James Holmberg is the co-founder of Vilas AI, a Minneapolis-based studio that designs AI fluency and large language model (LLM) programs for organizations.

I came up through the Twin Cities art world. A Minneapolis College of Art and Design graduate with two exhibitions at the Minneapolis Institute of Art, I’ve worked in software as a service (SaaS) since 2017 alongside coders and product teams. That mix trained me to look for two things: how people actually adopt tools, and how they develop judgment and voice. AI demands both.

I work with these models every day. I have been in them since 2022, essentially right after they went public. By late 2023, forward-thinking organizations were bringing us in to teach employees how to use them in the flow of their jobs. What I have learned since then is not a small detail. It is the whole story.

Related: You don’t need to be a techie to use AI.

Most AI education still looks like a syllabus or a curriculum. It focuses on tool training—here’s where to click, here is a prompt, here is a feature. That approach can get someone started, but it will not make them fluent, because AI is not a calculator and it is not software. It is a medium and an instrument. The output depends on the person using it.

Everyone can have the same guitar, tuned the same way, and you can still immediately hear the difference between B.B. King, Carlos Santana, and Eddie Van Halen. Same instrument. Completely different result. Not because the guitar is different but because the player is. That is what AI is like. A lot of people want to be Eddie Van Halen. Very few want to practice. That is fine. Most of us enjoy music more than we make it. But in the workplace, the gap between “listening” and “playing” is becoming a career gap.

There is a missing piece to what we term “AI fluency” that few want to name: the judgment layer. The ability to steer output, critique it, revise it, add context, and keep your voice. Without that layer, people either get generic results and decide AI is overhyped, or they produce work that sounds like a machine, and trust erodes. People comply with a rollout, but they are never taught how to play the instrument.

Related: Today’s HR balancing act means deploying AI and valuing human judgment.

Leaders often assume they can plug AI in, announce it, and everyone will figure it out. It does not work that way. Everyone has a full-time job. Nothing feels broken. This change is unexpected and, for many people, unwelcome. If you treat AI like a tool rollout, you will get surface use and uneven outcomes.

I foresee a reemergence of liberal arts skills. Habits that value communication, nuance, and the ability to ask better questions. For years, those skills were treated as optional. Now they are the difference between driving the system and being driven by it.

I’m writing this to start a necessary conversation. We are treating AI like a rollout, but it behaves like a medium that is flexible and malleable. The gap between these two facts is where people will get left behind. AI is moving so fast that if you blink, the ground shifts underneath you. Ignoring it will not freeze it in place. It will only widen the gap.

The takeaway: Treat AI like an instrument you learn by playing. Because hoping you can air-guitar your way through this is not a strategy.