As you will be aware from reading this newsletter, the DABUS saga on whether a machine can be named an inventor has taken some twists and turns this year. Before taking a glance back at the highlights, let me just say that I am amazed at the amount of debate this has stimulated in the IP community. I know of no other topic where so little factual enquiry (do inventing machines exist? when does a machine cease to be a tool and can make inventions? do any autonomous machines even exist?) has created so much interest in legal principles that may or may not become relevant one day. Therefore, for me the highlight of this topic this year was the publication of this article that looks at the state of the art in AI and ML and asks whether the current debate is anchored in fact or based on misconceptions about what AI does and does not do (you may have guessed by the end of this note to which side of the debate I lean). Or as the IPKAT asks somewhat more pointedly: is this a case of the emperor’s new clothes?
See here for an interview with Prof. Ryan Abbot, author of a book on legal challenges in the context of AI and leader of the Artificial Inventor Project. See here for a recent summary of the debate more widely (including a quote by yours truly).
As we reported here, an application listing DABUS as an inventor has been accepted in Australia and South Africa but the corresponding applications at the EPO and in the US and UK have been refused. The UK appeal decision makes for some interesting reading with some highlights here and here. Next in the time-line is the EPO hearing the appeal against the decision to refuse the application this week. Looking at the preliminary opinion from the Board of Appeal I think that the outcome is unlikely to change. No doubt, the decision, when it is handed down in the new year will make for fascinating reading.
Outside the courts and appeal boards, the UK IPO has launched a consultation on whether the law needs to change to accommodate inventions made by AI. See here for some thoughts. I am generally against making laws based on a hypothetical (the existence of autonomous machine invention in the absence of evidence suggesting this is possible), in particular considering that making changes in this area may land us in hot water in respect of international treaties and patent protection abroad. That said, the copyright and data use aspects of the consultation seem timely and useful.
Finally, what are people thinking – is autonomous devising of inventions / autonomous problem solving the stuff of science fiction or current reality? I have asked the oracle of LinkedIn™ polls and my small sample of 11 kind respondents is saying “science fiction” by a margin of roughly 2:1. Not a huge or very representative poll you may rightly say but interesting nevertheless, I think. Few will be disputing that AI and ML algorithms provide hugely powerful tools that help in problem solving, making discoveries and, yes, devising inventions. But they are just that. Powerful tools programmed to carry out specific algorithmic instructions under the control of the operator of the computer and the code of the software.
The fact that AI and ML algorithms can uncover correlations that were previously not apparent (like a microscope enable us to see structure not visible to the naked eye) and even that the results can be surprising or unexpected has no bearing on the question of inventorship. Computers programmed with AI may automate certain aspects of making an invention, but this is different from autonomously devising one. No doubt, though, the debate will carry on next year…