Reflections on Software Engineering in the Age of AI

Coders are fighting over whether AI is a miracle helper or just overhyped busywork

TLDR: The article says AI now writes a lot of software, but experienced developers still have to catch the big mistakes and make the final calls. In the comments, some cheered it as a game-changing helper for solo builders, while others slammed it as overblown hype that makes coding less enjoyable.

A software engineer’s essay about life in the so-called AI era lit up the comments with a very familiar internet battle: is artificial intelligence making programmers superhuman, or just making them supervise a very confident intern? The article argues that AI can spit out decent code fast, but still needs an experienced human to catch the dangerous stuff — legal problems, security holes, slowdowns, and future conflicts. In other words: the robot may write, but the adult still has to read the homework.

That set off a deliciously split reaction. One veteran coder basically said, “Speak for yourself,” bragging that AI helped him ship a side business he couldn’t have built alone in 2019. That’s the pro-AI camp’s big flex: the tools are turning one-person projects into real money-makers. But the skeptics came in hot too. One commenter flatly declared, “This is not an Age of AI”, calling the whole thing hype from “not-real-programmers” and insisting coding by hand is still better — and, importantly, more fun.

Then came the sneaky philosophical brawl. Some commenters argued the article is romanticizing old-school coding, because the real creative work was always deciding what to build, not typing every line yourself. Another dropped a brainy mini-bomb: maybe programming is really just organizational theory in disguise, with company structure shaping the code more than any magical machine. And yes, there was a side-eye moment too, as one reader seemed less worried about AI than about how the author is somehow both a full-time developer and a novelist. Honestly? Same.

Key Points

  • The article says AI coding ability has improved because it is trained on large amounts of source code and because programming problems often have testable outcomes.
  • keyPoints: The article contrasts a traditional software development workflow with an AI-assisted workflow centered on prompting, reviewing, and merging AI-generated code.
  • The author describes the developer’s role as shifting from directly writing and designing most code to supervising and editing AI output.
  • The article states that AI may miss legal, performance, coordination, and security considerations that affect production software.
  • The article argues that senior developers remain necessary because they provide institutional knowledge and systems-level judgment that AI lacks.

Hottest takes

"I couldn’t have done solo in 2019 without the tools" — AndrewKemendo
"This is not an Age of AI" — lilerjee
"the creativity was largely at the feature definition step" — danans
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.