AI is here to help lawyers- Not Replace Them

We were flattered when a recent BBC News article about legal technology included our company TrademarkNow. However, we were somewhat surprised that it questioned the use of AI for legal technology.

There are pitfalls with this approach in such a heavily regulated sector according to the article, and it raised the controversial issue of AI-driven technology offering legal advice - a big taboo and, in most jurisdictions, restriction in the legal field.

Mikael Kolehmainen, our CEO and co-founder, is a former practicing trademark attorney, so we are very familiar with the intricacies of designing software that offers legal advice vs. software that provides a much-needed, powerful digital tool to help lawyers productivity and efficiency. Our products have always focused on the latter, and we have always done our best to be as clear as possible about this. We do not give legal advice, we just help trademark attorneys and in-house counsel by providing them with the information they need, in a complete and organized manner, to help them make decisions and give advice more efficiently.

Why does artificial intelligence and machine learning often attract negative press attention? The prospect of sentient machines, a science fiction trope only a few years ago, is frightening to many people, and it certainly makes for a good story. In my opinion, however, that story is extremely unlikely to become true, and the main reason for that is the fundamental difference in information processing in humans versus computers: the former is fundamentally analog while the latter is digital. As a consequence, people are (at least comparatively speaking) good at handling completely novel situations with imperfect information. Coincidentally, this is also what allows trademark law to work reasonably well with a legal standard of trademark similarity consisting basically of just a single sentence in a statute, accompanied by a vast body of case law. It is just that finding the relevant information is slow and unreliable when done by humans alone.

Admittedly, there will be huge disruptions in certain industries sooner rather than later thanks to advances in automation technology. Many people still scoff at Google's, and other companies working on self-driving cars, however driverless vehicles are inevitable. That change in transportation will be transformative, possibly starting with commercial trucking before consumer vehicles, and while it may result in the loss of jobs, the biggest net result will be countless lives saved when a major part of our transportation system will no longer be subject to the perceptual capabilities and risk acceptance of human drivers acting as individuals.

Machine learning and AI are being used in other heavily regulated fields as well - in ways that machines aren't offering expert advice. In medicine, Lumiata uses big data and graph analysis to connect 160 million data points from textbooks, journal articles, and public data sets to offer physicians 'what if' questions versus the software itself making a diagnosis. In finance, Feedzai uses machine learning to detect payment fraud for credit card/debit transactions by studying millions of transactions and learning what fraud signals to pay attention to.

In the legal industry, Lex Machina has been referred to as "Moneyball for lawyers" instead of focusing on the merits of a patent case - something lawyers are uniquely qualified to study and determine - Lex Machina analyzes and ranks a wide variety of aspects of a patent case, including the patent owner's track record in terms of litigation frequency, likelihood of settling, duration of the process and the distribution of outcomes as well as the assigned judges tendency to rule in favor of the defendant or the plaintiff. Lex Machina's analysis is valuable information for patent trial strategy.

A significant task of every trademark attorney is to conduct extensive research looking for possible issues with trademarks that companies are considering filing or using. That research has traditionally been a very manual, messy, and sometimes imprecise process. It can (or should) include searches of legal databases, extensive internet searches, searching dictionaries and thesauruses in multiple languages to look at word and phrase meanings, searching local and international e-marketplaces and industry in-use databases. What we've done at TrademarkNow is automate that search process and return results in a matter of seconds using a sophisticated algorithm based on models of trademark law and linguistics to prioritize and rank the most relevant results for lawyers to consider and analyze in developing advice for their internal teams and clients. On top of this, our clients provide feedback on results, so our platform is always learning and improving, based on real feedback from attorneys.

While FUD and science fictional horror stories about the rise of robots will continue to make great headlines, that is not the day-to-day reality of the companies in multiple industries that are building machine learning and AI software platforms. Instead, those companies are harnessing the power of technology to build digital solutions to often complex problems and processes, whether it's global trademark searches or sifting millions of data points to aid doctors in making correct patient diagnoses. Machine learning and AI will be an integral part of the future we'll all be living in, and we are happy to be a part of it!

by Mikael Kolehmainen.