Start With the Magic Trick
The slide had maybe four Markdown files on it. Then the next slide had the instructions that were supposed to live inside those files, and the slide after that had instructions for the instructions. I looked up from my notes and saw the same expression on every face at my table: the flat, slightly embarrassed stillness of people who have quietly decided they are the problem.
These were not people who struggle with complexity. Around me sat lawyers in their fifties and sixties, established, genuinely brilliant, the kind of minds that have spent decades untangling statutes and reading their way to the bottom of problems that would flatten most of us. They are not necessarily technical, though, and that combination is its own particular trap. Someone who is used to understanding hard things, who has built an entire career on understanding hard things, has almost no practice at the specific discomfort of not understanding this one. So when the material sailed over their heads, they did not read it as bad teaching. They read it as a failure in themselves. They seemed to have no framework for the feeling except the oldest one there is: assume the fault is yours.
The woman beside me is someone I work with all the time, someone I lean on for her knowledge, and I watched the confusion deepen on her face until it turned into something worse than confusion. She had decided the problem was her. You could see it settle in. She was not failing, and nothing in that room was her fault, but she had no way of knowing that, so she carried it anyway.
To understand what went wrong, I want to be fair to the presenter. He is a genuine authority at the intersection of law and artificial intelligence, someone who has worked at the frontier of the field for well over a decade, since long before ChatGPT made any of it fashionable.
He clearly knew his material, and somewhere under all of it was something worth knowing. But twenty minutes in, I was still translating, and I have spent nearly four years with these tools myself, ever since ChatGPT walked out into public in 2022. I have personally taught around twenty people how to use various AI tools. When I glanced around and realized the other heavy users in the room were struggling right alongside me, the problem came into focus. If the experienced people are lost too, the room is not the problem. The pitch is.
What we were getting was a 400-level course delivered to a room that needed a 100-level one. That does not have to come from arrogance. It can come from the curse of knowledge, the strange amnesia that sets in once you have internalized something deeply. The further out toward the frontier you stand, the harder it becomes to see the people still gathered at the trailhead, and he has been out there a very long time. He was teaching from where he stood, not from where we sat. He had forgotten what it feels like to lack the basic framework that makes the advanced material make sense, so he handed us the advanced architecture of a thing before any of us believed the thing was worth building.
Watching that disconnect, I thought about a teacher who understood how to close it. Dr. Gregory Monahan was my main history professor at Eastern Oregon University, and my advisor besides. He was the kind of teacher who would turn up in costume as King Louis the XVI and answer our questions in character, as someone out of the century we were studying. He was funny, he was genuinely brilliant, and he made eighteenth-century French history, of all things, feel urgent and close to a room of undergraduates who had no earthly reason to care about it. He knew exactly where we were sitting, which was mostly nowhere, and he led us up to higher ground without ever once making us feel the climb. He did not begin with the apparatus of the discipline. He began by making us want to be in the room, and the rigor came later, carried willingly, because he had already convinced us it was worth the weight.
That is what I try to do when I teach technology. I have been teaching people technology since I was a film editor in my twenties, and the one principle that has never once failed me is that you have to find out where someone actually stands before you can move them an inch. You cannot teach a person to run who has not learned to walk, and you cannot teach them to walk while they are still learning to crawl. So I show people one thing first. Something fast, something a little magical, something that lands well enough to make them sit up. I do not open with structure. I open with a result.
The trick I reach for most is a deposition summary. Almost everyone I work with has lost a day to the same task: a transcript running three or four hundred pages, read line by line, the key exchanges flagged and typed back up by hand, an afternoon and an evening gone. So I sit down with someone, hand the model the transcript, and ask it in plain English for a summary of what the witness said about one issue, with a citation to page and line for every point. A moment later the summary comes back, each claim tagged to its place in the record, [42:15-18], [118:6-20], the way we would have marked it ourselves if we had been given the hours. I do not ask them to trust it. I ask them to check it. They pull up the transcript, find the cited line, and see it says exactly what the summary claimed. That is the moment. You can watch the arithmetic change behind someone's eyes, the lost afternoon quietly canceling itself, and the doubt goes out with it. They have learned nothing yet about how any of it works, and it does not matter. They want it now, and wanting is the whole game.
People do not adopt a new tool simply because they understand it. They adopt it when they want it, and wanting comes from feeling the payoff before they ever feel the friction. Belief has to arrive before mechanics, or the mechanics never get a chance to matter.
In Seattle, that order had been reversed, and the cost was visible around me. We met one roadblock after another, all before we had felt any reward. He had even given us zip files to work through, with sample documents, a strange prompt to run, and an answer key showing the correct results. I went through them, but the output did not make much sense to me. The trouble was not that it was wrong. An answer key only helps when you understand the question, and we had not been given enough grounding in the problem to appreciate the solution. It was the whole failure in miniature: the reward with the wanting stripped out of it. Meaning was the one thing that could not be zipped into the file.
A sequence like that does not teach rigor. It teaches the quiet fear that a lot of newcomers already carry in with them: that this stuff is fundamentally too hard, too much, not for them. Every unexplained obstacle confirms the thing they were most afraid was true. You can lose a beginner permanently in about ten minutes that way, and the tragedy is that the tool at the center of it is genuinely, almost embarrassingly, capable of delighting them.
The longer I sat with it, the more I realized that the missing piece was motivation. So much of the advice on using AI is about the difference between good prompting and bad prompting, better technique versus worse technique. That was not what went wrong in that room. The session had treated teaching the tool as though it would also create belief in the tool. Those are two different jobs, and too often the first is taught while the second is skipped entirely, as though motivation will simply show up on its own once the mechanics are clear enough. It will not. It works the other way around.
That distinction is why I can criticize the session and still respect what he has to offer. I still want to learn what he was teaching. I have the materials, and I intend to work through them, because there is clearly something real in there. That impulse helps explain the difference between my experience and a beginner's. I already believe. I had my magic trick years ago, so I can hit roadblock after roadblock and keep climbing, because I know what is waiting at the top.
He had the ladder. He just forgot to show anyone why they would ever want to climb it.