There is a lot of buzz around the potential for AI in patent practice, and speculation about what it could do. However, in the short term the biggest changes are likely to come from automation, rather than AI as strictly defined.
- Many tasks traditionally done by patent practitioners will be automated
- Both clients and IP offices will seek efficiencies from automation
- Within the next decade, AI and blockchain will be playing a role in patent practice
There is a lot of buzz around the potential for AI in patent practice, and speculation about what it could do. However, in the short term the biggest changes are likely to come from automation, rather than AI as strictly defined. Increasingly, routine tasks such as filling in forms, checking dates, titles and categorisation, and sending instructions or reminders will be carried out or assisted by computers. This is already established in some respects – notably managing annuities – but it will spread much further in the coming decade, driven by technologies such as blockchain as well as the growing value of Big Data.
Both clients and IP offices will promote this automation, as it leads to cost savings and (if implemented correctly) cuts down on errors. In addition, it will help to generate and analyse the vast amounts of data contained in the patent system, making it easier for companies to assess risks and opportunities. The implications of this data explosion, in particular for the way patent portfolios are managed and valued, are explored further in our report on the future of in-house patent practice.
For patent practitioners in private practice, the impact of automation will be profound. Most obviously, they will need to adopt new tools into their existing work practices. This will inevitably cause some disruption: it will require investment, either in developing bespoke systems or licensing-in software developed by other service providers. There will also be a need to integrate systems with those of key clients. As discussed below, increasingly firms will develop or even acquire standalone technology businesses to meet clients’ needs.
Above all, automation poses a threat to existing working practices and some roles. To begin with, this threat will be focused on the lower end of practice – patent administrators and paralegals, as well as trainees. “Some people who are using quasi-legal knowledge on manual tasks will be replaced,” says one practitioner. Freed from spending time on administration, these administrators will need to identify different roles where they can add value. “If I were a paralegal, I would be worried that my job could disappear quite soon,” states one practitioner.
But people should not underestimate the extent to which automation will affect all roles in the patent workflow. The 2017 launch of Specifio, which promises to create near-complete patent applications based on claims drafted by a practitioner, shows how AI has the potential to replace a large proportion of drafting work. Routine work around writing licences and contracts, as well as managing the copious records and deadlines in litigation, can also easily be automated. “It doesn’t need to be perfect: 80% is good enough for most purposes and saves huge amounts of time and cost”, as one IP provider says.
Machine learning will also soon be used as part of corporate technology and IP policies to measure threats and opportunities. As one in-house counsel contends: “AI will take off. It helps maintenance decisions, makes filing and prosecution strategy easier and aids decisions about when to litigate and what the risk is in contracts. We will see that in the next five years.” In particular, companies that provide patent analytics services are developing AI tools to provide these kinds of service – a trend that is further discussed in our report on patent service providers.
It is easy for IP legal practitioners to see such developments as a threat. Fearing the worst, many ask whether a patent application could ever be drafted and examined without any human involvement. The short answer is that the software to do that will surely exist soon (if it does not already), but it is likely be a long time before it is implemented due to the investment required, the risk profile of patent applicants and the slow pace of legislative change.
In that sense, existential questions about whether computers will ever be able to file patents are a red herring. The question that every practitioner should really be asking is how can automation and AI make me a better adviser to my clients and how can they help me to deliver solutions to the challenges I face?