Using innovative technology to enhance the delivery of legal expertise can bring law practice to a higher, more competitive level, says Dera Nevin, Senior Associate at Baker McKenzie and a legal technologist. But the prevailing view is that the uptake of technology in law firms is slow, with investments and adoption lagging behind other industries. Dera explains some of the challenges that underlie this stereotype and shares her insights into what may impact legal technology adoption by law firms in 2019.

1) Data Quality and Availability

Since joining the firm and its award-winning Whitespace Legal Collab in 2018, Dera has helped advise on the expansion and integration of technology into the delivery of legal services to stay aggressively relevant in solving problems for clients.

Good machine learning technology rests on the quality and availability of data.  As law firms move towards incorporating machine learning technologies into decision making, the quality of data is critical for producing results that do not have adverse impact on clients, the public or society.  “The consequences to client matters are too significant to risk on imperfect technology or data,” she said.


“The generalized criticisms of law being slow to adopt technologies may minimize some of the realities of operating technology in the legal context. Tech in law, especially technology that relies on dataset training, takes longer to mature then consumer-facing technology.  For example, some of the datasets we use as inputs to train machine learning may come from case law, on topics which are highly fragmented, unique, and specific. These datasets are the opposite of the large, aggregated sets like driving routes collected by Uber that are highly repetitive and common. Often, we don’t have the scale of data to train machines quickly and so legal technology can take longer to mature.”

“Law is not known for having stores of clean, structured data, but there is a movement to refocus efforts on this,” she said. “The concept of garbage in, garage out is real.”  Referring to examples of machine learning being used in criminal sentencing proceedings in the United States, in which the prevalence of human bias has been identified, she notes:  “We see in machine outputs reflections of the biases and moral judgements of humans that left unchecked can be problematic and harmful.”

2) Natural Language Processing

Because so much data in the law is text-based, natural language processing is another promising technology emerging in legal service delivery.  Natural language processing (NLP), a sub-field of artificial intelligence (AI) seeks to help legal professionals use text artifacts more efficiently, by having computer algorithms handle some of the “interpretation” and processing of text. Ultimately, NLP is focused on enabling computers to understand and process human languages.

“Humans don’t communicate in ‘structured data’ or Boolean searches, we communicate using words,” says Nevin.

“Imagine you are searching a corpus of text, or a body of laws,” she adds. “Your typical search will only as good as keywords you plug in, and the ability of the system to process those keywords.” She notes that two problems arise in that an individual may not know the words they need, or the words they need may not be available to the search technology (either because the words aren’t present, only the idea is, or the words may not be indexed and available to the search engine). “Imagine you are addressing a new situation like drone regulation. Those particular words associated with drones may not be found in case law but you need to find things that represent same idea or concept. With NLP, concept searches become available, so users can engage in complex searches without specialized formulas or exact keyword matches.”

The promise of NLP  is that the technology will help augment search concepts by enlarging concepts, but a potential risk is that it may introduce more and different kinds of extraneous information for lawyers to sift through if the system is improperly configured or poorly designed.

“This technology is very promising in the long run,” Nevin believes. “But, there is time associated with normalizing data. It is important to understand any current limitations of the technologies.”

3) Process and DocumentAutomation

There is a resurgent interest in technologies that facilitate automation of processes, especially where automation of document creation, filing and handling is possible, and where human judgement can be simulated through AI-based expert systems.

“Process automation encompasses a range of solutions, from the simplest mechanism of technology that helps autofill content in forms to speed up creation and reduce errors, to a logic tree operating in the background of an AI-based decision model,” Nevin states. The impact can mean improved speed or quality checking of work, for example, through Technologies that can identify mis-defined terms in contracts. Or, entire workflows behind legal solutions delivery can be standardized through decision trees. “I expect we’ll see greater standardization of service delivery through the use of these technologies as they evolve and mature.”