How AI is already reshaping lawyers and the practice of law
Gordon Calhoun was up to his neck in jargon. An experienced trial attorney, Calhoun was handling a hotly contested litigation involving reams of arcane data and terminology about pharmaceuticals and medical devices. Neither he nor anyone else on his team at Lewis Brisbois Bisgaard & Smith LLP in San Francisco was expert in the subject. But instead of spending the time and money to hire a consultant, Calhoun began feeding documents into a legal AI tool called Everlaw. Before long, the program returned summaries of the key points in plain English — no pharma wizard needed. Squeezing easy-to-understand summaries of the important issues out of dense jargon “is a huge advantage,” Calhoun said. Instead of relying on outside experts, “just about any litigation team can go in and handle that case.”
While much of the world is still trying to figure out how best to use generative artificial intelligence programmes and chatbots, the legal field has rushed to adopt the technology. Attorneys are taking advantage of AI’S voracious ability to scan through millions of documents and extract meaning from mountains of words. Programs such as Everlaw let lawyers speed up discovery and find needles in legal haystacks that would otherwise take many more days and dollars to unearth. “In a few years, it will be almost malpractice for lawyers not to use (AI),” said David Wong, the chief product officer at Thomson Reuters, at an event in April. The global data and publishing giant has set aside an $8 billion war chest for AI deals in upcoming years.
AI can even take a first pass at writing a legal brief — a bit like a first-year intern. But as with an inexperienced lawyer, it can also make mistakes.
AJ Shankar, the chief executive of Oaklandbased Everlaw, said his company advises customers to consider the software like “a smart intern” whose work has to be checked. That’s because no current AI program is free of what are known as “hallucinations,” where the bot makes incorrect assumptions and returns faulty information, Shankar said. As a result, AI can’t be relied upon yet to spit out flawless work. Those and other risks are prompting some customers to take a go-slow approach. The San Francisco District Attorney’s office, for instance, has been an Everlaw client for five years, using the software “to process and review voluminous documents in civil litigation” and in a small number of complex criminal cases, it said in written comments. But the office took pains to note that it isn’t yet using AI features of the software, such as to summarise or synthesise information within documents, search for or obtain documents it doesn’t already have, or to make charging decisions against suspects.
Still, Everlaw has seen significant customer uptake since the AI boom took off in late 2022 with the release of Openai’s CHATGPT, the same technology that undergirds Everlaw’s product. For litigation consultant Greg Mccullough, using the company’s tools is personal as well as professional. His parents lost their home during devastating wildfires in his native San Diego in 2007 and were part of a lawsuit against the power company. The help he gave them with their case caught the attention of the attorneys working on it, and now Mccullough, who is not an attorney, advises lawyers in wildfire cases. Hired to sift through what can amount to millions of documents to find relevant nuggets, he has worked on litigation resulting from some of California’s largest fires in recent memory, including the 2018 Camp Fire that killed 85 people, as well as the 2019 Kincaid Fire and the 2021 Dixie Fire.
“In the very early days, I did what most people do, which is open PDF files and grab stacks of documents and look” for documents that might be relevant, Mccullough said. “A lot of firms still do it this way.” Now he can use Everlaw’s AI functions to automatically rank documents in order of importance, rapidly return key names from a pile of information, and even search for specifics across thousands of pages like the number of a power pole that may be where a fire started. “I can search through 50 million pages in less than a second and be able to parse down to what I’m looking for,” Mccullough said. AI also has the potential to make legal advice available to laypeople.
Fresno County doctor Thomas Hull experienced that earlier this year, when he was awakened after an all-night emergency room shift by a knock on the door. A person there served him with two thick envelopes containing stacks of dense legal documents. “Am I being sued?” Hull wondered, his sleep-fogged mind racing with anxiety.