Attorney – and LawGPT subscriber – William McDermet recently sat down with us to discuss the evolution of AI in the legal realm. With a practice rooted in consumer protection and a knack for e-Discovery, Mr. McDermet explores the potential hazards and the tremendous benefits of the legal application of AI… and what the future holds for industry professionals.
How has your approach to applying AI changed over the last two years?
We’ve been hearing about this new wonder-tool for a while, but it’s really only just come to fruition. I’m an early adopter of IT, but I’m also a structural engineer, so I’m cautious by nature. I see generative AI as a powerful, dangerous tool – like a table saw. I can get a lot of wood cut quickly, but I could also cut my hand off! I can use AI to write a terrific brief, but only if I use it correctly. Attorneys have gotten nailed by judges for submitting garbage AI-generated briefs.
To what extent are you currently integrating or relying upon AI in your practice?
I specialize in document review, looking through thousands of documents, like manufacturing or environmental records. Traditionally, this work has been done by dozens or hundreds of attorneys looking through piles of documents. That manual approach has several obvious weaknesses – people are imperfect and inconsistent. With AI, we can zip though 250,000 documents in a weekend, and the results will be consistent. We have started using AI like LawGPT to work through document dumps to enormous effect, but I don’t envision a day when the human is removed. I use AI to sort the wheat from the chaff.
What are your most common applications for AI to date?
Plaintiff’s attorneys have long struggled with the problem of volume in Requests for Production. Defendants might use the strategy of flooding Plaintiff’s counsel with gigabytes of docs, so that the incriminating evidence is buried, never to be found. To some degree, that problem is now obviated – LawGPT can comb through that pile, and find the needle in the haystack, negating the old tactic of “burying your opposition.” This is particularly awesome for firms like mine, because we represent people who have been wronged, often by mega-corporations with vast resources. AI can level the playing field, and put the advantage back on our side, so we can bring justice to our clients.
What are the misconceptions about AI that you recognize amongst legal professionals?
- “AI makes stuff up, and you can’t trust it.” True, we have seen Generative AI suffer ‘hallucinations’, like citing to case law that doesn’t actually exist. But this is becoming less of a problem as the science evolves.
- “AI can write my brief for me.” No, AI lacks two things: creativity and judgment. The hardest part of writing is getting started. I can use AI to start the brief, to generate an outline, and I take it from there. It doesn’t have – likely will never have – the real creativity or the judgment I possess to write something compelling. Along the same theme, it lacks ethics. I think the most important class in any law school is Professional Responsibility. The AI program never took that class! For example, my computer can find evidence and write a brief about a cloud of vinyl chloride that wafted through a small town in Ohio – but it doesn’t care about the children in that town, or the difficult moral balance we seek when we ask tortfeasors (who are people, too) to make those children whole.
What advice would you give to attorneys who are hesitant or are newly embracing AI?
Your hesitation with this tool is arguably prudent. But legal professionals have no choice but to learn it. At first, judges were forbidding submissions made using AI, because the early programs made frequent mistakes. But we will soon see the day when a judge will ask why an attorney didn’t use AI. Gone are the days when we can have 150 attorneys spend 37,000 hours reviewing documents, and ask the judge to award us $9,000,000 in fees just for that review.
With legal writing, I’m convinced that I can do a better job if I use LawGPT to get me started. Therefore, if I don’t use it, I’m not providing the best level of service to my clients. AI can offer insight I may not have considered. I like to think of AI as an intern or new associate from a good law school, who has terrible judgment, and lacks practical experience. Much legal writing is very formulaic, especially discovery (RFPs and RFAs always use the same boilerplate language). AI is terrific for generating these items, subject to careful review from a human. Just like you might have an intern write these, but you would never simply send his work out the door without proper review, consideration, and revision.
I was taught to always look for your opponent’s mistakes. We will inevitably see opposing counsel misuse AI. But, if we aren’t proficient with technology, we may not recognize those errors and thus, miss an opportunity.
How has AI implementation – including prompt engineering – improved your skills as a legal professional?
Better issue spotting and all-around efficiency. Also, it has sharpened my writing skills. Prompts function best when the writer provides context. I often forget that my intended reader doesn’t have the same context I have in my brain, after closely dealing with a case for weeks. We “engineer” the prompt by rewriting it repeatedly. LawGPT encourages me to write better, and more thoughtfully, until I get the most useful output.
What do you envision for the future of AI in the legal realm?
Large Language Models are just getting bigger. Generative Pretrained Transformers (GPTs) will keep getting more sophisticated. Courts will soon be using AI to check your filings. Litigators will use it to predict litigation outcomes. And, hugely, these services, which are now quite expensive, will eventually become more affordable.
Always, always remember: AI is extremely confident in the correctness of its answers. It’s very convincing. But ultimately, it has no judgment. Learn it, use it, but always, always verify.
Ready to equip your firm with LawGPT?
