"AI" Changes Nothing for Devs, Despite what Crappy CEOs Think
AI won’t replace developers soon, but hype lowers salaries, fuels job anxiety, and misleads. Learn why LLMs are tools, not threats, and how to future-proof your software career.
In 2017 I was talking to some “data people” and one of them very confidently bet me 1 million dollars that within 5 years there will be no truckers, AI will take their job.
Fast forward 5 years and the trucker unemployment was as hard to find as my 1 million dollars. But a new “monster” has emerged: no developer will have a job soon.
My old boss predicted that he can replace mid-level devs by the end of this year (I already replied to that). We see articles about how jr devs already lost all ability to code…
I have been working in ML for over a decade now, which gives me both the context and the historical perspective and I would encourage you to breath. Seriously.
Source of Commotion
Money.
Oh wait, I think I need to explain more in this section!
You see it is much easier to raise money for the revolutionary technology than an evolutionary one. So the VCs (who are not that different from other MLM schemes) and “founders” (all of whom see a huge bag) keep inflating the hype bubble, but there is an ugly downside to our profession…
The hype isn’t just about selling a dream—it has real consequences. When VCs and founders push the narrative that AI will replace developers, it creates uncertainty in the job market. Companies hesitate to hire, salaries get squeezed, and junior devs struggle to break in. The fear trickles down, making developers feel like they’re on borrowed time, even when the reality is far less dramatic. And while the people selling the AI revolution cash in, it’s the engineers—the ones actually building and maintaining these systems—who bear the brunt of the uncertainty.
Coders are Harmed
I talk to hundreds of folks trying to get into ML development each month, they are not too different from traditional devs. All are scared about the future.
This makes people take lower salaries (not helped by current adjustment from lower interest rates, tax law, and pandemic exuberance) and hold on to jobs they hate much more than usual.
In the mean-time, the crappy CEOs whose fortunes have been constructed entirely on the backs of developers keep talking about how they will soon fire all of them. Guess what? They won’t.
The problem with AI
Today’s LLMs are stochastic processes that don’t have a grasp of how various concepts relate to each other. These are supposed to replace developers?
How is it possible that at the same time as these approximations are slated to take your job, the crappy CEOs keep screaming about how the only way to do this work is with in-person collaboration between multiple humans?
I have and will continue to use LLMs for coding.
They are terrific at boilerplate code.
They are incredible at “rename this variable in the whole file”
They are perfect for “I need to find the length of the array” (why does every language name their length function differently)
But that is not what a jr or mid-level dev does.
Furthermore, LLMs introduce more bugs into your code that are not trivial to diagnose.
Need to change the flow slightly? Nope, LLMs can only rewrite the whole file from scratch.
But what About Jr Devs
Story time.
My family came to US when I was 13. I took English at school since third grade and had really high score throughout. And you know what?
I could not speak English when we came to US. It took me 6 months to even open my mouth to attempt to speak English.
How could this be?
Did I cheat in my classes? No, I never cared about grades enough to cheat.
Was my teacher bad? No, I had the best teacher in the region.
But quite simply, academic foundations are not the same as practical experience. Newsflash: jr devs could never code just like I couldn’t speak English.
What should you do?
Do not make any career decisions based on LLMs or the words of crappy CEOs.
Do learn what LLMs are good for and use them as an extra tool to help you in your work.