By James Nurton and Gwilym Roberts
100 percent of jobs will be different.
That was IBM CEO Ginni Rometty’s prediction about the impact of artificial intelligence, made in her speech at the CES meeting in Las Vegas in January 2019. We believe her. We’re all beginning to see the impact of artificial intelligence tools across aspects of work, leisure and travel. The focus of the third and final article in our series on AI and IP is therefore: how will the work of IP practitioners be affected by AI?
Experienced practitioners – in-house, in private practice and in IP offices – can testify that our work has already changed fundamentally thanks to the use of dedicated software and automation. If IP Practice 1.0 was paper-based, IP Practice 2.0 has seen the widespread use of electronic filing and searching, and more recently machine translation and language recognition with tools such as Google Patents and the EPO’s Patent Translate.
Towards IP 3.0
Today, we are embarking on the era of IP Practice 3.0, which offers greater automation, more use of AI applications such as image recognition and deep learning, and the customisation of other technologies such as Blockchain.
Used properly, all these tools could increase efficiency, simplify tasks and offer greater value to clients. However, there are obstacles: IP rights are monopoly rights and there are justifiably concerns about over-simplifying the systems we use to protect and enforce them. Businesses need to know that patents or trade marks that have been granted are secure and enforceable: IP is one area where ensuring legal certainty is valued more highly than cutting costs. Similarly, IP offices have strict rules and procedures to follow to ensure fairness and defend the public interest.
IP offices explore opportunities
Despite these hurdles, we are already seeing AI tools proving useful in different aspects of patent and trade mark prosecution, as well as in other aspects of IP work. Notably, the World Intellectual Property Organization (WIPO), together with national and regional IP offices, is exploring the use of AI in at least six main areas:
Automatic classification of patents and goods/services for trade mark applications
Search of patent prior art and figurative elements of trade marks
Examination and formalities checks for trade marks and patents
Helpdesk services (automatic replies to client)
Machine translation, linguistic tools and terminology
Data analysis for economic research
In May 2018, WIPO hosted a meeting of IP offices on ICT strategies and AI for IP administration. In the same month, more than 350 people attended a European Patent Office conference on patenting AI, which raised questions such as how the patent system deals with inventions made by AI (see the IPKat blogpost “An Apology to the Machine”, written by Gwilym Roberts who attended the event). This is a topic that the EPO’s Standing Advisory Committee (SACEPO) is also addressing.
The UK Intellectual Property Office is already making use of big data to improve patent and trade mark searching and worksharing with other offices, as well as in patent analytics and classification. “At the moment, it’s about assistance rather than replacement. Above all these tools help us as examiners to find better prior art, and improve the filing and searching processes for our customers. We’re at the early stages of AI in the sense of machine learning,” says Nigel Hanley, Head of the Examining Group for Electrical and Optical at the IPO.
Faster computers and more data will have an impact on making workflows both more efficient and effective says Hanley. But he warns: “There are lots of problems to deal with. When you have narrow data sets, context is everything. For example, we estimated that there are 14 or 15 different interpretations of the word ‘cell’. There’s a big challenge to get the context.” The IPO has received a grant from the Regulators Pioneer Fund, which is being used to investigate using AI-powered preliminary searches for patents, trade marks and designs. This work includes looking at improved image searching, which until now has proven difficult. The Office is also in the early stages of working with Cardiff University on a patent feasibility study.
The value of AI is becoming evident in litigation too, says Dr Noam Shemtov, Reader in IP and Technology Law at Queen Mary University of London Law School: “In the US, we are already seeing companies that compile data on judicial decisions to help lawyers predict outcomes. Information is power and I think it is only a question of time before we see similar tools being used to analyse cases in Europe, particularly decisions from the EPO and EUIPO.” The use of such predictive technologies was the subject of a recent Financial Times article (‘Big data: legal firms play Moneyball’).
