Some friends of mine are saying that this might be the next big thing on IT. As someone who still thinks of starting a new career in this field (just got my associate level certificate in system analysis after a long time being a journalist who works in PR), especially considering that my company is becoming a bottomless pit, should I start building my stack on this before things get really sour at my current job?
In my experience, fixing someone elses mess is no fun at all, be it a person or AI generated.
It is good, solid work tho if that’s what you’re looking for (again regardless of whether it’s human or AI)
I would say focus on becoming an experienced, well-rounded developer who is familiar with all of the tools at their disposal. The “vibe cleanup” angle may be popular for a few years, but 10 years from now it will be something else.
Just get hired at any company that’s heavily pressuring employees to use Claude or whatever and employees are drinking the Koolaid. You’ll spend much of your time fixing LLM fuckups.
Maybe, but imo it will be a short-lived niche until people learn to use AI properly.
When I use AI for coding I try to catch its mistakes as I go along. I rarely use the agent mode and just let it loose. I’ve done that as an experiment a couple of times, with hilariously bad results each time.
People will adapt and learn when you need to babysit the AI and the problem will eventually diminish.
Just my take.
Wasn’t that the path of many senior devs long before ‘AI’ ate the world? I mean, getting called in to not just re-factor but re-architect an existing shitty code base seems in line with a job for someone who was there when the deep magic was written.
Honestly, cleaning up legacy shit code is already a thing, it’s called consulting.
It won’t be a dedicated career field. The AI bubble is at an all time high, and it works now. What people will realize, is that there is more to a piece of software than just the initial code / prototype. AI is amazing at prototyping, it’s fast and it gives the dopamine rush of bringing something online fast. What AI is not good at is actually creating production ready code. Maintainability, security, operations of AI slop code suck. Massively. Adding features by AI to a vibe coded codebase sucks, and all of this is amplified exponentially if the person vibe coding does not know their shit.
The question on vibe coding is not if it will break, but when. And when it breaks, it does not matter if it’s AI or just bad code. It’s a broken app that needs fixing, and that’s just your regular software engineering job.
ever heard the phrase ‘shoot the messenger’ ?
