Boult Wade Tennant
Bulletins » AI inventorship: dissent at the UK Court of Appeal?

The long running saga of the AI “DABUS” and its creator Dr Stephen Thaler reached yet another milestone this week with the handing down of a decision from the UK Court of Appeal. The decision, however, included a dissenting judgement from one of the UK’s leading intellectual property judges, questioning the role of the UK Intellectual Property Office (UKIPO) in the dispute.

The story so far
As set out in our previous bulletin on the topic Dr Thaler had previously filed a number of patent applications worldwide, naming an Artificial Intelligence (AI) called “DABUS” as the sole inventor, in an apparent attempt to force the question of whether patent offices will accept AI inventors. The UKIPO, the European Patent Office, and the US Patent and Trademark Office all concluded that AIs could not be named as inventors – a position which in the UK was upheld by the High Court. Dr Thaler had more success before the South African Patent Office and the Australian Federal Court, though the latter decision is still under appeal.

In the UK the UKIPO (and later the High Court) effectively rejected his applications on the basis that he, not least, failed to comply with the formal requirement to “identify the person or persons whom he believes to be the inventor or inventors”. They concluded whilst Dr Thaler had named DABUS as the inventor, as DABUS was not a person the above requirement was still not met. As such the UKIPO and the High Court felt that the UK patent applications must fail.

The Court of Appeal decision
By a two to one decision the Court of Appeal has decided to uphold the finding of the UKIPO and the High Court – that the UK applications are to be deemed withdrawn for non-compliance with Section 13 of the UK Patents Act 1977 (the requirement to name the inventors). The implication being that any UK application where an AI is named as sole inventor would be bound to fail in the same manner.

A dissenting opinion – form or substance?
What is most interesting however is that the decision was not unanimous – Lord Justice Arnold and Lady Justice Elisabeth Laing formed the majority view that the applications should fail, with Lord Justice Birss instead dissenting and arguing that Dr Thaler’s applications should have been allowed to proceed.

Whilst such dissent may give Dr Thaler hope in the possibility that the Supreme Court (where it is speculated the case will eventually end up) may overturn the effective refusal of his patent applications, Birss LJ’s dissenting opinion is unlikely to give any comfort to those who believe AIs should be considered as inventors.

Indeed, all three judges agreed that under the current UK statute inventors must be natural persons and that as a matter of law an inventor could not be an AI. Birss LJ in fact took issue with the supposed concession by the UKIPO that DABUS was the “actual deviser of the inventions”, a position Birss LJ indicated was not consistent with UK statute.

Where the judges differed appears to be more in the role of the UKIPO in enforcing this. Birss LJ appears to be of the view that it is not for the UKIPO to look behind the honest belief of the applicant when they indicate that inventor is an AI – and it is certainly not for the UKIPO to effectively refuse an application on this basis. Instead it should be left to anyone who feels that they should instead have been named as inventor to challenge the patent (or patent application) in the court should they wish.

In their majority judgement however Arnold LJ and Laing LJ took the alternative view – that the naming of an AI on the face of it cannot comply with the requirement for the applicant to “identify the person or persons whom he believes to be the inventor or inventors” and thus the statute left the UKIPO with no choice but to effectively refuse the application on formal grounds.

For now – until perhaps the Supreme Court is in a position to give a view – it is the latter practice that will prevail at the UKIPO.

A postscript on inventiveness
That the applications were refused on the grounds of non-compliance with formalities underlines how throughout the proceedings in the UK none of the more substantive issues – whether anything has been invented, and whether that was done by the AI – have been assessed or even challenged. This is a function of the order in which patent applications are often examined, formalities first followed by the generally more difficult and time consuming substantive examination – i.e. deciding if there even is an invention.

Into this vacuum however there has been some speculation as to what allowing AI inventors may mean for substantive examination. In particular, in the UK, an invention is only acknowledged if it is not “obvious” to the notional “person skilled in the art” (i.e. it must possess a so-called inventive step). Would not this notional person now become a notional AI should AI inventors be allowed? Could human inventors ever meet such a threshold as AI advances?

In his dissenting judgement Birss LJ has, as an aside, provided some thoughts suggesting that in his view there is little connection between the attributes of the notional “person skilled in the art” and those who we might deem inventors – a reminder that the “person skilled in the art has attributes no real human being does or could have”. However, given the majority judgment it seems that in the UK, absent a change in law or the intervention of the Supreme Court, such questions will remain academic for at least the time being.

Relevant sectors
Artificial intelligence and machine learning
Communications and networks
Computing and software
Consumer goods and retail
Electronics and electrical devices
Energy and green technologies
Food and beverage
Industrial manufacturing and processing
Medical devices and diagnostics 
Trade marks