Artificial intelligence has moved from novelty to daily habit in a remarkably short time. At our firm, it now touches a surprising amount of the work we do. We use it to write software that handles routine business tasks, such as culling the daily reports of accidents in our market. One of my partners built a small automated tool that pulls injury reports from the Nebraska Workers’ Compensation Court. We use it to surface recent court decisions in the areas of law we practice. And most usefully, we have used AI to build software that runs parts of the business itself, from practice management to the medical chronologies that anchor an injury case.<\/p>\n
So this is not a warning from a skeptic. It is a warning from a heavy user. Because for every hour AI saves us, I also see work product from others that was clearly produced by AI and just as clearly wrong. The difference between those two outcomes is not the tool. It is the person using it.<\/p>\n
Not long ago, an expert witness told me that in a Nebraska workers’ compensation case, the insurance company pays for an Independent Medical Exam. I pushed back. He held firm, and told me it was right there in a Nebraska statute. I practice here, and I knew the claim did not hold up. What I wanted to understand was why he was so certain about something I knew to be wrong.<\/p>\n
When pressed, he explained that ChatGPT had told him, and had even cited the statute. To his credit, he did not stop there. He went and read the cited statute, and it said what ChatGPT told him it said. From his perspective, he had done his homework.<\/p>\n
The statute in question, Neb. Rev. Stat. \u00a7 48-134, requires an injured employee to submit to a medical examination at the request of the employer or its insurer. The employer or insurer arranges that exam and pays for it. On its face, that matches what the AI said.<\/p>\n
But anyone who has practiced in this area knows what that exam actually is. The plaintiff’s bar does not call it an Independent Medical Exam. We call it a Defense Medical Exam, because it is requested, scheduled, and paid for by the defense, and it is typically conducted by physicians the defense has chosen and used before. It may carry the word “independent” in its name, but there is nothing independent about it. It reminds me of the health studies commissioned by tobacco companies in the 1950s and 1960s that, for some reason, could never find a link between smoking and cancer.<\/p>\n
That is the nuance the AI missed, and the expert missed it right along with it. The words on the page were accurate. The meaning of those words in practice was not what either of them believed. An injured worker who relied on that advice, expecting a neutral second opinion at the insurer’s expense, would have been badly misled.<\/p>\n
I like to think that someone well educated and well versed in a subject can use AI aggressively and still avoid being led astray by it. The reason is the context that person brings to the conversation. A practitioner reads an AI’s answer against years of experience, knows the vocabulary of the field, and recognizes when a technically correct statement points in the wrong direction. When a true expert uses AI, they not only notice when it is wrong, they know why it is wrong and can steer it back to where it is useful again.<\/p>\n
Someone without that background has no such filter. To them, a confident answer with a citation looks exactly like a correct answer with a citation. And AI is always confident.<\/p>\n
AI, much like an old-fashioned internet search, cannot yet replace the human expert in the loop. That is certainly true in the law, where I use it most. One character in the television series Better Call Saul described a certain kind of lawyer as “a chimp with a machine gun.” That line applies just as well to AI in untrained hands. The real danger is not that the tool is useless. It is that the tool can give an otherwise uninformed person the belief that they know more than they do. That misplaced confidence leads to very wrong and sometimes very costly mistakes.<\/p>\n
None of this means you should avoid AI. It means you should use it the way a careful professional uses any powerful tool. A few rules that serve us well:<\/p>\n
If you are running a business, AI can genuinely transform how much you get done. We are proof of that. But if you find yourself relying on it in an area where you would otherwise need to hire a professional, that is the moment to slow down. The purpose of the professional was never just to know the words. It was to know what the words mean, and what they leave out.<\/p>\n