How Should Lawyers Be Using AI Today? A Legalweek Informal Survey

As longtime subscribers to Marketing the Law Firm may recall, I go into the conference each year with a question for all of those with whom I have meetings, whether scheduled and formal or not so much. Sometimes the question and answers don’t amount to much and sometimes they are worth noting. The last time that happened was in 2023 when I asked about AI and legal in general.
The question this year was how law firms should (or could) be using AI as the technology stands today — adding that last qualifier to try and reduce the amount of projected future AI capabilities and to make it more useful to pardon the pun.
Some of the answers are mentioned below, but everyone I spoke with contributed to the overall results. 
Some of the comments on AI apart from uses were interesting on their own, and some of those are included as well.

Summarization

Not surprisingly, this was the answer mentioned most often. It’s the use straight from Hollywood central casting — it takes a mundane, time-consuming task that first-year associates dread and creates a first draft in seconds (emphasis on the first draft part — more on that later). As friend-of-the-newsletter and Board of Editors member legal industry consultant /researcher Ari Kaplan said: “It takes dense amounts of data and summarizes it into something practical and comprehensive.”
Using AI to summarize a pleading, a deposition, or notes from a meeting with a client or partner is an easy, low-risk use of AI. As Daniel Bonner of Level Legal put it, using AI for summarization provides “confidence that the tech works and has value and allows you to explore other use cases.”
The time saving aspect of AI summarization cannot be overstated. Not only does it literally save time in doing the work of summarizing a deposition but instead of sending the deposition transcript to a first year or paralegal and waiting for them to turn it around into something you can use, there are AI-fueled products that can provide a summary in seconds which you can use it right away in working on the case and — maybe most importantly, as Veritext’s Dave Dasilva pointed out — use to send the client a summary of the event immediately, rather than a week (or more) later. That plays really well with clients and puts a notch in the reliability and client-focus belt that goes to getting more work from that client.
Now, remember the emphasis on the first draft? That leads to another response from Legalweek folks …

Generating Draft Content

Similar to summarizing, this is a huge time saver. Need to draft a contract between a seller of goods and a buyer? How about a memo to your supervising attorney on the issues that need to be included in a complaint or arguments to make in an answer? Or how about writing an article for a trade publication? “AI can provide a template,” said Veritext’s Tony Donofrio. “It’s a content generator.  

E-Discovery

It wouldn’t be legal tech without a little e-discovery mixed in. “In e-discovery, you’re looking for patterns within documents, like a fingerprint check,” said ACTFORE’s Christian Geyer. 
“AI can replace or augment doc review — you couldn’t do 10-15 months ago,” said Epiq’s Brandon Hollinder.
“AI can provide a targeted review of hot docs to help determine which are privileged,” said FTI’s Ashley Allman.

Humans Needed

Every article on AI that we have published in the last three years has included a caveat that any AI-generated work product needs to be reviewed by a human. Think of AI as “abundant interns” – a term that came from Exterro’s John Vincenzo and Anthony Diaz that I intend to use from now on. You wouldn’t submit a report from a summer intern or first-year associate without reviewing it, right? Right?? The same goes for AI. As Casepoint’s Amit Dungarin said, there “has to be a human element.”
Anecdotally, case hallucinations from AI-generated legal research is down, and experts insist that they will soon be a thing we will look back on with knowing sympathy for those caught in it, but as recently as this past February, lawyers from Morgan & Morgan were sanctioned for submitting a motion in limine in a products liability case which contained fake cases (for details, see Bob Ambrogi’s coverage). So while you may hear that hallucinations are not a concern, don’t believe the hype. Even as AI technology improves and learns, I can’t foresee a time when you wouldn’t want to review any court filing before submitting it. In fact, as the judge in the Morgan & Morgan case points out, “using a fake opinion to support an argument is a violation of Rule 11(b)(2)” of the Federal Rules of Civil Procedure.
At least for now, AI-generated content is “imperfect,” as Kaplan said. However, he said that’s not a bad thing. “If it’s imperfect, that’s ok; perfection brings challenges.”

Extras

In addition to the top uses as gathered by this very informal and unscientific survey, some interesting elements and considerations for using AI in legal came through.
AltaClaro’s Abdi Shayesteh said that AI can be used as “a hiring tool, especially for summer programs. Standard assignments can be provided across all candidates and can be assessed objectively. You can also use AI to create simulations for summer associates and litigators.”
“AI can be used In-house,” said Xakia’s Anne Post. “Use it to gather feedback from clients – what is the problem they’re trying to solve? Is it worth the investment?”
“We need controls and guardrails around GenAI,” said Brian Corbin of Quislex. “There needs to be widespread adoption of controls by industry groups and law firms. It should be People, Process, Technology.”
“Prep for trial and create exhibits,” said FTI’s Ashley Allman “AI can synthesize and provide real insights fast. Trial lawyers will start using chat bots instead of carrying around case binders.”
I thought Prosearch’s Brian Meegan summed it all up nicely: “We’re just getting snippets of what people can do. We need to be more effective, targeted and strategic in designing prompts. We need to start using AI everyday — the  more use, the more you will find use cases. You need to get hands on practice.”
We’ll see what the rest of 2025 and 2026 brings and gather together again in March in New York City.