June 8, 2026
Ctrl-Alt-Existential Crisis
Replies to comments on my "LLMs are eroding my career" post
Programmers are spiraling as AI eats the ‘special skills’ they thought made them safe
TLDR: A programmer says artificial intelligence is already replacing the expert knowledge that once made his job feel secure, and the comments exploded. Supporters say he’s bravely saying the obvious, while skeptics argue tech has survived this kind of panic before — making the fight less about code and more about whether this time is truly different.
A programmer’s gloomy follow-up to his viral “artificial intelligence is eroding my career” post has turned into a full-on comment-section cage match. His core confession hit a nerve: the hard-won knowledge that once made him valuable now feels increasingly askable on demand by chatbots. That sparked a wave of readers saying, basically, “Yep, he’s right, and people are in denial.” One of the loudest cheers came from users arguing that the real world only cares whether things work, not whether they were built with craft, care, or human pride. Ouch.
But not everyone was ready to hold a funeral for programming jobs just yet. Some commenters rolled out the classic comeback: isn’t every tech panic ‘different this time’? Others pushed back on the author’s claim that demand for software has a ceiling, saying people have declared software “done” before and been hilariously wrong. So yes, the thread had everything: fear, coping, and a little economic philosophy.
The spiciest part? The author didn’t just complain — he admitted he’s already become what he jokingly hates to call an “AI-native engineer,” using multiple bots to check and re-check work while quietly maneuvering around rushed workplace habits. That confession made readers split hard: some praised him for being brutally honest, while others saw it as proof that the future belongs to the people willing to “ride the beast.” The mood was half support group, half apocalypse watch, with a side of grim workplace comedy about drowning in long documents, cutting corners, and trying not to let the robot drive the money truck.
Key Points
- •The author wrote a follow-up post to address comments on a viral earlier post about LLMs affecting their software career.
- •LLMs are described as not fully automating highly specific areas such as local tax regulations, while still reducing the value of some accumulated domain knowledge.
- •The author says newer models, agent-friendly documentation, and an AGENT.md file improved AI agent performance on workplace tasks involving accounting and ledger details.
- •The post describes practical workarounds used in a fintech environment to slow down risky AI-accelerated delivery, including generic design docs, extra E2E testing tickets, and splitting sensitive work into more task cards.
- •The author says they actively use agentic tooling and multi-model adversarial code review, but worries that continued model improvement could commoditize software engineering.