When I was a young prosecutor, I remember putting the lead detective on the stand during a murder trial and introducing nearly every photograph taken during the investigation. If there were 100 photographs, I was going to introduce all 100. I didn’t want to risk leaving something important out.
Of course, many of those photographs weren’t particularly relevant. Some were duplicative, others added very little, and I’m sure the jury was bored to tears. But I thought I was being thorough.
As I gained experience, I learned to introduce maybe 25 or 30 photographs instead. I became comfortable leaving the other 75 out because I understood which ones mattered. That judgment came from trying cases, watching jurors, and learning what was actually needed.
I’ve been thinking about that experience lately because generative AI can behave like a shotgun precisely when we need a rifle, and like a rifle when we could benefit from a shotgun.
Consider the briefs we are beginning to see from lawyers and self-represented litigants. GenAI makes it remarkably easy to produce a 35-page brief when 15 pages might have been far more persuasive. Every conceivable argument can be developed, every available authority cited, and every point expanded well beyond what is necessary.
I recognize my younger self in that approach. I thought including everything was the safest way to try a case. It took experience to understand that being thorough and being persuasive were not always the same thing. As the old adage goes, if I had more time, I would have written a shorter letter. GenAI may give us that time, but we still have to know what to cut.
Oddly enough, the same technology can create the opposite problem when we’re doing legal research.
I recently had coffee with a lawyer who was reminiscing about learning the law by flipping through books. He explained that when he searched for a particular statute, he inevitably encountered the provisions surrounding it. Sometimes he discovered an exception or a related issue he hadn’t considered. Other times he came across something entirely unrelated that proved useful years later.
He learned a great deal simply by looking around.
Today, GenAI can take us directly to the statute or case we need. That’s an extraordinary improvement over spending hours searching through books. But when the tool delivers precisely what we asked for, we may never encounter all the things we didn’t know to ask about.
The shotgun produces more than we need because we haven’t exercised the judgment to narrow it down. The rifle takes us straight to the answer, potentially bypassing the experiences that help us develop that judgment in the first place.
Of course, GenAI can help us shorten briefs and explore related law. I’m not suggesting we return to law libraries or spend hours doing work that technology can accomplish in seconds. But I do wonder what happens to the next generation of lawyers if they never have to wrestle with the material themselves.
I learned which photographs to leave out by first working through all 100. My friend learned the law not just by finding what he was looking for, but by discovering what he wasn’t.
We should embrace the efficiencies these tools offer. But we should also recognize that some of the work we’re eliminating helped us become better lawyers.
A tool can help us find the target or cover the entire field. It takes experience to know where to aim, and what to leave alone.
