My Current Love-Hate Relationship with AI

By gmurai, May 22, 2026

It’s early in the relationship, I know. But still, there are some things that bug me. Yet, I also know that it’s a relationship in which leaving is not an option, and even if I could, it’s not to the point where it’s so bad that I would do so. So, if you would, let me gripe a bit.

While there’s been much discussion about AI and, at least in my neck of the woods, a fair amount of discussion about how lawyers can, should, and must use AI or risk becoming discarded into the dustbin of history, much less has been written about clients’ use of AI.

Increasingly, I’ve gotten the sense that my clients are using AI. For example, I had a client ask for confirmation that if he disagreed with an administrative decision that he could file a writ of mandate, and if so, whether that deadline was 30, 60 or 90 days after the administrative decision. The answer to the first question was yes, and as to the second question, the answer was 90 days. This was from a client who, smart as he is, probably didn’t know this off the top of his head.

I suppose one could get defensive about it – that the client was second guessing or quizzing me – but most lawyers are forged in a kiln in which they are trained to question authority, and with widespread access to AI, that authority now includes themselves.

In fact, I think good things can come of it. It forces a lawyer to brush up on the law – which, in my case, is over two decades old – and forces lawyers to reconsider the “why” in why they do things the way they do. In short, it keeps your sword sharp, which is important when you’re on the battle field.

But, while I think there are personal benefits for lawyers who have clients who ask sophisticated questions, I’ve also seen a downside. And it is this: I really like my clients. And what I mean, is that I like them as people. And while I wouldn’t describe my relationship with most of my clients as one of friends, I certainly want the best for them, and that means doing my upmost to obtain the best possible solution for them while recognizing that I come at a cost. And this is where I have my gripe with AI.

AI can provide accurate information or knowledge. What it lacks, however, is what I would call wisdom. And there is a distinction. For example, while it’s true that lawyers can legally appear remotely at depositions while their clients appear in person – and this, I suppose, can save money – it’s not a wise decision and is (to my mind) a proverbial example of being penny-wise, pound-foolish. Why? Because it can suggest to the other side that the attorney doesn’t care enough about the case or their client to be there in person while their client is, or worse, can leave a client exposed and alone in the presence of adverse counsel without his or her attorney’s ability to “read the room” and provide the client with effective counsel.

AI, in its present iterations, can’t provide that kind of insight, at least not without the right prompts, which as any lawyer doing research on Westlaw or LexisNexis or inputting queries into discovery databases knows, is extremely important. “Junk in, junk out” as they say.

So, that’s my current love-hate relationship with AI. AI can provide correct answers, but wisdom, insight, or truly thoughtful responses at least (in my opinion) not yet, and when clients believe that correct answers alone dictate the most optimal courses of action, the results can not only be less than optimal, but even sub-optimal.