Artificial intelligence (AI), also termed as machine learning, is a feature of technology that is becoming increasingly important in the legal sector and has attracted significant attention in the last year. Firms are increasingly becoming aware of investing in innovative technologies to keep themselves relevant and to upgrade their offerings to their ever-demanding clients across the globe.
As a result, firms like Baker & Hostetler, Clifford Chance, DLA Piper, Latham & Watkins, Linklaters, Slaughter and May and Wachtell, Lipton, Rosen & Katz have all started experimenting with AI in workﬂow processes in their firms. Dentons, for example, is working with ROSS Intelligence, RAVN Systems and Beagle through Nextlaw Labs to explore the application of AI in real-world legal work.
BRAVE NEW WORLD
The most common use for AI in the legal field occurs in the due diligence phase. This includes expert legal research, contract review and e-discovery.
”We are delighted to be working with leading data scientists on using AI and machine learning in transactional due diligence. This year has also seen the uptake of technology-assisted review in e-disclosure,” says Jane Bradbury, head of knowledge and information at Slaughter and May.
In the legal space, AI synthesises massive amounts of unstructured data like legal documents or cases, identify trends and specific pieces of information which are most relevant to lawyers, and becomes "smarter" by learning the preferences of its user and receiving feedback on the quality of its results from experts.
Some of the AI technology is still in the early stages but experts believe that machine-learning tools in the legal sector will appear in three waves.
The first wave is in tools that can form a good view of what a document is about. “These can be mainly seen in the growing number of advanced e-discovery tools in the litigation and investigation area. These e-discovery tools can analyse millions of documents and can, for instance, replace the first level of review to decide which documents might be relevant to the case. This can result in large cost savings compared to the traditional way of reviewing these documents. This has permanently changed the way investigations are being done,” adds Bas Boris Visser, the global head of innovation and business change at Clifford Chance.
The second wave is in tools that can dig deeper inside the documents and understand the structure and then can find and extract specific data or provisions.
Finally, the third wave is in tools that can understand natural language and supposed to be able to reason like a human being. The best known example is IBM's Watson. “What I expect IBM wants Watson to be able to do is to function like a paralegal or junior lawyer,” Visser says.
The key AI-related technologies currently being used by lawyers fall under the categories of machine learning, natural language processing, cognitive computing and robotics.
Machine learning involves identification of trends in data through supervised and unsupervised learning. The natural language processing technology aids in creation of meaningful text language from structured data, or analysis of text based on the interpretation of emotion or meaning.
The cognitive computing type predicts human thoughts and processes within a computerized model. Robotics is the automation of repetitive manual tasks such as data extraction and entry through manipulating computer software.
Dentons-supported Nextlaw Labs' first portfolio company ROSS Intelligence has developed a legal research app called ROSS, powered by IBM’s cognitive computer Watson, which uses machine learning and natural language processing to assist with contract drafting and review, allowing lawyers to do their jobs more efficiently.
WILL LAWYERS BE REPLACED?
The use of all these technologies obviously enhances client offerings, as speed and accuracy can be improved.
“In some matters today, a lot of time is spent finding the right information and reading through contracts, some of which may not be relevant. Technology will help ensure the quality of the data that is being reviewed and considered. We should be clear that we don’t see these technologies as replacing lawyers. These tools will give the lawyers access to a better pool of information on which they can use their experience and judgment to offer advice and solutions,” says Matt Peers, the director of information systems and strategy at Linklaters.
Using AI, for example, in due diligence, modifies certain practical steps in the due diligence process, rather than the legal analysis itself. “It sits within a workﬂow process that is designed to assist lawyers in spotting issues and allocating tasks. The software can process and classify large data sets quickly, but is not a replacement for humans. Rather it helps to sort and prioritise documents for expert review by lawyers, not robots, and speeds up the time taken to spot potential issues. The aim is to allow lawyers to use their time as effectively as possible to identify and analyse critical legal issues and communicate these issues to the client in the early stages of a transaction,” explains Bradbury.
Nextlaw Labs’ CEO Dan Jansen says that currently firms are not at the stage where they can completely hand off tasks over to computers for processing and trust the results would completely be accurate.
“It takes years of expert guidance and training the machine before that can happen. However, the benefit and promise of AI in legal field is that the repetitive and time-consuming tasks such as legal research and document review will be significantly reduced, allowing lawyers to direct their time and attention to more important work and ultimately providing significant cost savings to clients,” he adds.
Andrew Cooper, the regional IT manager and head of business infrastructure, Asia, at DLA Piper points that law firms need to think carefully about how far they push the idea of using AI tools as value-added innovation.
Firms today are running controlled pilots by reworking old matters with technology to understand what would be different if the technology was used in the delivery. “The challenge is in choosing technologies that you can work with,” says Peers.
One of the big barriers to widespread adoption is that many of the technologies don’t work out of the box. “They need specific lawyer time to set them up, configure for different clients and applications and show them how to learn. Until there is a compelling case and certainty about the products that will be taken forward, it is hard to secure lawyer time to do this work,” Peers adds.
Jansen comments that the legal industry as a whole tends to be resistant to change, and when it comes to the subject of AI, there is resistance stemming from the fear that robots will eventually take over lawyers' jobs.
“But this is simply not the case,” he says. “Technology is about augmenting and amplifying the intelligence of lawyers already in their jobs to help them complete their tasks more efficiently and focus on higher value work and improve their client service. It is about man plus machine not machine replacing man. Clients are demanding innovation and we are seeing the market respond to this demand with an increasing amount of interest in AI among some of the major law firms.”
