The Second Language.

My cousin John said something on a call a couple of weeks ago that I haven’t been able to put down.

He said learning AI is like getting dropped into a country where nobody speaks your language. However you got there doesn’t matter. You’re just there. No coaches. No tutors. All you can do is listen, watch, and try not to embarrass yourself.

For months you translate every word in your head before you say it. Then one day you walk into the corner store, and a whole sentence comes out of your mouth in the other language. “Can I get a pack of cigarettes and a lighter?” You didn’t build it word by word. You didn’t think about it at all. It just came out.

That moment is the proof. Not the studying. Not the vocabulary. The sentence you didn’t have to think about.

I told John I was going to use that, because it’s exactly right.

What the other essays left out

I’ve written a lot about the line. In The Blast Door I said the window to get across the AI competency line is closing, so get across it. In The Blast Radius I wrote about how far your AI reaches once you stop asking it questions and start instructing it. In The Crossfire I wrote about what happens to the people standing between two others who already crossed.

Every one of those essays tells you to cross the line. None of them tells you how you know you did.

This one does. And the answer isn’t a certificate, a course, or a number of hours. It’s a change in how you think.

Thinking in the other language

They say that when you learn a second language as an adult, there’s a bridge you cross. Before it, you think in your own language and translate. After it, you start to think in the other language.

That’s what happened to us. We rewired our brains, and now we think in AI.

I want to be specific about what that means, because it’s not “I’m good with ChatGPT.”

John described it better than I can. He said he started thinking in very precise, clear steps, because early on you had to, or the model wouldn’t do what you wanted. Every day he was practicing the same move: take a concept and break it down into its individual parts. Do that long enough and you stop noticing you’re doing it. That’s systems thinking. You look at a problem and you see the parts, the order, and the outcome, all at once.

Here’s the part nobody expects. Once that becomes natural, the steps stop mattering.

Once that becomes natural, the steps stop mattering.

Outcomes, not steps

For a hundred years, work has been built around tasks. Each person does their piece. A leads to B, B leads to C, and eventually somebody arrives at G. The whole company is designed around who does which letter.

AI doesn’t need the alphabet. You might tell it the tasks so it understands the job and doesn’t come off the rails. But it doesn’t have to do A, B, C, D, E and F to get to G. You tell it where you’re going. It gets you there.

That’s what the rewiring really is. You stop thinking in steps and start thinking in outcomes. A process that took 30 steps can take 3, and the fluent person sees the 3 first.

People who haven’t crossed still describe the steps. People who have describe the result.

When’s the last time you built a prompt?

I asked John that on the call, because I’d watched him do this four times in an hour without noticing.

A couple of years ago, prompting was the skill. People sold courses on it. You built long, careful instructions and hoped the model followed them.

I don’t do that anymore. I can’t remember the last time I built a prompt. I say what I want, and the system I’ve built does the rest. It knows who I am, it knows how I work, and it has a skill for most of what I ask. It’s a smart build on the backside now, not a clever sentence on the front.

That’s the acid test. If you’re still crafting the perfect prompt, you’re still translating. If you just say what you want and it happens, you’re thinking in the language.

Why the old way fights it

I’ll say this in defense of the people who are resisting. We’ve done it the same way for a hundred years, since the Industrial Revolution. When people came into the workplace, you needed procedures. Methods. Documented policies. Things to keep everyone in line. Their dad did it that way. Their grandfather did it that way.

So when a company tries to put AI on top of its old procedures, it usually keeps the procedures and bolts a model onto every step. That’s translating word by word. I wrote about the bigger version of this in The Blast Radius: the reach comes from instructing, not from asking it to follow the old script faster.

This is radically different, and there’s no shortcut around it. You’re going to have to learn to think differently about this.

Too far

One warning, and John gets credit for this one too.

He joked that the day he stands up in a meeting and says he’s “quibbling a bit,” something has gone wrong. If you talk with these models every day, you can start to sound like them.

That’s not fluency. That’s an accent. The goal isn’t to talk like the machine. It’s to think clearly enough that the machine understands you the first time.

The goal isn’t to talk like the machine. It’s to think clearly enough that the machine understands you the first time.

On your own clock

So how long does it take?

I don’t think it has a time frame, and I’m not going to give you one. Everybody crosses the line on their own clock. What I do know is that a new way of thinking takes time to form, the same way a language does. You don’t get there by reading about it. You get there by living in it, every day, until one day a sentence comes out that you didn’t have to build.

And the door is wider than it has ever been. John put it this way: getting to expert used to mean swimming a mile-wide river. Years of school, years of apprenticeship, and most people never started. Now that river is about two inches wide, and the swim costs twenty dollars a month.

You still have to get in the water.

You still have to get in the water.

Second nature

That’s where this series lands. The consequence clock says you work at the speed you’re checked, not the speed you’re capable of. The blast door is closing. Your blast radius is how far your work reaches. The crossfire is what happens to the people standing still between two circles. And this is how you know you made it across: you stopped translating.

When John and I talked about it, I said the only thing that came to mind.

It’s our second nature now.

If you want a place to start living in the language, that’s what One Window is: your job, run from one place, with the AI doing the work around the work and you keeping the send button. The software is free.

If you’re learning the language yourself, start there: kerzie.ai/one-window

If you lead a sales team and want the whole team thinking in it, we build that with you: kerzie.ai/services/sales-teams