A counterpoint to the hype

My thoughts and favorite points of someone else’s writing from the web:
The most common take on AI in programming these days, and the one that is plastered like graffiti all over LinkedIn, especially by non-developers; is that it’s essentially going to replace programmers. Or that using it can make a developer 2x or 10x or 100x as productive.
This developer, who only goes by Jj, is clearly skeptical of that viewpoint. I share some of that skepticism, though less than I used to. But more importantly I think this is a viewpoint that people are hesitant to share because of the AI hype and not wanting to seem like they’re stuck in their hidebound ways.
Chapter 1: My Coworker, The Programmer
Section titled: Chapter 1: My Coworker, The ProgrammerA shell of a man. More of a parrot than a person. My boss, a true believer in the sacred rite of Pair Programming, chained myself and this “programmer”-colleague together like conjoined twins from different planets. We shared a keyboard, but not a brain. Lord, not even close.
“Hold up. I’ve got an idea. Gimme the keyboard real quick.”
An idea. Yes. The same way a toddler has “an idea” to stick a fork in a wall socket. I was halfway through constructing something beautiful; a lean, elegant piece of logic that sliced through complexity like a blade through butter-and here he comes, pounding the keyboard like it owes him money, pasting in code he Frankensteined from a stack overflow comment written by an Uncle Bob disciple in 2014.
Did he know what our system did? No. Did he read the ticket? Absolutely fucking not. Did he feel confident mutating global state with reckless abandon? He absolutely fucking did.
“Why aren’t you pairing more with him? He types twice as fast as you.” Of course he does. So does a cat having a seizure on a mechanical keyboard. But that doesn’t mean it should be writing production code.
What it isn’t
Section titled: What it isn’tThis isn’t about tools or productivity or acceleration. It’s about the illusion of progress. Because if that programmer-if that thing, that CREATURE-walked into your stand-up in human form, typing half-correct garbage into your codebase while ignoring your architecture and disappearing during cleanup, you’d fire them before they could say “no blockers”.
Copilot isn’t a peer who can fly the plane if you pass out. It’s just the ghost of a thousand blog posts and cocky stack-overflow posts whispering, “Hey, I saw this once. With my eyes. Which means it’s good code. Let’s deploy it.” Then vanishing when the app hits production and the landing gear won’t come down.
Things AI is good at
Section titled: Things AI is good at- Writing a small function in an unfamiliar language
- Finding potential flaws in an architecture design
- Giving you a starting point to build on
- Providing a close-enough description of an unfamiliar codebase
- Surprisingly quick at listing out your blind spots
Chapter 3: You as a Programmer
Section titled: Chapter 3: You as a ProgrammerI like to code. Not supervise. Not hover over a synthetic lobotomized chatbot like some drooling silicon intern.
“But I just use AI for boilerplate!” you whimper, clutching your Co-Pilot subscription. Listen to yourself. If you’re writing the same boilerplate every day like some industrial-age cog monkey, automate it yourself. Write a library. Invent a macro.
“I’m just moving fast!” you say. Yeah. Straight off a cliff. AI isn’t helping you build something novel. It can’t. It only knows what’s been done before. It’s autocomplete with a superiority complex.
When you outsource the thinking, you outsource the learning. You don’t know your code. You’re babysitting it.
The problem isn’t just laziness. It’s degradation. Engineers stop exploring. Stop improving. Stop caring. Eventually, you’re living in a cathedral of technical debt, and every user pays.
Chapter 4: The Computer as a Machine
Section titled: Chapter 4: The Computer as a MachineYou’re human. You can know - actually know. But the bot? The bot? The bot has no clue. The bot has zero understanding. But it will copy the advice of a sweaty stranger from an ‘08 StackOverflow thread.
If you want to sculpt the kind of software that gets embedded in pacemakers and missile guidance systems and M1 tanks - you better throw that bot out the airlock and learn.
This is a profession. Take pride in your life’s work.
Chapter 5: Conclusion
Section titled: Chapter 5: ConclusionThe real horror isn’t that AI will take our jobs. It’s that it will entice people who never wanted the job to begin with. People who don’t care for quality. It’ll remove the already tiny barrier to entry that at-least required people to try.
And what’s worse, we’ll normalize this mediocrity. Cement it in tooling. Turn it into a best practice.
Defer your thinking to the bot, and we all rot.