AI won't replace human devs anytime soon

Original Article ↗ Hacker News Discussion ↗

Role of AI in Development Workflows

  • Many see LLMs as useful tools (autocomplete, boilerplate, documentation, diagrams, “second brain”) rather than replacements for developers.
  • Several experienced devs report marginal or task-specific gains; others say tools like Cursor/Claude make them dramatically faster on greenfield or small projects.
  • Some compare this shift to IDEs or cloud: didn’t remove devs, but changed the work and roles (e.g., sysadmin → “DevOps”). Others say LLM impact is less clearly positive than IDEs were.

Limits and Failure Modes of LLM Coding

  • Common complaints: hallucinated APIs, wrong patterns, inconsistent changes across a codebase, inability to robustly update SDKs, CI, or handle non-trivial bug reproduction.
  • Many stress that checking AI output can be slower than writing code yourself, especially in large, messy, brownfield systems.
  • Some report AI-generated code that’s brittle, incoherent, or security-sensitive, creating “landmines” for teams that lack expertise to review it.
  • LLMs are often likened to junior devs: can produce snippets and tests, but need supervision and don’t truly “understand” systems or downstream effects.

Impact on Skills, Juniors, and Careers

  • Strong concern that juniors leaning heavily on AI won’t build deep skills, becoming stuck on easy tasks and unable to handle complexity later.
  • Others argue that juniors are already used for “grunt work,” which AI can increasingly handle, making entry-level roles more vulnerable.
  • Counterpoint: juniors bring fresh perspectives and domain insights that AI can’t; real value is in domain understanding, architecture, and organizational problem-solving.

Job Market, Economics, and Hype

  • Several comments suggest layoffs and “AI will replace devs” narratives are primarily about lowering labor costs and exerting downward pressure on salaries.
  • Some see AI displacing certain creative and low-skill roles already; others note data is anecdotal and contested.
  • A common view: if/when AI can truly replace devs, it will also replace many computer-based jobs, implying broader economic upheaval, not just in software.

Non-Programmers and Low-Code/No-Code

  • Non-devs report that LLMs let them build CRUD apps, glue APIs, and prototype quickly—similar to historic low-code tools and spreadsheets.
  • Consensus: this expands who can “write software,” but also increases demand for skilled devs to clean up, scale, and secure what non-experts create.