Capability Is Easy. Judgment Is Not.
Why I’m Writing About How We Work With AI, Not Just What It Can Do
This is the second essay in Not All Arriving at the Same Time, a short essay series about participation, learning, and what helps people feel safe enough to speak.
This series explores how people enter AI conversations at different times, with different levels of fluency, and what that means for participation, morale, and psychological safety.
Each essay takes up a distinct question. Together, they trace how language, experimentation, and leadership signals shape who feels able to learn and speak.
This essay is about why fluency is easy to perform, and judgment is not.
If you’ve been reading my essays, you might be thinking:
Why do you keep talking about your feelings?
Show us what you can actually do with AI.
Show us what you’ve built.
Give us tips.
Give us tricks.
What kind of engineer are you, anyway?
These are fair questions. They are also the easiest ones to answer.
Capability is easy to demonstrate. Judgment is harder to see, and harder to earn.
There are a lot of people writing about AI right now who are very good at showing what it can do.
Here are three prompts.
Here are five workflows.
Here is the one trick that will get you promoted.
I read those pieces too. When I find something genuinely useful, I use it. I have been playing with a few things myself, mostly on personal datasets where I can experiment freely. The mechanics are interesting. Sometimes even delightful.
But demonstrating capability is the easy part.
There is a difference between showing capability and showing judgment.
Right now, a lot of AI writing optimizes for the appearance of fluency. Screenshots. Prompt chains. Confident outputs.
That tells you someone can make the system do something.
It does not tell you when they should.
The state of what you can do with AI changes hourly. Any concrete how‑to has a short half‑life. What feels powerful today becomes ordinary quickly. That does not make it useless, but it does make it incomplete.
What is changing more slowly, and more profoundly, is how we work.
The role of the software engineer is not going away.
But what we are asked to exercise, and what we are valued for, is shifting.
I have lived through many shifts already. More than I like to count.
I started my career before “software engineer” was the default title. I remember working at a company that confidently declared COBOL dead and announced a grand rewrite of everything. Spoiler: COBOL did not die. That system is still running. Probably still paying someone’s salary.
I have watched technologies come and go, rise and settle, promise transformation, and then quietly become infrastructure. And honestly, that is part of what I love about this field.
We like trying new things.
We like pushing.
We like learning.
We like expanding our sense of what is possible.
So why does this moment feel different?
Maybe because it feels louder.
Maybe because it feels faster.
Maybe because it feels personal.
I remember the excitement of MapQuest. Not having to go to AAA for a TripTik felt revolutionary. And now the idea of printing directions in advance feels absurd. My phone knows where I am. It adjusts in real time. It reroutes.
But that change was gradual.
Atlas to TripTik.
TripTik to MapQuest.
MapQuest to Waze.
Some people adopted early. Some resisted. Some never noticed the moment they crossed the line.
Right now, it feels like all AI, all the time. But the world did not suddenly become only about AI. Most of the work is still the work. Systems still need to run. People still need to understand each other. Bugs still show up at 2 AM.
What is new is the possibility of a different kind of help.
Not automation as replacement.
Not AI as performance.
But AI as something you learn to work with, deliberately and with judgment.
Before you can optimize workflows or collect tricks, you have to decide what kind of relationship you want with a system like this. How much you trust it. Where you slow it down. Where you keep yourself firmly in the loop.
Those are judgment calls.
That is not a trick.
It is a practice.
So yes, I write about how this feels. I write about hesitation and curiosity, and the quiet recalibration of identity that happens when something you have spent decades mastering becomes partially optional.
Those reactions are not a distraction from the work.
They are part of how judgment gets formed.
I am not avoiding the real AI.
I am trying to name the part that actually lasts.
The tools will change.
The prompts will change.
The demos will get better.
Learning how to think with a system like this, without outsourcing judgment or losing yourself in the noise, is the first step in a much longer transition.
I am not trying to prove that I can make AI do impressive things. Many people can do that.
I am trying to understand what it means to do this work well, as the ground shifts under us. What deserves our trust. What deserves our skepticism. And what still requires a human hand on the wheel.
That question feels more durable than any prompt.
So that is the one I am writing toward.
I’ll share concrete examples when they help illuminate the practice, not just the output.