LLMs are a technological leap without a ramp

The career opt-out you might not know you're making

James J. Boyer of Engineering Leadership Bites is back again with another guest post after last year’s well-received piece: No more coding vibes in the efficiency era

I’m staring at LLM output, confused. I’d just furiously launched a stream-of-consciousness prompt into my terminal window, and it didn’t do what I expected.

Oh. Claude wasn’t running. 🙃

I sat there for a second, realizing my bash prompt had absolutely no idea what I was talking about. I laughed, fired up the Claude CLI, and did what I was apparently trying to do.

I was less embarrassed and more fascinated.

Technological shifts don’t usually jump-scare you; there’s a ramp.

Before git, I versioned with folders. project-v2, project-final, project-seriously-this-is-the-last-one. We all had a system. Mine was terrible, and yours was too. The folder habit immediately felt primitive when commit history arrived. I could explore without consequence and undo without panic. I retired the habit deliberately.

Docker was similar. “It works on my machine” wasn’t something I let myself say anymore. A clean before and after with a conscious crossing.

Delayed reflexes

This time was different. I didn’t decide to reach for AI every time I had a parasitic thought, quietly slurping away at my focus, getting fat and greedy in the back of my skull. I didn’t make a deliberate choice to stop pounding out notes in Obsidian of half-formed ideas I’d never look at again, and instead start dumping nutritional slop into a contextual stew I had neither the time nor energy to schlep over to the stove yet.

It just became where my brain goes now.

The way I write and how I work through something have changed, too, whether that’s code or otherwise. It’s more important for the idea to vacate my head than to clinically dissect it, so I’ll clean it up later. This is messier, but more honest. I couldn’t tell you when that happened. I just noticed that it had one day.

The farmer probably can’t tell you exactly when he stopped thinking about the horse plow either. At some point he just stopped reaching for it. That wasn’t a decision so much as a fact he discovered.

What’s different this time is that the previous shifts gave you enough runway to notice yourself adapting. With cloud and mobile, the ramp was gradual. The old habit had time to deprecate cleanly, and you got to watch the transition.

This one compressed all of that. I typed a prompt into an empty terminal. The reflex had been there for a while, I just wasn’t home when it moved in.

Just missed it.

Real people will care about this

I could not explain Web3 to my mom. Not because she isn’t smart (she is), but because she’s stubborn (that’s where I get it from). She wants to understand the “why”. But the value proposition of Web3 requires you to already believe a set of premises that most humans will never, ever care about. It’s a people problem. AR and VR suffered the same fate. The mainstream consumer looked at a headset, or a bulky pair of sunglasses, and shrugged. End of story.

When AI showed up, I was skeptical. Same initial posture as the “busts”. I went looking for the exit ramp. Tested the edges. What is this actually bad at? What does it excel at? Where are the seams?

I found some early, and I kept coming back to challenge it. It had gotten better and the seams kept closing.

I was surprised too. When ChatGPT first impressed me, I exclaimed to my then-disinterested wife: “this is going to change the fucking world because real people will care about this.”

This was the first time we had the conversation we thought we should’ve been having with Siri over a decade ago. It passed the test.

Opting out is a career mortality event

The companies building on top of this are not selling a vision of the future that you have to build faith in. They’re selling you something you can use right now, today, to do your actual job faster. The people adopting it are pulling away from the people who aren’t, and that gap is growing rather than shrinking.

I watched a colleague shrug this off a couple years ago: “I’m gonna sit this one out. Someone else will figure it out, and I’ll follow along.” I understood the impulse because change is hard. Paradigm shifts are exhausting, and if you get deep enough into your career, you’ll want to make the same call.

But I couldn’t. I’d already gone looking for reasons to dismiss this, and hadn’t found enough of them.

What upsets me about this is that it isn’t turning into “another tool in the toolset”. I wanted it to be that. Instead, it’s becoming something closer to: this changes the job. Maybe the entire industry. And people who opt out aren’t just missing a skill, they’re opting out of what the job is becoming.

I’ll say it plainly because everyone else is hedging: The people choosing to sit this one out are making a career-ending decision.

Outcome engineering

CEOs, founders, and investors writing the checks don’t care about the craft of software engineering. I know people who’ve built their entire identity around that craft. I have genuine respect for it. But the people with the vision, the capital, and with the ambition to build something that actually matters are chasing outcomes.

They will hire the person who builds faster and ships faster. If that person uses AI to do it, nobody is going to apologize for it. That’s the part that stings. Not the technology, but the implication.

There’s a person in every neighborhood who builds furniture in their garage because they love it, not to compete with IKEA. Sometimes that’s enough.

You can be that person too! Make your peace with what that means though. Keep your skills sharp, build things you’re proud of, and enjoy the craft for what it is.

But don’t pretend you’re also going to thrive in a market that is structurally changing around you. Don’t be the person who shows up to a job interview in 2027 and explains that you chose not to engage with tools that have become table stakes, expecting the person across the table to nod, thump their chest twice, and say “we respect that.”

They won’t. They have someone else to talk to in twenty minutes.

The opportunity in front of us all

The good news is that the window isn’t closed yet. It’s still early enough that getting started isn’t embarrassing. These tools are genuinely weird. Hilariously imperfect. You’ll struggle with them at first. But that’s normal, and it means that you’re learning.

What isn’t fine is my colleague’s decision. The conscious, deliberate choice to wait for someone else to blaze the trail so you can follow later. That’s a reasonable strategy when you’re observing a slow-moving body in space, light years from any real and imminent threat to your safety or job security.

But your colleagues who are embracing this are approaching the event horizon at a quickening pace. They’re soon to be spaghettified by the sheer momentum and rate of change, body and limbs going where their minds, their habits, their instincts, must dare to follow. The folks watching from the window of the ship have signed a resignation letter that has yet to be drafted.