Years ago I wanted to make a point about knowledge and perceptions to a client. To do so I picked up a Sharpie, and drew a Chinese character on a sheet of paper. As I did so, I told them about how they work, the symbolism of the strokes, and that grouped together they make an idea, a joint meaning. There are 214 of these building blocks and thousands of combinations. I held up my character, with the confidence of knowledge and asked "Do you know what this means?"

All I got was maybe a little mumbling, and mostly heads turning left and right. Soft no's.

They were expecting a bright answer from me, instead they got: "It's nothing. Meaningless. It is a scribble. I made it up, I don't know Chinese." Chuckles ensued.

Nobody could tell, and there really was no reason they should have. This is what happens when we look at something from the far side of a skill or knowledge we do not have. It may be a poem, a work of art, an image, a logo or a headline. We think it is fine. We can even be talked into believing it is significant. Because we try to find the known in the unknown. Think Rorschach, the famous ink blot tests. But in truth, we really do not know what we do not know. We are blind to our own ignorance.

"We do not know,
what we do not know."

‍

What does this have to do with AI? Well, first let me say:

"Yay!". Because the tools genuinely are magic. New ones almost every week, and work that was never possible is possible now. Forget the speed for a second and look at what actually comes out. It may be a vibe coded functioning prototype of an app. Or a thirty second spot that looks ready to traffic, and it is only a comp. The Animatic. The Rip-O-Matic. Are dead. The output may be a Hollywood style film with effects that used to need dozens of people and a budget to match. Things that were seemingly impossible or out of reach for many of us are now all within reach. Yet again, much of our industry has been democratized, and with democratized access, the measuring rod rises.

Then I say:
"Boo!". The temptation that comes with it. We now want one person to be the whole orchestra.

And here is where it gets challenging for those tasked with saving money, often because we don't know if it is a Chinese character or a scribble: the answer may be fewer people, or for many it is more about different skilled people.

The answer depends on your goals, and should look more like a question, a series of process questions: What is the task? How long does it take right now? What skill does it actually need? and Who or how is that skill provided? What are the costs associated with the skill at the level we need? If we get that understanding right, then we can refine the process and make decisions so that the product doesn't suffer, on the contrary, it is enhanced. That is duller than "AI replaces the department" and saves us "Bocoup Dollars"(1), but I believe it is the way to look at it, not only because it works, but to ensure success in the long game.

One of the many reasons the distribution of work matters is because of what happens when you are the one doing it. This is a different blindness from the scribble, and in some ways a worse one, because it catches us on the things we do know. Closeness fosters near-sightedness. Your mind corrects as you go. You stop seeing what is missing, because you think you know what it is supposed to look like. An old drawing trick is to hold the work up to a mirror. Seen flipped, different angle from how you drew it, allows all your mistakes to jump out at once. You trick the brain to bypass its corrections. We can check our own work, but it is always good to have a fresh set of eyes. That is why even the best authors have editors. That is not a weakness. It is just how making great things works, and it is the reason good teams put additional trained eyes on anything that matters.

AI has a similar blind spot, only that it comes at us all at once. You get apparently finished results in seconds. Confident. Sharp. Done. Looks good. Sounds good. Makes sense. But none of that means it is good, and none of it means the person looking at it is trained enough to tell. Remember the scribble? You don't know what you don't know.

"Are you looking for a short term solution, or longer term market value and growth?"

‍

Without even getting into the sea of sameness or the thousands of versions of the same error, we need to consider an honest watch-out. Many people reach for these tools to cut cost, others to supplement with a skill they don't have, and very often to involve fewer people and run shorter timelines. Those are real uses and not wrong, but they may be chasing different goals. Is it short term solution, or longer term market value and growth? Quick buck or brand building? This is not trying to provide a reason to slow down. It is an invitation to be clear about where the real ROI is, and where we want the gains to actually get us.

The numbers back this up. Salesforce found that 75 percent of marketers have adopted AI, and 84 percent admit they are still sending the same generic(2), one-way campaigns they always did, just faster. The ones seeing real returns are not the ones with the newest tool. They are the ones who did the unglamorous work first, unified their data and built real judgment into what relevant even means. Distribute the skills. Refine the process. The gains are real, and they already go to whoever is doing just that.

The skills needed are not disappearing, they are executed differently, under different titles and with different tools. The faster and more democratic the tool, the greater the importance of someone who can see what was missed.

How to build the team for this, how to bill fairly for work that takes an hour instead of a week. We are learning, have more questions than answers, and our clients even more. Let's use the tools to feed our superpowers, what they are genuinely great at, and learn to keep the human in the loop for what they are not. There is a real place in the middle, and every team needs to find their own version of it. The first step is to really understand your process and the skills it requires.

The fun part: We not only get new toys every day, but we also get to reinvent what a team looks like.

(1) Bocoup. Beaucoup. Glad you were paying attention.

(2) Salesforce, State of Marketing (10th Edition), 2026.

‍