies all the detailed elements of a computer in his claim. In the end the test must be what contribution has actually been made, not what the inventor says he has made.
So does this mean that AI inventions are no longer patentable in the UK?
In short, the answer is no.
Rather, the High Court has confirmed that the UK IPO correctly applied the legal test as to whether the claimed invention was excluded from patentability. Further, as we have previously commented on here and here, under certain circumstances, AI inventions can be patentable in the UK and indeed in other countries as well.
So what can we learn from this case?
Firstly, in a communication from the UK-IPO during early prosecution of the application, the examiner was of the view the application provided very little specific technical detail upon which to base a search and invited the applicant to withdraw the application and to request a refund of the search fee. So in essence, the UK IPO had serious concerns at an early stage that the specification did not disclose the invention in sufficient detail to allow the skilled person to reproduce the invention.
Therefore, it is always worth considering whether specific examples, such as specific data types can be included in the patent specification to illustrate how these are used by the AI to achieve the claimed benefits.
Further, it is also worth considering if evidence can be provided in the specification which shows that the alleged technical contribution does indeed result in an actual technical contribution when implemented on a computer. The evidence may be in the form of specific examples and technical data, applied to specific use cases, which show how the claimed AI invention results in an actual technical contribution when running on a computer. Alternatively or in addition, it is worth explaining in the specification how the AI invention has been developed to take into account the underlying technical operation of the computer system on which it is run. This may help in showing the presence of an actual technical contribution.
This article is for general information only. Its content is not a statement of the law on any subject and does not constitute advice. Please contact Reddie & Grose LLP for advice before taking any action in reliance on it.