What's gonna happen to software engineers?

Original Article ↗ Hacker News Discussion ↗

AI coding tools and agents are reshaping software development, raising doubts about whether engineers will be replaced, merely overworked, or pushed into new, more architectural and product-focused roles. Many contributors report higher productivity but also greater fatigue, loss of craftsmanship, and declining code quality as “vibe-coded” AI output floods codebases. Opinions diverge on long‑term impact: some predict a culling of routine web and implementation jobs, others expect more demand for experienced engineers who can design systems, verify AI output, and manage the broader technical and organizational complexity.

Shifting Role of Software Engineers with AI

  • Many see AI as just the next tool in a long chain (punch cards → high-level languages → LLMs); “engineering” persists even if “coding” changes.
  • Expected shift from large coding teams to fewer engineers acting as architects, integrators, and domain experts, with more value on product understanding and domain-driven design.
  • Some predict engineers move toward “alignment” and coordination roles (technical PM/TPM–like), while AI handles more of the typing.

Productivity, Overwork, and Skill Atrophy

  • Widespread agreement that AI makes individuals more productive yet more exhausted; work volume expands rather than freeing time (Jevons paradox).
  • Several describe juggling many more concurrent tasks and feeling constantly overwhelmed.
  • Noted risk of skill atrophy: weaker typing, weaker mental modeling of large codebases, faster decline for those heavily dependent on tools.

Code Quality, Vibe Coding, and Maintainability

  • Strong concern about “vibe-coded” systems: fast output, high bug rates, poor tests, weak understanding of the code by its nominal owners.
  • Some report being called in to “finish” AI-heavy or low-discipline projects that are painful to stabilize or maintain.
  • Others say AI is great for internal tools and small utilities where correctness and security stakes are lower.

Training, Experience, and Verification

  • Open question: how to train juniors when many “junior tasks” are now trivial for LLMs.
  • Experience and “taste” in design, testing, and saying “no” to bad features are seen as key differentiators.
  • Programming is a strong AI use case because compilers/tests provide immediate feedback; other domains lack such verifiers.

Job Market and Economic Impacts

  • Conflicting views:
    • Some expect more dev jobs as software gets cheaper and demand explodes.
    • Others foresee fewer IC roles, especially in low-complexity web-agency work, or a translator-like collapse to “orders of magnitude fewer.”
  • Several tie outcomes to capitalism: productivity gains likely captured as more output per engineer, not shorter weeks or better conditions.

Sentiment and Emotional Impact

  • Perception that HN sentiment has recently shifted more skeptical of AI hype, though this is subjective.
  • Some enjoy AI-augmented craftsmanship and polish; others feel deep loss of pride and increased alienation when LLMs write most of their code.
  • A few express resentment toward past developer arrogance and see current anxiety as a form of reckoning.