If you think that AI is going to be the magic bullet that solves all of the challenges in your firm, then we have some terrible news.
Yes, AI is changing many aspects of our daily lives, and law is no exception. While we’ve been mechanizing things for centuries, AI can work at remarkable speed, but the quality of its output still depends heavily on the instructions, information, and oversight provided by humans. An increasing number of professionals are using it as the world’s most powerful assistant.
But think about assistants that you’ve had challenges with in the past. Yes, they’ve been hungry and eager to learn. If you ask them to go and grab something, they’ll run out the door at lightning speed in order to get you what you want. Yet you also know the frustration that can come when they return with the wrong thing – whether that’s the wrong coffee order, the wrong book from the shelf, or the wrong precedent case. It’s not their fault because they didn’t know any better, but the end result is the same.
AI is similar and often used as that hungry assistant. The challenge is that AI still doesn’t know what it doesn’t know. If you’ve heard of hallucinations in the legal context, that’s essentially when the system is making up answers believing they’re correct instead of admitting that it doesn’t know. If a human assistant did that for you, they would be fired immediately. So why are we giving AI such a free pass?
The Human Element
There may be great uses for AI within your practice. However, if you think that it will quickly replace the humans you’ve hired, you may want to reconsider your position.
To be used effectively, AI still requires a great deal of human training and intervention. At the outset, you’ll need to make sure your staff is trained on how to appropriately use AI. Setting out a firm policy can help educate staff on what usage is okay, and what usage is not. For example, AI may be helpful for some research, but unsupervised case law research can lead it to hallucinate cases.
Which leads to one of the main stipulations – AI needs to be carefully checked by knowledgeable humans. You’ve likely seen the news stories of lawyers who are penalized severely in court for their over-reliance on AI-suggested cases that turned out not to be real cases at all. Without safeguards, AI can fabricate an entire precedent-setting decision, and then encourage reliance on that very fabrication.
In the discovery space, AI needs careful supervision to ensure that it’s making the correct privilege calls. It can be a helpful tool for completing and sorting privilege logs, but privilege can be tricky. Overly cautious redactions may raise the eyebrows of opposing counsel, and sloppy redaction practices that are not carefully checked may reveal things detrimental to your position.
Decisions, Decisions
There is no question that AI is changing the legal landscape yet remember that AI is not making decisions. Those ultimately come down to humans, and our own will. AI may select a document for inclusion, but ultimately the decision whether or not to follow that call or to rule against it and include a document for production is ours.
The same can be said in multiple aspects of legal practice. AI is simply following our commands, but it does not in itself make any actual decisions. The practice of law involves making decisions routinely, on a rapid-fire basis. Deciding whether or not to ask a question, how to follow-up on a piece of evidence, what a witness’ body language can mean – those are all human points of analysis that AI will not take away anytime soon.
If you’re wondering where to begin, start small. Identify one repetitive, low-risk workflow, establish clear policies and guardrails, keep knowledgeable humans involved, and measure the results before expanding AI into additional areas of your practice.
Want to learn more about how we’re incorporating AI here at Everest? Visit our AI page to learn more.
