Cathy Cowin, a solo practitioner based in Fresno, Calif., isn’t new to technology. Long before becoming a lawyer, she earned a degree in computer systems and applications. So when generative AI began creeping into the legal space, she understood both its power and its risks.
“ChatGPT scares me because it’s out there in the wild,” says Cowin in a recent law.com vodcast, referring to its relative lack of guardrails concerning law-practice matters. “[But] it became clear that [generative AI] was going to… become commonplace.”
When LexisNexis launched a bounded, law-specific generative AI tool (Lexis+ AI Protege), Cowin felt ready to fold the technology into her practice – despite her instinctive hesitation as a solo to invest in new tools.
“Overhead creep is something that I’m always really cognizant of,” says Cowin. “But in the case of Protege, it’s allowed me to take cases where I wasn’t completely comfortable with that area [and] handle really tough questions from clients on the fly.”
“It’s a non-negotiable for my budget,” she adds.
Minutes to a Simple Yes or No
Asked for a favorite AI use case, Cowin talks about a time she needed to boil down a thick stack of banking documents into a simple yes-or-no answer. In the vodcast, Cowin describes uploading the documents into her AI tool and getting an answer “within minutes” – replete with relevant sections specifically highlighted.
AI isn’t perfect, according to Cowin – sometimes turning up cases not “quite on point” for her research purposes. But even then, she says, she’s not particularly put out.
“Often in that context, there’ll be something to the side on the screen that will take me to something that is relevant,” says Cowin. “I haven’t had a negative experience where I’ve been completely misled.
In the vodcast, Cowin talks about upleveling her research and analysis by treating AI as a sounding board – a stand-in for the associate she doesn’t have.
The associate analogy, for Cowin, also encapsulates the risk element of AI in law practice.
“It’s a tremendous resource, but you have to look at it like an associate that you still need to keep track of,” she says.
That said, generative AI is known for defaulting to sycophancy. Cowin’s secret for ensuring trusted answers from AI: pretending to be opposing counsel.
Getting Safe and Comfortable
Despite her initial caution about unbounded generative AI, Cowin points out that the need for human quality control isn’t much different than with other computer applications and online tools.
“The internet [is] something like the world’s most efficient idiot; if you ask it for chips, you’re equally likely to get an Intel chip for your computer or Doritos,” says Cowin. “We need the opportunity to observe [new technology] safely.”
To that end, Cowin recommends LexisNexis’s live demos and training sessions for taking lawyers from skeptical to proficient.
“Then [you can] launch off to where you feel comfortable… that you know how to do the correct queries,” says Cowin. “And you’re going to get great results.”
Joe Stanganelli is a writer and recovering attorney. He is managing director of content advisory Blackwood King LC.
To read more insights, thought leadership and other video perspectives from Lexis+ Ai, click here.