A new client found me recently, and he found me through AI. He had described his situation to a popular AI assistant, and it suggested he talk to me. He was upfront that he had also used that same AI to figure out how to handle his legal problem on his own. Here is the part that should get every business owner’s attention: the AI had given him the opposite of the advice I gave him in our consultation.
He asked me to read the transcript of his AI session. I did. What I found is the whole reason I am writing this.
Everyone Is Using AI Now, Including Lawyers
I told him what I will tell you. I am not bothered that he used AI. Clients use it. Attorneys use it. Judges use it. After 25 years of practice, I have watched a lot of tools come into this profession, and this is a powerful one. It is not going away, and pretending otherwise helps no one.
So this is not a warning to stay away from AI. It is the opposite. AI can be a genuinely useful starting point for a business owner trying to understand a problem. The question is what you do with what it hands you.
The Facts Were There. The Adversarial Thinking Was Not.
When I read the transcript, the AI had not malfunctioned, and my client had not left out the facts. The relevant details were right there in what he had typed. The AI read them, gave a clean and confident analysis, and still reached the wrong conclusion. It had everything it needed and missed the point anyway.
That is the real difference between a client using AI for legal advice and a lawyer using it. It is not the tool, and it is not always the input. It is the guardrails and the follow-up questions. A lawyer’s job, even with every fact on the table, is to ask the next question the facts demand, the one that tests the position against how the other side will respond.
What the AI Missed: The Other Side’s Argument
Here is what happened next, and it is the part I keep thinking about. During our consultation, I wrote out a single question for my client to paste back into his AI session. It pointed the AI directly at the issue it had skipped.
The AI’s answer changed. It acknowledged it had overlooked the point, noted that the other side would argue exactly what I had flagged, and concluded that the position needed to be reevaluated. The new position it landed on was the position I had given him in our meeting.
Consider what that means. The same tool, with the same facts, reversed itself the moment it was forced to weigh what the opposing party would say. It had delivered a confident answer without ever doing the one thing litigation is built on, which is adversarial thinking. AI retrieves and organizes information well. It does not instinctively ask how the other side will attack a position, unless someone who does that for a living tells it to look.
Why a Confident Wrong Answer Is So Dangerous
A wrong answer that sounds unsure is easy to distrust. A wrong answer delivered with total confidence is the dangerous kind, because you act on it. My client was prepared to make a decision based on a polished, well-written analysis that happened to be backwards. In a business dispute, that is often the kind of move you cannot undo. You send the email, you take the position, you sign the document, and now it is evidence.
AI does not know what it does not know, and it will not volunteer what you did not think to ask. Even with every relevant fact in front of it, it will not necessarily flag that the real exposure is a clause you read past, a deadline that has already started running, or the argument the other side has been quietly building.
How to Use AI Without Getting Burned
I am not telling you to stop using AI for your legal questions. I am telling you to use it the way a professional does.
Use it to get oriented, not to make the final call. Treat its answer as a first draft, not a verdict. And before you act on anything that carries real risk, have it pressure-tested by someone whose job is to think about how the other side will respond. The most valuable question in a legal matter is rarely the one you started with.
That is what you are actually hiring a lawyer for. Not to look things up. To know what to ask, what is missing, and what your opponent is going to do about it.
If you are using AI to navigate a business or legal problem and you want someone to pressure-test it before you act, contact the Law Offices of Scott D. Wu at (626) 799-1858 for a consultation. Bring the AI transcript if you have one. I am glad to read it.