Michael Ford (post until Oct. 31/19)
Legal Suppliers

Making the business case for AI in law firms

The role of artificial intelligence (AI) in the legal profession is expanding, and lawyers should be thinking about the opportunity it presents, says Carla Swansburg, vice-president and general manager for the Canadian operations of the legal technology company Epiq.

Machine learning and AI are "creeping up the value chain in legal services," and while advanced technology can help with tasks such as scouring documents for key concepts and words, it won’t replace the need for human input or oversight, she tells AdvocateDaily.com.

"Typically, when we talk about AI in legal services we are talking about machine learning: the use of algorithms that learn from being repeatedly applied to data," she says. "It's really a search tool on steroids.

"In terms of the practice of law, we're creating new ways to use a law degree. It's shifting from the traditional model where we need all kinds of junior lawyers to do the highly repetitive process work,” Swansburg adds.

The work performed at the lower end of the value chain is getting commoditized with technology, and Epiq’s goal is to help clients organize their data and information to require less effort to pull, categorize and review information, she says.

A firm must be able to find "those needles in haystacks" quickly and efficiently if it faces a lawsuit or a regulatory investigation, Swansburg says.

In eDiscovery, machine learning is often referred to as Technology Assisted Review (TAR) or Continuous Active Learning (CAL) applications, she explains.

Epiq uses a number of tools including NexLP, a platform that employs machine learning to cull large data sets down to more limited, actionable document sets for review, including emails and contracts. NexLP can identify negative or positive sentiment by using machine-assisted descriptive and predictive analytics, Swansburg says.

CAL involves a team collecting a significant amount of data and reviewing it for relevant documents. The documents are coded, and the tool uses that information to search for the most relevant or potentially relevant records, she says.

"In essence, machine learning is pattern-matching," Swansburg says. "So take a million documents that have been collected that might be relevant to litigation and then you can use one of these tools to narrow the review field.

"But essentially humans categorize the material and based on relevant tags or topics, the algorithms will know there are also other words to describe a particular topic," she says.

Those algorithms search the data and look for words associated with a specific term, theme or concepts, delivering the requested data while cutting out all other material deemed irrelevant, Swansburg says.

In the "old days," she says a human would have to manually scour all the documents to find the needed data.

"With the rise of data platforms, social media and email use, it got to the point where it was prohibitively expensive to search like that in litigation," Swansburg says.

New technologies started to be developed and applied within the profession in part because the market for legal services changed after the 2008 financial crisis, she says.

"Because of what happened over the last decade, the law has become a buyers' market and clients have become empowered," giving birth to technologies that increase efficiencies while lowering costs, Swansburg says.

Procurement processes are being applied to legal services, driving the "need to bring costs down, she says.

"Reviewing documents tends to be the largest cost of discovery, and with technology, the costs can be reduced significantly," she says. "Toronto is a hotbed for emerging legal technology. That confluence of events all led to increasing sophistication in machine learning and AI tools."

It’s also important to note that a number of courts have vetted the automated processes processes in some circumstances and found lawyers using the technology are meeting their professional obligations, Swansburg says.

"We see this is the way the market is going and so our goal is to stay ahead of trends and keep abreast of how much we can deploy technology to provide efficiencies in this work," she says. "That's a big part of what Epiq does.”

In some firms, Swansburg says there can be an over-estimation of the short-term impact of machine learning and AI and under-estimation of its longer-term effect.

"The use of machine learning is going to continue to creep up the value chain and encroach even more into what was traditionally done by humans," she says.

But that doesn’t mean the need for intelligent humans is going away any time soon, Swansburg says.

“The algorithms used by machine learning and AI must be ‘taught’ before performing their tasks, and the accuracy of the output must be confirmed, which is best accomplished through human review,” she says. "The best outcomes will continue to be machine augmented.

"I don't think technology will overtake what people do, but rather it will help create efficiencies when they’re deployed under the supervision and support of humans."

Lawyer Directory
BridgePoint Financial Services (post to 5.31.19)Toronto Lawyers Association (post to 6.30.19)MKD International (post until Sept. 30/19)Feldstein Family Law (post until May 31/19)Legal Print & Copy Inc.VR Law/Victoria Romero (post until June 30/19)Ryerson LPPWilliams Family Law