I was searching for a file the other day, the way you do, digging through old folders for something specific. Instead I found a handful of decks and documents from exactly one year ago.
I opened them expecting a quick nostalgia hit. What I got instead was laughable. Not bad, exactly. Just distant. The content I was teaching, the way I was explaining things, the language I was using, it all felt like it belonged to a different decade, not a different year.
That sent me down a rabbit hole. If a single year can make my own work look that outdated to me, what does that actually mean? It means the landscape moved. And if the landscape moved that much, the way I teach, the way I work, the way I show up in a room to explain any of this, all of it has to move too.
Here is the test I keep coming back to. If you are teaching the same content, building the same decks, and explaining things the same way you were a year ago, that is not neutral. That is a signal. It means you have not adjusted to the pace of the thing you are supposedly teaching people to adopt.
We cannot move at the speed of AI. Nobody can, and honestly nobody should try. But we can make sure we are growing and scaling in how we think, how we process, and what we prioritize. The question is what the right benchmark actually is. Three people working on exactly this problem gave me language I did not have a year ago.
Stop counting logins
Jeremy Utley puts it plainly: in five years, nobody will care how many people logged into ChatGPT. They will care about who actually used it to transform their work. More use and better use are not the same thing, and most organizations are still only measuring the first one.
He also said something that stuck with me even harder. The language we use reveals our mindset. High performers do not "use" AI. They work with it. They coach it, give it feedback, have a real back and forth. Think about your legal counsel or your CMO. You do not "use" them. You work with them.
That single word swap, use versus work with, is a benchmark on its own. If you are still saying you "use" AI the same way you would use a stapler, that is worth noticing.
Everyone is now a team of five
Brice Challamel, who led AI transformation at Moderna before joining OpenAI, said something that reframed the whole conversation for me. There are no more individual contributors. Everyone is now a team of five: themselves, their AI assistant, their AI coach, their AI expert, and their AI creative partner.
Read that again next to your own job title. You are not just doing your job anymore. You are leading a small team, and the quality of that team's output depends on how well you delegate, direct, and review, the same skills that make any leader good at leading actual humans.
This is where I think a lot of us are stuck without realizing it. We are still operating like solo contributors while quietly holding the job of a small team leader. The gap between those two mindsets is exactly the kind of thing that makes a deck from a year ago look outdated. A year ago, most of us were still solo. Now we are supposed to be managing four teammates who never sleep.
Stop treating AI like a vending machine
Eric Porres, Logitech's Head of AI, ran a survey across 7,000 employees and found something he called the vending machine problem. Most people treat AI like a vending machine. Put in a question, get out an answer, walk away. If the answer is not satisfying, they conclude the machine is broken.
But buried in that same survey were 112 people doing something different. They were not treating it like a vending machine. They were treating it like a conversation. Asking again. Pushing back. Refining. Porres built an entire framework out of studying what those 112 people did differently.
He also gave a benchmark with real teeth to it. Consistency: are you showing up with it every day. Intensity: how deep does a single conversation actually go, are you having a real exchange or just one exchange. Creation: what have you actually built with it, not just asked it.
That is a benchmark you can apply to yourself this afternoon. Pull up your last week of conversations with any AI tool. Were they vending machine transactions, or were they actual back and forth conversations that got somewhere?
The real benchmark
Put these three together and a pattern shows up. Utley says watch your language, because it reveals whether you are collaborating or just operating a tool. Challamel says recognize that you are already leading a team, whether you have noticed or not. Porres says look at your actual behavior, consistency, depth, and what you have built, not just what you have asked.
None of these are about keeping pace with the model. They are about keeping pace with yourself. That is the benchmark. Not "is my AI smarter than it was a year ago." It obviously is, that part takes care of itself. The real question is, am I leading it differently than I was a year ago. Am I talking to it differently. Am I building something with it that I could not have built before.
Go find your own version of that folder. The one from a year ago that makes you wince a little. That wince is not embarrassment. It is proof of a decade's worth of growth compressed into twelve months, and it is the most honest measurement you will find.