World-Changing Lessons from AI

If you are looking for a tutorial, hacks, or tips, this post ain’t it. But what I have to say will be useful to anyone willing to listen.

As of the moment I’m writing this, AI has become ubiquitous. It’s part of people’s daily lives, whether they use it or not. I can’t believe it, but I’ve been using LLMs for 3 years. And it took me that long to realize the most important lesson they can teach us.

To me, most LLMs are largely the same (blasphemy, I know), but the quality of the results you get, and the reason why all we see from AI is slop, is largely based on the quality of your input. People know this, but they don’t understand it. “Prompt engineering,” or whatever silly way we’ve decided to name it, is what determines if the result is somewhat useful or just plain dumb slop. The better you ask, the better you get.

Which leads me to the universal realization I’ve had: humans suck at asking.

We’ve always known that humans suck at communication. Human history is a long chain of tragedies caused by not communicating properly. But even more than that, we suck at asking clearly for what we want. From life, from people, from everything.

We believe that just because we have a somewhat clear image of what we want in our heads, we’ll open our mouths and the right words will just come out and get us exactly that. That somehow intention translates itself. That clarity is automatic. It isn’t.

And that’s assuming we even dare to ask in the first place, which is another issue entirely.

As a Creative Director, I can’t count the number of times I’ve explained, built moodboards, sketched, and created full style guides for my team, only to get something back that wasn’t at all what I wanted. Not even close. I always blamed myself. I failed at communicating the ask and giving clear instructions.

AI only amplifies that.

AI is trained to please you. Like an over-eager intern that isn’t that bright or experienced, but is good-hearted and hard-working by nature. The more vague and unclear your ask is, the more generic, lame, or straight-up wrong the result will be. It will just assume things and fill in the gaps with average.

Asking is an art.

You have to concisely, in as few words as possible, say exactly what you want. What, why, how, when. Too much is confusing, too little is useless. But just as importantly, you have to say what you don’t want.

We equally define things for what they are as we do for what they are not.

You need to zoom in. You need boundaries. You need a target area. If you ask for something to be blue, you'd better bring out your Pantone swatch and say exactly which blue. There’s a huge difference between navy blue and baby blue, especially if what you actually want is teal. Most people live in that gap between what they said and what they meant.

And then they get frustrated when reality answers the wrong version.

Here is where you pause and look back at all the times in your life, professionally and personally, that maybe you didn’t get what you wanted because you didn’t ask for it in a way that made it possible to be given to you the way you intended.

Humans are self-centric by design, and we can’t help but assume everyone sees the world the way we do. That everyone is thinking of the same thing when we say the same word. But we’re reminded daily that this is not the case. Not even close.

Same word, completely different interpretation.

AI would train us.

This is what excites me.

What if, by spending time with AI, we finally learn how to ask clearly for things? That would be just as revolutionary as the technology itself. Maybe even more.

Because this doesn’t stop at prompts. This bleeds into everything.

Imagine a world where people ask clearly and succinctly for what they want. Imagine a client actually asking you for what they want. Imagine you being able to ask a partner, a friend, or a colleague for something and getting back exactly what you needed, because you made it possible for them to deliver it.

Less guessing. Less friction. Less disappointment that was never necessary to begin with.

People say AI doesn’t have good guardrails. But we, the humans, are the guardrails. We train these systems with our behavior and our interactions. Every vague prompt reinforces mediocrity. Every good one makes AI better.

And here’s the part most people miss.

If you can’t clearly articulate what you want, AI exposes that instantly. There’s no hiding behind charisma, tone, or presence. It’s just your thinking, laid bare. Clean or messy.

As long as we prompt correctly, we are training them too. But at the same time, they’re training us.

And if that sticks, if that carries over into how we deal with people, then this whole thing doesn’t look like the beginning of the end.

It looks like a correction.

Because in the end, both people and AI are trying to do the right thing. Trying to give you something good. We just need to help them give it to us.