How new technologies are reshaping the law
This article is part of the sixth annual special Technology & Law edition of LawNews, put together by ADLS’ Technology & Law committee.
Media regularly produce lists of the jobs most and least likely to be automated by robotics and artificial intelligence, but not many of these feature the legal profession.
That could be about to change.
MinterEllisonRuddWatts has deployed an AI tool in the form of a Microsoft Word plug-in that it helped fund. The tool, AuthorDocs, uses neuro-linguistic programming and machine learning to speed and improve the laborious contract review process.
At the global launch of the product in Auckland last week, the tool’s New Zealand developer McCarthyFinch indicated it was preparing several more technologies for rollout, some of which could have a more direct impact on jobs than AuthorDocs.
It seems intuitive that it will be a long time before a robot can carry a complex argument in court or design creative new business structures.
But then you might once have argued that robots could not deliver empathy and human engagement, something being fast disproved by avatar technologies from the likes of New Zealand’s own Soul Machines.
Investment in legal tech platforms reached US$1 billion last year across 40 identified deals, according to a report by analysts Tracxn Technologies. Of that, US$362 million was invested into legal software using AI.
The tasks most suited to automation with current commercial AI are those involving repetition or research with high levels of accuracy, said Josh Comrie, the founder of New Zealand conversational intelligence technology developer Ambit AI.
The legal profession has many such functions, from research to conveyancing and document preparation and checking.
“Many of these tasks are far better suited to either a process or an algorithm,” Comrie said. “In my opinion, the holy grail for law is a detailed and actual cognitive insight into what a client wants to achieve and then how to go about representing this in some form of outcome.
“While this may eventually occur, the real, realisable current benefit for the profession is in removing the low-level heavy lifting of junior solicitors.”
This, however, would pose a long-term challenge to the profession; that is the kind of work experience that produces senior lawyers.
The development of AI and robotics offers opportunities as well as threats.
Digitally-progressive law firms can prosper by developing and deploying technology to disrupt their own businesses and the industry through improved efficiency and accuracy as well as automation.
As with other industries and professions, the jobs – or, perhaps more accurately, the “tasks” – that will fall to the robots first will be very specific.
The legal discovery process, for example, is already being boosted by smart technologies such as natural language processing.
But some early tests are showing robots can achieve better consistency and accuracy than humans in relatively complex tasks, especially when a lot of written material needs to be processed.
Last February, legal automation company LawGeex released a report comparing the performance of 20 experienced United Nations lawyers to its AI system.
“Few would be surprised that artificial intelligence works faster than lawyers on certain noncore legal tasks,” the report said. “However, lawyers and the public generally believe machines cannot match human intellect for accuracy in daily fundamental legal work.”
In a test of a daily legal risk assessment task performed on contracts from the Enron dataset, however, human lawyers achieved an average performance accuracy rating of 85%, while the average of AI tool was 94%.
Almost as pertinent, the average time human lawyers required to complete the process was 92 minutes, while the AI system turned out its results in 26 seconds.
Perhaps the most notable thing about this test and several others is that it did not focus on mundane tasks or commoditised legal work, but on the core function of legal advice.
The core technology used was proprietary legal language processing (LLP) and understanding models that learn “legalese”.
“The LLP technology allows the algorithm to identify these concepts even if they were worded in ways never seen before,” the report explained.
Lowndes Jordan partner and technology law specialist Rick Shera also said it would likely be the low-hanging fruit – commoditised, easily-repeatable tasks – that will be impacted first.
Strategy work such as anticipating the other side’s responses and positions are much less likely to be automated.
“Those are decisions that humans will have to make and may never be replaced by AI,” Shera said.
Tasks such as trademark applications can be automated through database integration. The intellectual property office, IPONZ, has already rolled out a beta tool allowing people to make their own inquiries about desired trademarks.
As with any new technology, AI adoption could be slowed because it is always a challenge to move people to new ways of working, Shera said.
However, even law firms that don’t outwardly adopt AI could find themselves using it because many of the profession’s software vendors are starting to build it into their products by way of upgrade. AI, for instance, is already embedded and being widely (if largely invisibly) used as part of common email and other desktop tools.
That could mean most businesses won’t actually have control of their own AI technologies, Shera said, potentially raising more concerns and trust issues about who has access to, and use of, personal and confidential data.
Shera said other automation technologies such as document management are likely to have as big an impact on legal operations as AI in the short term.
At the launch of AuthorDocs, MinterEllisonRuddWatts CEO Andrew Poole said after investing more than $1 million in the company and a huge amount of time, he believed the tool would deliver immediate benefits to all lawyers, but particularly in-house lawyers and those in private practice, enabling them to work smarter and faster while focusing on quality.
“Our firm had a choice about whether we would be disrupted or be part of that disruption and we are much happier to be part of the disruption,” Poole said.
“We recognise that change is coming. We’re not quite sure what the magnitude of that change will be in legal services, but there is no doubt change is coming.”
Poole said as well as wanting to service clients better and see a return on its investment, the firm wanted to be part of AI to understand it and not to be a bystander.
“We don’t know whether AI is going to revolutionise the practice of law. Some people say it will. But certainly this tool, if not revolutionising it will certainly be of huge benefit to lawyers as a whole.”
James Schellhase, CEO of McCarthyFinch, said AuthorDocs was designed for every day, not heavy-projects use, within Microsoft Word “where lawyers work” and where engagement with the client also happens.
AuthorDocs was built to deliver “time to value” and requires no training, he said, but it was also a gateway product before McCarthyFinch progresses into the more advanced AI solutions it would soon be offering.
These include a contract approval and workflow automation product, another automating the drafting of contracts, and yet another that extracts key information from masses of documents.
McCarthyFinch is far from alone in its efforts. AI is breaking out all over.
Many of these tools are those lawyers can use rather than tools designed to replace lawyers.
Canada-based Blue J Legal, for instance, is developing an AI system that can help tax professionals gauge the strength of their legal position by applying AI to previous judicial decisions and findings.
That said, it seems likely in the medium to longer term fewer lawyers will be required as a result of AI, especially at entry level and in areas with repetitive, lower-level work.
Rob O’Neill is a freelance journalist specialising in technology