How Artificial Intelligence Is Improving the Justice System

Saving time and money in document management, but with the added benefit of identifying, interpreting and then projecting conclusions which may be able to assess guilt and innocence in criminal cases or the veracity and utility of historic data and projections in civil cases. JL

Rowland Manthorpe reports in Wired:

Barristers were wading through 3,000 documents a day. RAVN processed 600,000 daily - with fewer errors than the lawyers.“Where someone has scanned a 300-page document, it's not uncommon to put one in upside down by mistake. We need to deal with that real world of messy datasets. It cut out 80% of the work. It also saved us a lot of money." Using data buried in documents, its system could suggest likely outcomes of mergers and acquisitions, or true estimates of project costs.
The Serious Fraud Office had a problem. Its investigation into corruption at Rolls-Royce was inching towards a conclusion, but four years of digging had produced a massive pile of documents: over 30 million, including everything from spreadsheets to emails about staff away days. Now these needed to be sorted into “privileged” and “non-privileged”, a process that meant paying junior barristers for months of dull, repetitive paperwork.
“We needed a way that was faster,” says Ben Denison, chief technology officer at the Serious Fraud Office. So, in January 2016, the SFO started working with RAVN. East London startup RAVN (pronounced Raven) builds robots that sift and sort data. Hardly unusual – but its system, which mixes applied AI techniques such as computer vision with more conventional database management, can digest not only neatly presented material, but also information that comes in a less structured format. “We have things like an upside down detector,” says co-founder Peter Wallqvist. “Where someone has scanned a 300-page document, it's not uncommon to put one in upside down by mistake. We need to deal with that real world of messy datasets."
The two teams started to feed material from the Rolls-Royce case into the AI. By July they had trained an algorithm, and with the agreement of lawyers on both sides, they set the robot to work. The barristers were wading through 3,000 documents a day. RAVN processed 600,000 daily, at a cost of £50,000 - with fewer errors than the lawyers."It cut out 80 per cent of the work," says Denison. "It also saved us a lot of money." For Rolls-Royce, it had the opposite effect. In January 2017, the engineering company admitted to "vast, endemic" bribery and paid a £671 million fine. "It's hard to imagine a better outcome," says Wallqvist. “It's a good trend that governments are brave enough to pull the trigger on things like this.”For RAVN, the trend has been good for some time. Its co-founders – Wallqvist, Jan Van Hoecke, Simon Pecovnik and Sjoerd Smeets – met at Autonomy, the first UK unicorn, where they worked together on early versions of AI-powered database management. But by 2009 they were growing frustrated. “We felt innovation had stagnated,” says Wallqvist. “There seemed to be very little interest in taking that next leap.” So, the next year, the four took it themselves, using their Autonomy contacts to get RAVN off the ground.
Today, the self-funded firm has fifty-one staff, revenues of three million, and around seventy regular clients, mainly city law firms. One satisfied customer: BT, which recently signed a “very significant” deal, and credits RAVN with annual savings of £100 million, thanks to automated checks that ensure its contracts with suppliers are accurate.

Plus, of course, there's the SFO, which is using RAVN in increasingly sophisticated ways. That means allowing it to make subjective judgements, such as pointing investigators to data it thinks is relevant to a case, perhaps even before they've thought to ask. "This is potentially very valuable," says Denison.
Wallqvist believes the system can go even further and make not just assessments, but predictions. Using data buried in documents, he says, its system could suggest likely outcomes of mergers and acquisitions, or true estimates of project costs. "We've gone to the level of figuring out and structuring data. Now we have the ability to surface that record of the past to predict the future." Today, Watson; tomorrow, Holmes.