The outcome we probably expected. The Court of Appeal reaches a verdict on the Emotional Perception AI case
The Court of Appeal has now handed down the judgment in, arguably, the summer of 2024’s most anticipated UK patent court case – Comptroller-General of Patents, Designs and Trade Marks v Emotional Perception AI Limited. The judgement follows the May 2024 hearing in which both sides put forward their arguments on appeal.
The very readable judgement provides a nice introduction to artificial neural networks (ANNs), which may be welcome background for those not familiar with this trending area, before delving into the case in question.
In brief, the invention related to a file recommendation system for providing a media file recommendation to a user. The invention offers suggestions of similar music (or other media) in terms of human perception and emotion irrespective of the genre of the file – arriving at this by putting the file through a ANN.
The appeal was from the High Court judgement. Somewhat surprisingly, the High Court found that in the case in question no computer program was involved, and even if the provisions of s1(2) of the Patents Act did apply to the case in question; in other words, that neural networks do engage the computer program exclusion, then there was a substantial technical contribution so the subject matter was not excluded as a computer program ‘as such’.
The High Court decision was surprising in that it went in a direction different to many other jurisdictions for this type of subject matter and how cases had been previously examined in the UK. This shift in approach led the UKIPO to update its guidelines on examining computer implemented inventions (CII) relating to ANN late last year.
Clearly this judgment on appeal was highly anticipated, as it would determine how machine learning ANN inventions are handled here in the UK: if the Court of Appeal upheld the decision it would make the UK a more favourable jurisdiction for patent applications in this area.
However, the Court of Appeal has instead gone in the opposite direction to the High Court and overturned the decision. In doing so, many would argue that it has sensibly taken the approach that appears to be more consistent with previous case law, finding that the application is in fact excluded from patentability.
In the appeal the Court of Appeal considered two questions:
- What a computer program is and whether there is a computer program in an ANN; and
- Was the [High Court] Judge wrong to hold that the claimed invention involves a substantive technical contribution.
For the first question, the Court of Appeal defined a computer as “a machine which processes information” and a computer program as a “set of instructions for a computer to do something”. The Court of Appeal confirmed that whether a computer program is fixed in a permanent form or not doesn’t alter the fact that it is a set of instructions for a computer to do something and that an ANN however implemented is “clearly a computer – it is a machine for processing information”. Therefore, their view is that the weights of the ANN are a program for a computer and should be considered under the exclusion – drawing similarities with the EPO’s general approach.
Turning to question two, the Court of Appeal confirmed that ANN inventions are in no better or worse position than other computer implemented inventions – they may still be patentable if they make a substantive technical contribution outside the exclusions. Considering the invention in question, the Court of Appeal found that it is the “semantic similarity of the files here which gives rise to their recommendation … is not a technical matter at all”. The Court of Appeal stated that what mattered in this case is the correct characterisation of the data being transferred and found that brings the issue back to “the aesthetic and therefore non-technical quality of this aspect of the contribution”.
This decision brings the UK close to where it was prior to the High Court decision last year, and is consistent with the approach prior to this. While a more generous approach to the exclusions might have been nice for applicants seeking inventions in the field of AI, consistency and predictability are also good for applicants. It is worth noting that the UKIPO’s approach to the exclusions differs from the EPO. Where the UKIPO consider the contribution of the claims is wholly within the exclusions they will refuse to perform a search and no examination will be performed. The EPO’s approach would instead be to search and examine giving the applicant the opportunity to respond (and argue there is a technical effect). This means that AI applicants are often better off filing at the EPO rather than the UKIPO if faced with the choice.
Given that this decision returns matters to something near to the status quo, it seems unlikely to be the subject of further appeal. It is instead likely that this decision will be cited for some time to come against computer implemented inventions filed at the UK patent office.
At Barker Brettell we have a team of European and UK patent attorneys with a wealth of experience in patenting computer implemented inventions and AI technologies. If you have an invention in this area that you wish to protect please contact the author or one of the team who would be happy to help.
