Robot power strengthens legal services

While many talk of robots replacing lawyers — a report from professional services firm Deloitte has even predicted 114,000 legal sector staff could lose their jobs to artificially intelligent programs — the reality may prove to be more complex. Automation may instead transform the nature of people’s work rather than necessarily reducing the demand for their skills.

Much digital innovation is taking place among in-house legal teams, which are under pressure to increase productivity and lower risk levels.

For AIG, the insurance group, data mining is being used to improve the hiring of external legal service providers. When looking at its suppliers — more than 1,500 with over 30 business lines in 90 countries — the company realised it would never be able to obtain the best services at the best value using a manual process.

AIG now has automated tools that can send requests to suppliers for information on everything from the geographical regions they cover to their areas of expertise. The information is fed into a regularly updated database.

This data-driven approach — along with organisational change, the creation of a purchasing unit and consolidating the company’s approach to purchasing — has generated savings of $200m a year on legal services.

Technology can help to drive other efficiency gains, particularly when it comes to laborious contracting work.

“There’s a core productivity issue in law,” says Tim Pullan, founder and chief executive of ThoughtRiver, which provides contract intelligence software to the legal sector. “If you gave a lawyer a contract to review today, it would take the same time [to do] on average as it did 20 years ago.”

However, by applying “machine learning” — the ability for software to train itself without being programmed — to the review of contracts and other legal tasks, teams can save time that is better used elsewhere. This is the aim of Dentsu Aegis Network, a London-based media, digital and creative communications business. It is working with ThoughtRiver to use machine learning to increase automation and lower risks in contract departments. After scanning thousands of documents, the system — which is currently being tested — can identify specific language and word patterns that assess the value and risk of each contract.

For example, a contract governed by Singaporean law with Singaporean courts specified for dispute resolution would be categorised as low risk. A contract based on China’s governing law, under which dispute resolution would take place in courts in Beijing, would be deemed higher risk. Nick Tomlinson, Asia-Pacific general counsel at Dentsu Aegis, says this information is depicted graphically on a dashboard based on the set parameters. “When you run the assessment report it highlights the relevant text in the document.”

Teams can then decide how to apportion contract workloads by distinguishing those that are high risk and high value — whether in monetary terms or because they relate to important clients — from ones that are low risk and low value.

“It helps us think about who is the right person to review specific agreements and, if it’s low value, low risk, whether the contract needs a full legal review or could be handled by a contract manager or paralegal, or eventually a non-human review,” says Mr Tomlinson.

A similar principle lies behind the development of UK company Riverview Law’s Kim, a legal virtual assistant, which the in-house services provider licenses to its customers. Kim uses artificial intelligence to help in-house teams improve and speed up their decision making.

Kim has three levels of complexity. In the first, automated process-level assistants prioritise the management of workloads. “You now know what work you’ve got, who it goes to, how long it takes and what it costs,” says Karl Chapman, chief executive of Riverview Law. “And when you have that type of data, you’re laughing because you suddenly have control.”

At the next level, artificially intelligent advisory assistants automatically send information to legal teams about the cases they are working on. Smart assistants go a step further and offer potential solutions based on analysis of thousands of past cases.

The system can identify improvements and efficiency opportunities by analysing and learning from past contracts that can be applied to new ones. It sends these to the legal team, which can review them and make decisions accordingly.

But while the new technologies may generate cost savings, in-house lawyers stress that their real power lies in reducing risk and helping to deploy human resources more effectively. This is the main attraction for Mr Tomlinson. “Our business is growing so part of the objective is to scale the growth of the legal team with the right level of resource and smarts — rather than simply just adding more bodies,” he says.

He dismisses predictions that artificial intelligence will replace lawyers. “It’s not necessarily firing up the robots and reducing our headcount,” Mr Tomlinson says. “I’m not doing myself out of a job.”

Wikipedia: Crowd control

Sometimes a general counsel needs to look outside his team for legal advice. Take Geoff Brigham at the Wikimedia Foundation, the non-profit group that supports and runs Wikipedia, the online encyclopedia written by volunteers.

When Wikimedia wanted to revise its terms of use policy, it asked its community to make edits to the group’s proposal, just as members edit Wikipedia entries. After 200 changes, the draft was put to the board.

Mr Brigham stresses the importance of being prepared to accommodate stakeholders’ suggestions. “There were certain provisions they were concerned about,” he says. “And in another world as a general counsel, I might have insisted on keeping those provisions. But as general counsel, I made decisions on what made sense for our community.”