The EPO has now issued its first-instance decision refusing the DABUS applications in which the applicant tried to specify DABUS, a computer program, as the inventor of two European patent applications. This follows a similar decision reached at the UKIPO (see our note here).
Leaving to one side the factual questions around inventorship by machines (more on that below), this begs the question: Is this decision correct?
In my view the answer is undoubtedly “Yes, it is”. The EPO and the UKIPO are applying the provisions around the designation of inventorship correctly, both at a formal and substantive level. Regarding the substance, under current law, an inventor is the first owner of an invention, which may be transferred by way of agreement or employment. Therefore, an inventor must be able to own and dispose of property. The inventor must be a legal person to fulfil the functions foreseen by patent law.
It follows that recognising a program or machine as an inventor, would require either a change in law that enables machines to have legal status or a change in patent law that dispenses with the requirement of an inventor as the first owner of an invention.
Neither change is within the powers of the patent offices or its appellate bodies. Having this argument with the patent offices could be seen futile aside from the publicity that it clearly has raised. What would be needed is legislative change.
Before that can be contemplated, one should ask the question whether it is in fact established that a program or machine can be an inventor, a deviser of an inventive concept? Machines programmed with machine learning techniques can be an extremely powerful tool in discovering correlations in data and knowledge, even unearthing new connections not previously known.
However, it is less clear whether machines are currently, or even foreseeably, able to autonomously reason in a way that results in an inventive concept, i.e. a purposeful answer to a question to improve an existing technology, and to do so in a general purpose architecture that does not need specific adaptation in order to be able to “invent” to address a given question? The answer may or may not be yes but there seems to be so far merely an assumption that this is the case, without any clear evidence being published to support this notion. At present the information relating to “machine inventions” that is readily accessible is somewhat neulous, so that it is difficult to assess whether a change in policy should be contemplated or not.
What is needed to take this debate further is for those who make the case for “machine inventors” to make it more clearly and transparently so that a rational evaluation of these claims by the patent community and by policy makers is possible.
For more information or advice please contact Alexander Korenberg or your usual Kilburn & Strode advisor.