Dr Shemtov predicts that the take-up of such tools could eventually have an impact not only on practitioners, but also on judges and examiners. “No judge likes to be seen as biased. If you are being profiled, you might start to behave differently,” he says. However, he believes it will be ‘a long, long time’ before computers can actually replace judges: “Removing humans from the process would require a complete reassessment of our perception of what justice is.” And, as Hanley says: “IP law is not about finite sets and rules like chess is. A lot of it can be subjective: what does a particular word mean? How do you construe a claim?”
Impact on IP practitioners
IP practitioners have already seen the benefits of automation and some AI tools, for example in the wide use by patent attorneys of EPO-generated classification codes and in the suite of services for trade mark and design searching and classification available through the European trade mark and design network.
As we move into IP 3.0, we will have to change further the way we work. Inevitably, routine tasks can be automated. In all areas of IP, due diligence and dispute resolution will benefit from software that can quickly review large numbers of documents, including registered IP rights and case law, and then organise and prioritise them.
At Kilburn & Strode we are embracing new technologies where they can enhance the services we offer to clients. For example, we are running a trial using bots on non-critical case matters. These bots replicate keystrokes involved in certain tasks so we can render those tasks instantly and error-free. We are also starting to upgrade our data gathering and analysis to track trends in patent system usage and ensure we’re ready for what comes next.
New AI-based services: threat or opportunity?
Everyone in the profession is monitoring new services that are being launched in this field. For example, Iprova is a company that uses machine learning to invent new products and services, while TurboPatent and Specifio are two US companies that offer patent-drafting services based on AI. Other start-ups, such as IP.we, are exploring the potential of Blockchain for tracking IP ownership, licensing or prosecution history (maybe one day, it will no longer be true to say, as one former judge did, that “life is too short” to review the entire file history).
Some practitioners see the development of such services as an existential threat, but IP Practice 3.0 does not mean that attorneys will be replaced but that they will have to adapt, just as they have done in the past. Greater automation of routine tasks should deliver more efficiency and free practitioners up to do the more skilled and valuable work that they have been trained for. As we do so, we will also benefit from the insight of professionals in different areas – such as software developers and project managers.
Beyond IP 3.0
IP 3.0 is just around the corner. But some practitioners are already thinking about IP 4.0, in particular asking: Will we see AI tools drafting, prosecuting and granting patent applications, perhaps for inventions that have themselves been created by artificial intelligence and which are targeted at consumers that are themselves using AI, such as virtual assistants? Such a scenario would require considerable regulatory and organizational change, which suggests it is a long way off. But we should remain open to new tools and new ways of working, addressing questions such as: Should we change how we prosecute patents? Are there better ways of conveying information about an invention than using claims, a description and drawings? Do we have the balance between words and images correct? How can computer-based tools improve the quality and reliability of the work we all do? And how will the profession change when computers can invent and examine?
Nigel Hanley predicts that these challenges will raise both philosophical and policy questions. For example on questions of ownership and entitlement, he says: “Can a machine own or invent? Who owns an invention – the person who owns the machine, the one who built the machine or the one who programmed it? These will not come into view in the short to medium term, but they certainly will in the longer term.” There are also questions about trust – who writes the software? – and ethical standards, particularly where technology is being developed using common standards and spanning different jurisdictions.
Dr Shemtov argues that the increased use of AI in research could ultimately lead to challenges to established terms such as “person skilled in the art”, depending on the direction taken by developments in the science that underpins AI technology. He says: “Within about half a century, we may have to consider whether the benchmark will instead be a ‘machine skilled in the art’. How can you judge if something is obvious to a person when a person is not involved in the invention?” He adds that similar questions are being debated in copyright law. The first AI-generated portrait was auctioned in December last year, and AI is also being used in music and design. “In most countries, this presents a serious issue as the originality threshold in copyright law is arguably not met,” says Dr Shemtov. “My hunch is that there will be movement in the law there – you need to promote creativity. And history teaches us that the law is always playing catch-up with technology.”
If you would like to have a word with Gwilym or Nigel on the subject of AI or other topics related to your Intellectual Property, they would be delighted to hear from you. Contact us at firstname.lastname@example.org