February 28, 2026
Ctrl+Alt+Career?
Everything Changes, and Nothing Changes
Coders face an AI takeover while the crowd debates jobs, taste, and who foots the bill
TLDR: AI may soon write most code, pushing engineers toward high-level design and coordination. Commenters joke about a product-manager takeover, worry rising AI costs could sideline small creators, and some refuse AI entirely—making this pivot both thrilling and terrifying for the industry.
Software’s new era has the crowd buzzing: the article claims AI will soon write 90–100% of code, turning engineers into conductors of bots rather than craft coders. Cue Deep Blue dread and a chorus of spicy takes. One camp, like jongjong, is chill: “always an architect, not a coder,” meaning the job was never about typing, it was about taste and outcomes. Another camp is panicking-turned-comedic: dventimi quips we’re all becoming product managers (PMs), sparking jokes about the “PM apocalypse” and replacing keyboards with roadmaps.
Then come the philosophers. AreShoesFeet000 bets on a comeback for symbolic AI—the old-school rule-based stuff—as the real revolution, poking holes in the “everything is neural networks” hype. Meanwhile, magpi3 drops the money bomb: AI feels cheap today, but what happens when the bill arrives? Micro-entrepreneurs riding AI might get priced out, and commenters fear a future where only big corporations can afford the robot workforce.
Not everyone is buying the hype: brendanyounger flatly refuses to use large language models (LLMs), preferring decades-old dreams of better user experiences over mass-produced code. Throughout, folks agree on one thing: taste and architecture matter more than lines of code. The memes? “AI interns who never sleep,” “tabs vs spaces is dead; it’s budgets vs taste now,” and a lot of nervous laughter.
Key Points
- •The article argues that LLMs are transforming software engineering from handcrafted coding to AI-generated code, making code production cheap and abundant.
- •It claims that rejecting LLM usefulness is no longer tenable and that AI could write 90–100% of code for many teams; some engineers at leading AI labs already avoid manual coding.
- •Core engineering practices—small stacked diffs, continuous deployment, automated testing, and easy rollbacks—remain vital and become even more important with AI-generated code.
- •Engineering “taste” and intuition are shifting toward architecture; juniors must develop architectural judgment early, as frontier models can produce clean code when guided (e.g., via AGENTS.md).
- •The author is confident these dynamics hold for the next <5 years but is uncertain beyond, noting the term “Deep Blue” (coined by Adam Leventhal and Simon Willison) to describe engineers’ anxiety about automation.