Automated data analysis has changed how scientists think about health research—and is helping uncover inefficiencies in transport systems. Now it’s making waves in the legal sector.
In January, prosecutors in Germany presented a court with data from an iPhone suggesting a suspect was climbing stairs when it was alleged he had been dragging a body down a steep bank. It isn’t the only case of its kind. Pacemaker data has been used in US courts as evidence against a wearer and data from fitness trackers has been used to indicate how, for example, a person’s lifestyle has changed after an injury.
The amount of data being created by the devices people carry around has grown exponentially— and had a knock-on impact across multiple professions. For criminal lawyers, it has increased the potential evidence that’s available, and this increase in data has been mirrored across all branches of law. It’s a trend that has the potential to completely disrupt the legal profession. “The biggest impact”, says James Rogers, a solicitor at Norton Rose Fulbright, “will not be caused by big data alone—rather it will be the combination of big data with other innovations, such as artificial intelligence (AI) or distributed ledger technologies”.
Data created through the legal process is already being digitised and combined with other technology. It has the potential to make the jobs of lawyers more straightforward. “We will be able to find pertinent cases more easily through better recommender systems that leverage machine learning”, says Wolfgang Alschner, an Assistant Professor in common law at the University of Ottawa. “Another area is document and compliance review, where computers can find patterns in large amounts of texts more reliably and quickly.”
Everlaw, a legal tech startup that has raised more than $25m from Silicon Valley investors, is applying AI to historic legal reports to help lawyers prioritise and analyse documents. It is using computer science to visualise data from documents and predict the most relevant cases for lawyers to inspect.
For such a system to analyse court cases and predict decisions, it needs data. But it must be the right type of data. It needs to be easily read and understood by machines—messy handwriting can be tough to interpret, but clear check boxes on forms are easier for software to understand. If big data is machine readable, then AI systems can save the legal industry money and time.
Rogers says it will initially have the greatest impact on areas that are “relatively formulaic and process or volume-driven”. This includes disclosure, verification, first-tier due diligence review and similar processes. However, AI won’t be using data to formulate complex adversarial arguments at any point soon. “Human elements: intuition, judgment, emotion, memory, perception and so on, are all integral to that process”, Rogers says.
It’s already been proven that data can be put to work to speed up lengthy legal processes. In one instance, the UK’s Serious Fraud Office (SFO) found itself with more than 30m documents to examine during a corruption investigation into Rolls-Royce.
“Typically, this would have involved contracting independent lawyers to review potentially legal professional privilege documents individually—a lengthy and expensive process”, says Peter Wallqvist, the VP of Strategy at RAVN, a startup owned by document management company iManage. The firm set its machine-learning system to work processing the documents. It sorted more than 600,000 documents per day, which Wallqvist claims is 2,000 times faster than a lawyer.
“The program was able to interpret and understand material from a range of sources, including emails, financial records, data tables and photographs—but, unlike a human team, it could work efficiently for days on end without interruption,” Wallqvist says. The SFO won the case—its biggest investigation—and Rolls-Royce was ordered by a court to pay £497m, plus the costs of the SFO’s investigation.
For the moment, AI and analytical platforms create an advantage for legal firms and clients that can afford to pay for it. But it also has the potential to democratise the sector. “It will make legal services more accessible to lay people by driving down costs and by fuelling the development of self-help applications directly targeted at non-lawyers,” Alschner says. “All this will transform the role of the lawyer and their day-to-day legal practice.”
Stanford law student Joshua Browder has been pivotal in improving how legal data can be combined with technology to make the law easier to comprehend. He created a chat-bot ‘lawyer’, DoNotPay, which allows users to contest parking tickets they have received. Browder says the system talks people through the legal steps required to make a complaint and automatically files their issue. It has won more than 400,000 cases. The DoNotPay system has since been expanded to tackle legal loopholes used when flights are booked, to help customers get automatic refunds.
But there are some concerns around the use of big data. Data scientists, across multiple fields, have shown that automated systems inherit any bias that’s contained in the data they are trained upon. In some cases, AI has replicated racist and sexist language it has seen in data sets. Alschner argues that legal data needs to be freely accessible, while also ensuring that privacy laws are respected.
“Law is a public good,” Alschner says. “But most legal texts, from court decisions to filings, are only accessible through commercial service providers that do not generally share data in bulk.” The solution to the problem? “Large legal information providers can lead the way by sharing more of their data with academics, governments and legal analytics companies,” Alschner adds.