Using AI Generated Code Will Make You a Bad Programmer

Tool vs. craft: what is the job of a programmer?

• One side emphasizes “I’m hired to solve business problems,” so AI codegen is just a faster bus to the destination; hand-writing abstractions is mostly ego or hobby.
• Others argue there’s a “craft” dimension: system structure, clarity, and understanding every line are an expression of expertise that AI undermines.
• Several distinguish “art code” for fun from “engineered code” for work; AI threatens the former far more than the latter.

Effects on skill, learning, and juniors

• Many worry heavy reliance, especially by juniors, will prevent them from acquiring deep skills or the ability to catch AI’s mistakes; the pipeline to future seniors looks unclear.
• Analogies: using solutions manuals in math, taking the bus instead of running, CNC vs. hand tools, loom vs. weaver. You can get more done, but hands-on practice atrophies.
• Others use AI explicitly as a tutor: generate examples, then read, adapt, and debug them; report learning Rust, frontend, ESP32, even Spanish faster this way.
• Some argue reading code is inherently harder than writing; if juniors don’t write enough, their reading fluency will stagnate.

Productivity vs. quality and maintainability

• Pro-AI commenters claim 2–5x or 3–4x productivity boosts, rapid refactoring of “slop” with few regressions, and strong help on boilerplate and integrations.
• Critics see duplicated code, security flaws, hallucinated APIs, and huge, unreviewable diffs; they argue this leads to fragile systems no one fully understands.
• Several use AI only for snippets, then manually integrate and review, treating it like Stack Overflow on steroids.

Jobs, power, and the future of work

• Some see AI as an inevitable force-multiplier; not using it will make you unemployable, as with earlier shifts (C vs. asm, IDEs, VB6).
• Others fear broad white‑collar replacement, especially for junior roles, and concentration of power/wealth among AI owners. Unions are mentioned as a possible counter.
• There’s disagreement whether AI has already “taken” junior jobs or is just an excuse amid macroeconomic changes.

Are LLMs “stochastic parrots”?

• One camp insists they are sophisticated pattern matchers with no real understanding; powerful but not intelligent.
• Another argues we don’t truly understand how LLMs work or what “understanding” means; given some complex successes, dismissing them as “just parrots” is seen as denial.
• Some note that the term is often used rhetorically to shut down nuance rather than analyze concrete capabilities and limits.