Clifford is using software tools like Kira Systems and Leverton on real projects. Visser says, “We are piloting it broadly within the firm. Due to, for instance, language restrictions or specific features in certain jurisdictions, there are certain areas or jurisdictions where it might take longer for it to be more fully implementable.”
FIRMS AT RISK?
Early adopters face the challenges and opportunities of helping shape the development of AI in law. This requires serious investment of time, energy and resources. Machine learning requires expert guidance and constant quality feedback, but it is worth the investment.
AI in the legal sector also pose challenges around how and where data is stored, how products are priced, how risk profile changes for clients and law firms, and how organisations get the right skills mix to be able to deliver work using a range of technologies.
The traditional matter delivery model has evolved over many generations and is well established. “Changing this involves taking some risks on the side of firms and clients, and in the delivery of most matters time is important, and hence no one wants to lose time or introduce additional steps,” adds Peers.
On similar lines, Bradbury says time is needed to “train” the software to recognise the key legal clause and document types. “The more time that is invested in training the software, the better the results should become,” she comments.
Some lawyers may not favour the idea that software can do the job of a lawyer for obvious reasons, but at the moment technology isn’t doing that. “It is really only reducing the amount of time spent on the lower-end work, for example, finding provisions in a set of contract documents. It doesn’t take away the need for the human legal expertise to be applied to the significance of those clauses,” says Cooper.
According to law firms, clients are not only positively keen to see innovative uses of AI technology in the conduct of their work by law firms, but in-house client contacts are also interested in technology that can support research or contract review in their own legal teams. Says Peers: “Clients will hopefully work closely with law firms to see where mutual benefits can be found from using technology. It is likely to involve an investment and leap of faith on both sides.”
Clients are equally excited at the prospect of seeing higher quality work delivered much more quickly, thus reducing their amount of legal spend. “They are excited at the idea of receiving AI-powered tailored solutions which directly address their needs,” Jansen says.
In the future, law firms expect to see some very interesting use cases emerging. “From an industry-wide perspective, the rise of AI and legal tech will change the conversations partners have with their clients and law professors with law students.
AI will become a more commonly requested feature in clients request for proposals (RFPs) and a more prevalent subject taught in law school classrooms,” says Jansen.
Some of the technologies will hit the mainstream very quickly, for example, the ability to ingest lots of contracts, group them and make them available for electronic search is likely to be common in the next couple of years.
Already a number of software companies and law firms are working to get products to market quickly.
SIRI, CATCH MARKET CHEATS: WALL STREET WATCHDOGS TURN TO AI
Artificial intelligence programs have beaten chess masters and TV quiz show champions. Next up: stock market cheats.
Two exchange operators have announced plans to launch artificial intelligence tools for market surveillance and officials at a Wall Street regulator tell Reuters they are not far behind. Executives are hoping computers with humanoid wit can help mere mortals catch misbehaviour more quickly.
The software could, for instance, scrub chat-room messages to detect dubious bragging or back slapping around the time of a big trade. It could also more quickly unravel complex issues, like "layering," where orders are rapidly sent to exchanges and then canceled to artificially move a stock price.
A.I. may even sniff out new types of chicanery, says Tom Gira, executive vice president for market regulation at the Financial Industry Regulatory Authority (FINRA).
"The biggest concern we have is that there is some manipulative scheme that we are not even aware of," he notes. "It seems like these tools have the potential to give us a better window into the market for those types of scenarios."
FINRA plans to test artificial intelligence software being developed in-house for surveillance this year, while Nasdaq and the London Stock Exchange are close to utilising it.
The exchange operators also plan to sell the technology to banks and fund managers, so that they can monitor their traders.
Artificial intelligence is the notion that computers can imitate nuanced human behaviour, like understanding language, solving puzzles or even diagnosing diseases. It has been in development since the 1950s and is now used in some mainstream ways, like Siri, an application on Apple Inc's iPhone that can engage in conversation and perform tasks.
While financial firms are already applying artificial intelligence software for everything from compliance to stock-picking, it is only starting to become useful for market oversight.
"We haven't really let the machines loose, as it were, on the surveillance side," says Bill Nosal, a Nasdaq business development executive who is overseeing its artificial intelligence effort.
Market surveillance generally relies on algorithms to detect patterns in trading data that may signal manipulation and prompt staff to investigate.
But the sheer volume of data can lead to an overwhelming number of alerts, many of which are false alarms. FINRA monitors roughly 50 billion market "events" a day, including stock orders, modifications, cancellations and trades. It looks for around 270 patterns to uncover potential rule violations. It would not say how many events are ﬂagged, or how many of those yield evidence of misbehaviour.
The "machine learning" software it is developing will be able to look beyond those set patterns and understand which situations truly warrant red ﬂags, says Gira.
Machine learning is a subset of artificial intelligence in which computers figure out new tasks without having been programmed to do so. In the case of market surveillance, that would mean the computers "learn" which trading patterns lead to enforcement charges, in order to ﬂag the right ones.
FINRA plans to test the new tool next year alongside its existing systems to compare the results. The regulator has already moved its surveillance systems to Amazon.com Inc's web-based Cloud, giving it more computing power to quickly analyse massive data.
Nasdaq is working with cognitive computing firm Digital Reasoning, which it invested in earlier this year. LSE has teamed up with International Business Machine Corp's Watson business and cybersecurity firm SparkCognition to develop its A.I.-enhanced surveillance, Chris Corrado, chief operating officer of LSE, says. Watson has become something of a household name, having bested contestants in the game show "Jeopardy" in 2011.
The technology would not necessarily prevent events such as the 2010 "ﬂash crash," when the Dow Jones Industrial Average temporarily plunged more than 1,000 points. – John McCrank, Reuters