The updated document SCP/36/5 was presented. This background document on patents and emerging technologies is an update of SCP/30/5 “Compilation of laws and practices relating to patentability of AI-related inventions”.
It includes a wealth of information on:
- Laws and practices relating to patentability of AI-related inventions
- Background and introduction to AI technologies, from machine learning to neural networks, deep learning and generative AI
- Patent protection of AI-related inventions
- AI as a tool in prosecution/administration of the patent system.
The document can be found here: https://www.wipo.int/edocs/mdocs/scp/en/scp_36/scp_36_5.pdf
Addressing fundamentals
In the introduction to the update of this document, it was put forward that fundamental questions of patentability have not been sufficiently addressed so far -- AI-related inventions generally are considered as a subset of computer-implemented inventions and more guidelines and court rulings are needed to clarify the situation and to decrease the legal uncertainty and inconsistencies.
There are certainly many questions regarding the road forward. For example:
- What about when AI develops further?
- How to view prior art, and the skilled person?
- How will this affect the patent system going forward?
These are questions that FICPI is actively considering and which have been discussed at recent FICPI Executive Committee meetings, and captured in speeches at recent FICPI events, as well as in discussions with IP Offices and third-party events.
The best way to get involved in the discussions about these topics in FICPI is to join one of the Study & Work (CET) groups. AI aspects are generally relevant to the IP field in many different ways, both for patents, trade marks, designs and beyond.
At the moment, the growing importance of AI in our field is not sufficiently well understood, apart from the fact that it will likely become an important factor over time.
Use of AI in R&D and for generating new technical solutions
At the WIPO meeting, we heard from a number of experts on the use of AI in research and development, and from companies which are using AI to make the IP process more efficient.
Iprova positions itself as using “the inventive power of the world’s information to invent with unprecedented speed, disruption and efficiency”.
We heard from Julian Nowlan, CEO of Iprova, who holds an MSc in Artificial Neural Networks amongst other qualifications. Their website quotes him as saying, “In my view, by 2025 over 50% of the world’s inventions will be created using a data-driven approach.” His presentation at the SCP meeting positioned Iprova as using technology such as AI to amplify the human ability to invent and that digital tools can be harnessed to achieve a timing advantage. He believes that in the future, inventions will be produced faster and more frequently.
The next presentation came from ETH Zurich, a public research university in Zurich, Switzerland. Dr Gisbert Schneider is a Full Professor at the Department of Biosystems Science and Engineering where his research activities focus on the development and application of adaptive intelligent systems for molecular design and drug discovery.
He noted that while chemistry and pharma companies are currently lagging behind, “big pharma” companies are developing in-house AI capacities. He does see a shift in the drug discovery cycle towards the use of bots, for instance to use generative AI to identify drug molecule hypotheses. This is efficient due to the massively large discovery space for possible molecules (of the order of 10100). He added that AI can present surprising and novel suggestions to the wetlab scientist for evaluation.
However, there is a challenge in that there are only a few million known molecules that are available for training the AI models. Therefore, he said, we still currently need a human-in-the-loop to frame the problem to be solved in more detailed terms.
Dr Rachel Free, a European patent attorney with an MSc in Artificial Intelligence and a DPhil in vision science, from CMS Cameron McKenna Nabarro Olswang LLP, presented an interesting case study. She showed us a hypothetical R&D case for development of a novel multi-modal AI application, and highlighted steps in which generative AI can be used as a resource, for instance for getting inspiration, information and task prioritisation.
Conclusion
From attending conferences and meetings, I see a certain fuzziness about the term “AI” being used in the IP field generally. It is often used as a sort of catch-all for all kinds of computer automation, including conventional rules-based approaches as well as non-generative and generative neural network-based approaches. This is the case both for AI-related inventions and AI tools.
This fuzziness sometimes results in discussions assuming that the future of AI development will continue more or less at the pace of recent years (extrapolation), and that we can safely assume it to become much more “intelligent” and “creative” over the coming few years.
Having said this, there are also people in the industry having very deep knowledge in these topics, including with the patent offices. This is a great resource in putting these new abilities to best use going forward.
My view is that discussions would probably benefit from a sober discussion regarding the limitations of current state-of-the-art transformer-based AI technology and to what extent these limitations can be assumed to exist also in the medium term or possibly be dealt with.
Such a discussion could perhaps result in more concrete thoughts about how to put AI tools to work in hands-on situations, and what to reasonably expect from this.
Next steps
FICPI members can log in to view the full meeting report, which also includes highlights on current use by Patent Offices in Japan and France of AI tools.
View presentations from FICPI events including the FICPI Open Forum 2024 in Madrid:
- Drafting without borders – High Tech
- AI & IP: an arranged marriage?
- How patent law firms can support their clients’ innovation growth
Read the report on the Executive Committee meeting workshop (London, 2023):
Read the report on the Executive Committee meeting workshop (Goa, 2024):
Explore the FICPI Study & Work Groups and get more involved with FICPI’s work