Artificial Intelligence is a technology that has arrived, even though many of us may be unaware of it as we go about our daily lives.
But what is AI? This is open for debate, but at a high level AI relates to a computer which is able to act in an intelligent manner, in a similar way to humans.
In some areas AI intelligence is superior to ours. For example, a supercomputer including machine learning can take the rules of chess and, within a matter of hours, learn by repeatedly playing itself using those rules to determine a winner. The knowledge gained by this learning or training exercise can then be used by the supercomputer to beat even the best grand master chess players.
Other examples of AI are self driving cars and AI can now be found in even smaller products such as Apple’s Siri Personal assistant or Amazon’s Alexa smart home hub.
Although AI does have its drawbacks, the benefits are clear, and it can make a positive change to our daily lives. Therefore many companies are making substantial investment in AI to provide products which include these benefits. Naturally, these companies are seeking to protect their investment and the European Patent Office, as well as many other patent offices, have seen a big increase in the number of patent applications being filed by those seeking to protect their inventions in this area.
But are those AI related patent applications ever likely to be awarded patents? In Europe at least, the answer is usually yes, provided that the usual requirements laid down in the EPC in respect of novelty, inventive step, clarity, and sufficiency of disclosure are met.
However, patenting this technology is not without its difficulties.
Helpfully, the European Patent Office recently held a conference to discuss the impact of AI on the patent system. This conference provided a useful insight into the EPO’s approach in assessing patentability in this field as well as in the related field of computer-implemented inventions.
For AI related patent applications, the EPO uses the well-established two-hurdle approach for mixed-type inventions which have both technical and non-technical features.
In this assessment, the first lower hurdle to be passed is to determine whether the claimed subject matter has a technical character. Therefore, for AI related inventions, a question likely to be asked is:
Does the AI (and the Machine learning) have method steps which contribute to the technical character of the invention?
If yes, then the second higher hurdle to be passed is to determine whether an inventive step is present.
When assessing inventive step, and hence patentability of such a mixed-type invention, all those features, including the features which when taken in isolation are non-technical, but do contribute to producing a technical effect serving a technical purpose, are taken into account. Such features thereby contribute to the technical character of the invention.
However, features which do not contribute to the technical character of the invention cannot support the presence of an inventive step.
But how might one show a technical effect? The guidance from the conference was to include in the patent specification as much information about the technical effect and the corresponding technical features as possible. The inclusion of test or experimental data in the specification might be used to verify any technical effect. This increases the plausibility of the argumentation in relation to inventive step.
But of course, the entire assessment of patentability is dependent on what is meant by features which have a “technical character” and what a “technical effect” is.
Helpfully, the EPO will be providing additional guidance as to how they assess patentability of AI in the 2018 update of the Guidelines for Examination which will be published in October 2018. This will feature examples relating to AI as well as detailed information on the technicality of computer implemented inventions based on decisions by the EPO’s boards of appeal.
However, the conference did consider whether three broad areas of AI might, in principle, be patentable.
Firstly, core AI was discussed. However, core AI often relates to algorithms as such, which as mathematical methods are not patentable in Europe.
Secondly, trained models and machine learning were discussed. It was concluded that:
Comparative examples and parameter ranges might be needed and inventive step practices from other areas such as industrial chemistry might be relevant. It was suggested that the EPO could be more lenient regarding the technicality conferred by specific datasets and allow the “second use of a model” by analogy to second medical use claims in pharmaceutics. Uses should not be considered equivalent if arrived at by different means.
Thirdly, AI as a tool in an applied field, defined via technical effects were discussed. It was concluded that:
In fields such as autonomous vehicles and healthcare, AI might be claimed as a tool embedded in a larger claim, defined as a functional feature providing a solution. In such cases, the claim as a whole should be assessed for inventive step, and not just the functionally defined AI tool.
The above examples provide some useful guidance as to particular areas of AI which might be patentable. However we shall have to see how closely EPO examiners follow such an approach. Nevertheless, the conference provided an extremely useful insight into the approach that EPO examiners are likely to use when assessing patentability of AI related inventions.
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.