When artificial intelligence tools first appeared in legal workflows, many attorneys saw them as novelties. Today, they’re quickly becoming necessities. According to the American Bar Association’s 2025 Legal Industry Report, 31% of legal professionals said they personally used generative AI for work-related tasks — up from 27% last year. However, only 21% said their firms had formally adopted the technology.
Attorneys report using AI for everything from legal research to scheduling and business decision-making. But even as adoption accelerates, AI’s presence raises fundamental questions. What ethical standards apply? Where should lawyers draw the line between assistance and overreliance? And how can today’s law students build the fluency they’ll need in a transformed profession?
A brainstorming partner
Adam Shepherd, an associate at Weinberg Wheeler Hudgins Gunn & Dial in Atlanta, began using AI tools during his third year at Emory University School of Law and has since adopted them into his practice after graduating in 2024.
Shepherd views AI as a supplemental thinking tool.
“I use it like a brainstorming partner,” he said. “But it’s just that, a partner. You still have to lead.”
His approach centers on idea generation, not execution.
“I’ve seen people get in trouble when they just say, ‘Hey, give me a motion to compel,’ and the model hallucinates,” Shepherd said. “I think that goes back to the ethics we all learned in law school. You have to check and verify. It’s not about replacing the lawyer. It’s a tool to help you generate ideas, switch off your mind for a second, and let something else spur your thinking.”
One of Shepherd’s most frequent uses of AI is for administrative polish. He uses ChatGPT for what he calls “nonbillable ministerial tasks,” such as proofreading emails and refining time entries.
“It really cuts down on having to go back and proofread every punctuation mark,” Shepherd said. “I do read and make sure that it’s what I wanted to say. It’s about saving time, especially on things like professional emails, making sure the language is polished when responding to partners or clients. That’s where it helps.”
Security and ethics remain a priority. Shepherd said his firm avoids inputting any identifiable or confidential client information into public AI tools.
“We’re very careful not to put any kind of information that could identify the plaintiff or defendant,” Shepherd said. “Certainly, no confidential or medical information.”
For sensitive documents, Shepherd’s firm uses closed-source platforms like Claude.
“It’s basically an AI just tailored for a firm,” he said. “You can put confidential documents in, and it operates like ChatGPT, but within a protected space. That helps with risk management, especially in a litigation environment.”
‘You’re in it as a lawyer’
Beyond personal use, Shepherd has played an active role in helping his colleagues better understand the technology. He recently gave an internal presentation to demystify AI and explain its practical benefits.
“I told them, even if you watch something like ‘Suits,’ you’ll see someone flipping through law books,” he said. “That’s outdated. It’s like comparing a calculator to writing math problems by hand. We’re at the same kind of shift now. AI is the next step, like the jump from books to Westlaw. If we don’t adapt, we’re going to be left behind.”
He illustrated how AI platforms can aid in complex legal questions. When a partner asked whether a motion to dismiss stayed the requirement to answer other counts, Shepherd turned to the AI-powered features of Lexis.
“There’s no clear-cut statute on that,” Shepherd said. “It depends on case law, the jurisdiction, procedural posture. That’s when AI tools like Lexis can give you a head start. They pull up cases and treatises. You still have to check if it’s dicta or holding, if the facts line up, but it speeds up the process.”
For Shepherd, the tool doesn’t replace judgment.
“You’re in it as a lawyer,” he said. “AI gives you a boost, but you still have to know what you’re doing.”
Patent attorney Michael Alexander also sees AI as a key part of his modern practice. Based in Pittsburgh, Alexander has more than 20 years of experience working with inventors and startups at the cutting edge of materials science, automation and artificial intelligence itself.
— Michael Alexander, patent attorney
In a field where precision and deadlines collide, AI offers him a way to streamline initial drafting.
“I’ve incorporated AI into my patent drafting workflow, especially for early-stage drafting and claim iterations,” Alexander said. “It’s also been a helpful brainstorming tool when preparing responses to the United States Patent and Trademark Office. The biggest value so far has been in speeding up idea generation and refining language, especially under tight deadlines.”
He noted that most of his colleagues use AI for revision and polish, but a few are pushing it further — sometimes too far.
“Most attorneys I know are using AI to polish their writing … improving grammar, tone or structure,” Alexander said.
The risk, he cautions, comes when attorneys mistake form for substance. Legal writing generated by AI may look correct, but it lacks legal judgment.
“A smaller group is using AI to actually generate first drafts,” he said. “Right now, AI is mostly imitating what a legal document looks like, without knowing whether it’s good.”
For Alexander, the next step in AI’s evolution will hinge on feedback loops — teaching models not just how to draft, but how to learn from outcomes.
“Once it starts learning from which arguments succeeded or failed, then it’ll truly begin to understand what makes a strong legal document,” he said.
Even with these limitations, Alexander believes the ethical imperative is shifting. Failing to leverage AI when it could clearly help a client may one day be seen as negligent.
“Honestly, nobody fully understands how these models work, and that includes the experts,” Alexander said. “But attorneys don’t need to be computer scientists. What they do need is practical experience and good judgment. The key is building intuition: knowing when AI helps, when it’s bluffing, and when to take the wheel.”
That pragmatism is essential in his view of legal ethics. While misuse of AI could lead to breaches in confidentiality or accuracy, a refusal to use it might soon fall short of the bar’s expectations.
“Attorneys must be vigilant about confidentiality, accuracy and the limits of delegation,” Alexander said. “But there’s also an emerging ethical obligation to use AI when it measurably improves our work. Under the ABA’s duty of technological competence, it may soon be unethical not to use it, especially if clients are harmed by slower, costlier or less accurate work as a result.”
The thirst for knowledge
For law students aiming to stand out, both attorneys stressed the same value: curiosity.
“Be curious. Try the tools,” Alexander said. “Don’t wait for a law school class to give you permission. The best way to understand AI is to start using it. There’s a real opportunity to be ahead of the curve.”
Shepherd echoed that mindset, saying much of his skill came from experimenting on his own time.
“Prompting is very important,” Shepherd said. “I like learning about this stuff. I’m always tinkering, trying new ideas, watching tutorials. It’s kind of like learning a new area of law. You try something. It doesn’t work. You adjust.”
That adaptability, Shepherd believes, will define the next generation of lawyers.
“Never lose that thirst for knowledge,” Shepherd said. “Stay curious. Experiment. Fail fast. And keep going.”
He wrapped up with a metaphor that reflects how he sees the role of AI in the legal field.
“ChatGPT might give you vanilla ice cream,” Shepherd said. “But your job is to turn it into a sundae.”
