December 19, 2025
Artisanal Bugs vs Robo-Coders
Engineers who dismiss AI
‘Cheating’ or changing the game? Coders split over AI helpers
TLDR: A once‑anti‑AI coder now says modern tools can write code while humans still do the job. Comments erupt: some show real wins, others call it garbage, and many worry about accessibility and losing skills—turning AI coding into a workplace identity clash and a productivity arms race.
Engineers are throwing elbows over AI coding tools, and the comments are pure soap opera. The author admits they once refused to touch bots—preferring artisanal bugs—and even called a coworker “cheating” for using Claude Code. Then they dropped a post claiming AI can write your code, it can’t do your job, and expected “robots replace us” panic. Instead, skeptics stormed in with “can’t,” “trash,” and “it writes my bugs,” daring anyone to prove modern tools aren’t just confident nonsense.
Cue the receipts. dkdcio says the vibe is changing and links to the prior piece: the post. fwsgonzo flexes a two‑week emulator build—AI as setup crew, human as finisher. On the flip side, harimau777 drops a gut‑punch: ADHD makes constant bot back‑and‑forth impossible; they fear being forced out if the industry demands it. That sparks a mini‑battle over inclusion versus “adapt or fall behind.”
Meanwhile, ryandv wants proof, begging for a “Destroy All Software”‑style screencast on gnarly legacy code—bring receipts or it didn’t happen. And plastic‑enjoyer worries about skill atrophy: using AI feels like losing sharpness. The mood? Identity crisis meets speed run. Fans say today’s tools are the IE11 vs Chrome jump; purists say it’s still auto‑generated spaghetti. The only consensus: everyone’s watching—and arguing.
Key Points
- •The author’s 2022 test of ChatGPT led to skepticism due to incorrect results.
- •A coworker completed a project in one week using Claude Code, highlighting potential efficiency gains.
- •Modern AI coding tools like Claude Code and Cursor are described as capable of understanding full codebases, refactoring multiple files, and iterating toward completion.
- •The author acknowledges AI tools remain imperfect (bugs, over-abstraction, occasional API hallucinations) and require review.
- •The author argues engineers who refuse to try modern AI tools risk falling behind and urges them to test current tools.