LSPN Connect: artificial intelligence and IP
IP Australia introduces machine learning for TM examiners
Andrey Popov / Shutterstock.com
Artificial intelligence is reinventing trademark tasks and leaving the lawyering to the lawyers, says Charles Hill of TrademarkNow.
To judge from business book bestseller lists and recent popular conference topics, the question ‘how can artificial intelligence (AI) improve the way we work?’ has become a major topic of interest in today’s economy—and in the field of IP law, too.
Clear trends are now emerging on how lawyers should be adopting these technologies to improve performance—and success stories abound. For example, many trademark professionals report having used AI to completely reimagine their traditional work processes, with positive effects.
Human-machine best practices
Today’s programmers are using many different AI techniques—including expert systems, machine learning, deep learning neural networks, natural language processing, and voice and image recognition—to drive better business performance. AI is evolving very quickly and proving very valuable, across many different industries and in the field of law.
How can IP law leaders best derive value from these new technologies? AI consultants teach that we should view human-machine interactions as a form of symbiosis, where each side focuses on its own strengths, and each side is ‘augmented’ by the other.
Humans gain ‘superpowers’ in fast data analysis and repetitive tasks, while machines learn from human training and guidance.
"Leading-edge legal teams are similarly finding that they can deliver better results by using AI software to work much more flexibly."
This symbiosis works best via a human-machine division of labour rather than full automation of human tasks. Essentially, we should aim to leave repetitive tasks and routine cases to machines, while leaving the ‘harder stuff’—resolution of ambiguous information, handling of customer services and other nuanced social skills, and the exercise of judgement in tough cases—with human professionals.
This kind of partnership is driving better performance across virtually every industry, such as improved agricultural yields, medical diagnoses, stock picking, auto manufacturing, and yes, IP law, driving what some pundits have described as ‘the fourth industrial revolution’, referring to (1) mechanisation and steam power; (2) electricity and mass production; (3) electronics and automation; and (4) AI and robotic systems.
AI is being used not just to drive efficiencies in existing process flows, it has also allowed leaders to reimagine such ways of working entirely. For example, auto manufacturers have reinvented the traditional assembly line over the last decade, using machine learning ‘cobots’ that workers can move around the manufacturing floor on the fly. This allows a far greater customisation of cars coming off the line, using real-time information flowing in from dealerships around the globe.
In our own industry, leading-edge legal teams are similarly finding that they can deliver better results by using AI software to work much more flexibly.
The traditional process of clearing a new trademark, for example, can be thought of as similar to an assembly line—a series of discrete steps done over and over again to deliver a service (ie, safely launching a new brand) (Figure 1).
First, brainstorming generates a pile of candidate trademark names, which a legal team ‘screens’ to quickly and cheaply knock out obvious problems. ‘Good’ candidates are then thoroughly researched for similarity conflicts in a ‘full search’. Finally, senior people review the results of the searches and give their opinion on whether to greenlight the brand for launch or filing.
The worst thing that can happen here is launching a brand that has conflicts that weren’t caught. But the next-worst outcome is ‘rework’—cases where at the end of all of the screening and searching, all good candidates have been eliminated, and the team has to go back to the drawing board, driving up both time and expense.
Reinventing how lawyers work
Using better tools for both search steps within the exact same process can help, speeding these searches up, catching more conflicts earlier, and preventing expensive outsourced ‘full’ searches and ‘rework’ later.
Many of TrademarkNow’s clients report that they’ve simply added our AI to their existing way of working, with positive effects.
Other TrademarkNow clients have experimented with wholesale changes to the process. More than a quarter (28%) are now sharing direct use of our trademark search platform with marketing and creative groups. Benefits include “getting better candidates from marketing right out of the gate”, and improving ties between groups: “once our branding department found out that we (legal) could deliver search results faster, they were more willing to send us proposed names for products”.
Some clients report that they are eliminating screening and full search as separate process steps altogether, eg, by using fast AI similarity-search clearance software in the screening phase itself. This can result in faster resolution of product and marketing teams’ questions around the potential viability of candidate brands, which in turn can prevent those groups’ becoming wedded to certain ‘dream candidates’ during the process. As one corporate legal professional told us,
“I don’t ever want businesspeople to fall in love with a name, so I usually just go straight to the clearance search”.
The results of all this are often very impressive. As Mariya Yao, Marlene Jia and Adelyn Zhou report in their 2017 book “Applied Artificial Intelligence”, IBM (a TrademarkNow client since 2014) has “found the use of AI has cut down on the total time needed to analyse trademark search results by 50%”.
Thoughts on the future
Many AI experts agree that fears of large-scale AI-driven job losses in the legal industry are overblown. We agree: while machines will replace much manual labour in data-intensive process steps, they will continue to help paralegals and attorneys to do less ‘grunt work’ and more actual lawyering, customer service and sales. While workflows, like that of trademark search, will continue to speed up, human (lawyer) opinions will continue to be needed.
We predict that AI will enable in-house counsel to become a more responsive and effective partner to the rest of the business; joint marketing and legal processes will become tighter; corporations will be able to watch more of their marks and oppose more threats; and law firms will be able to serve a greater number of mid-size clients than ever before.
All in all, the future looks bright for firms who are embracing new technology. AI will do more good than harm for lawyers over at least the next decade.
Charles Hill is head of product at TrademarkNow. He is a product management professional with over 20 years’ experience managing large scale software development and product launches. He can be contacted at: firstname.lastname@example.org
TrademarkNow, AI, trademarks, machine learning, neural networks, data analysts, automation, medical diagnoses, IP law, manufacturers, in-house counsel, IBM