I hate AI side projects

Original Article ↗ Hacker News Discussion ↗

Raising the Bar and Changing Norms

  • Many expect the flood of low-effort AI projects to eventually push norms higher: only projects with real time investment, novelty, or depth will be worth posting.
  • Suggested heuristic: if it took only a couple of hours with an LLM, it’s probably not “Show HN” material; multi‑day or multi‑week, novel work still is.
  • Counterpoint: some feel the bar per unit time hasn’t changed much; AI just compresses timelines.

Signal vs Noise and Evaluation Difficulty

  • Core frustration: spaces that once had mostly interesting work (Show HN, Product Hunt, dev communities) feel “drowned” in shallow, “vibe‑coded” AI apps and blogposts.
  • AI can cheaply generate convincing READMEs, tests, and docs, making outward quality signals less reliable and increasing the work needed to assess projects.
  • Others argue the problem isn’t side projects or AI, but our filtering; we need better ways to judge effort and quality in a high-volume environment.

Value and Impact of AI

  • Supporters: AI democratizes building; non-experts can now ship things, learn faster, and complete projects that would have been impossible given their time/skills.
  • Experts report big productivity wins on boilerplate and exploration, but stress that architecture, algorithms, and deep understanding remain human.
  • Critics say AI mostly makes things faster/easier, not better; much output is low-quality “slop,” and many users would barely miss AI if it disappeared.
  • Concrete disagreements over AI search (e.g., Google’s AI mode): some find it faster and good enough; others see frequent, confident errors and a net downgrade in information quality.

Motivations, Authenticity, and Gatekeeping

  • One camp sees many AI side projects as thin, money‑chasing products with little rigor, crowding out unique, authentic, exploratory work.
  • Another camp pushes back on gatekeeping: there were always bad side projects; new builders deserve the joy and learning even if they rely heavily on AI.
  • Some note AI has forced a personal reckoning: realizing one’s work isn’t “special” but embracing coding for personal satisfaction rather than impact.

Curation, Prompts, and Platform Responses

  • Several propose better curation mechanisms (weighted voting, curated tiers, “qualified curators”) to restore “differential amplification of quality.”
  • A concrete proposal for AI-generated Show HNs: require submitters to share prompts or AI interaction logs, since that’s the real “source” code.
  • This is seen as promising by some but logistically hard: prompts are conversational, numerous, and intertwined with discarded ideas; full disclosure may be infeasible, though partial summaries might still help.