Grief and the AI split

Original Article ↗ Hacker News Discussion ↗

Craft vs result framing

  • Many like the “craft lovers vs result chasers” framing, but a lot of commenters see it as an oversimplification or outright wrong.
  • Several argue the real axis is what you care about: enduring quality, correctness, and maintainability vs speed and “good enough for now”.
  • Others say the deeper split is between those who enjoy understanding and designing systems vs those happy to delegate that to tools.

Quality, risk, and maintainability

  • Strong worry that unreviewed or lightly reviewed AI code is just a new way to accrue massive tech debt and hidden bugs.
  • Some report seeing AI-heavy codebases that are huge, verbose “slop” requiring rewrites; others report quality improving when AI is used under strict docs, tools, and human review.
  • There’s tension between contexts: quick MVPs and experiments vs regulated, safety‑ or money‑critical systems where AI shortcuts are seen as unacceptable.

Productivity and workflows

  • Many report substantial productivity gains: faster prototyping, debugging, refactoring, writing tests, resurrecting old side projects.
  • Others say the productivity story is overstated; coding was never the main bottleneck compared to understanding problems, design, and coordination.
  • “Agentic” workflows (agents modifying codebases) divide people: some see them as a new platform layer, others as inherently brittle and non‑deterministic.

Emotional responses and identity

  • A recurring theme is grief, but for different things:
    • Loss of daily hands‑on coding as a paid craft.
    • Loss of professional identity built around being “good at code”.
    • Fear of obsolescence vs excitement about new creative possibilities.
  • Some worry AI use erodes deep focus and reasoning, making people reach for tools instead of thinking through hard problems.

Economic, ethical, and power concerns

  • Concerns about:
    • Job security and the collapse of “knowledge work” as a moat.
    • Proprietary AI platforms capturing dev tooling, with per‑line costs and data/control implications.
    • Hype‑driven abandonment of basic engineering process (reviews, tests) and the long‑term fallout.
    • Use of AI in critical domains (healthcare, finance, infrastructure) without adequate safeguards.

Future of craft and work

  • Some think craft will “move up a level” (requirements, architecture, system prompts); others think the craft dies and so does its paid livelihood.
  • Many expect a messy correction: current FOMO‑driven overuse will blow up, leading to more sustainable, process‑bound AI usage that still needs skilled engineers.