Project ideas from Hacker News discussions.

The safest job from AI may be writing

📝 Discussion Summary (Click to expand)

Theme 1 – AI‑generated writing misses human nuance and defaults to cliché
- “The reason LLMs use the same cadence and cliches is that their RLHF does not emphasize writing well, the way it emphasizes, say, coding. And the reason for that should be pretty obvious: coding is where the big money is at.” — enraged_camel
- “No, they use ‘the same cadence and cliches’ because they inflate a short and ambiguous prompt into long and specific prose by making statistical assumptions about what best fills in the gaps… a popular model will always become exhaustingly common, painfully familiar, and cliche.” — swatcoder

Theme 2 – AI works best as a writing copilot (research, outline, editing) rather than a replacement for human authors
- “AI today is most effective when it's not vibing, but rather copiloting a skilled operator… The trick is to get them to make suggestions at a higher level than just writing alternatives… feeding them through GPT and Claude… drastically increased the speed at which I can knock out a completed piece.” — tptacek
- “The workflow that I've adopted… has been to leverage AI to help with the research and outline process. Once I'm happy with the structure I go and I write what I need to write.” — ramijames

Theme 3 – The effect on writing jobs is uneven: low‑end, routine copywriting is at risk while high‑end, creative, or non‑screen work remains relatively safer, yet clear retraining paths are lacking
- “Workaday copywriting is dead (was moribund, now has been shot in the head).” — curiinor
- “Exactly. And it's not only about whether a specific job is safe from AI, but whether it's safe from the effects of those that aren't safe: competition from laid off people and downward wage pressure… at the moment we have no clue of what workers should be retrained to. Not writing.” — gruntled‑worker
- “The safest jobs are the ones that can't be done behind a screen. I'm a technician and there's no robot that could replace me and there won't be for a long time.” — ThrowawayTestr


🚀 Project Ideas

Generating project ideas…

VoiceGuard: AI Writing Style & Empathy Coach

Summary

  • Real-time feedback on AI‑generated drafts to reduce cliché, boost empathy, and preserve author voice.
  • Core value proposition: Helps writers produce higher‑quality, human‑sounding content faster while retaining creative control.

Details

Key Value
Target Audience Professional copywriters, content marketers, novelists, and anyone who uses LLMs for writing assistance
Core Feature Style & empathy scoring engine that highlights clichés, tonal flatness, and suggests rewrites aligned with the writer’s target audience
Tech Stack Python backend (spaCy, HuggingFace Transformers, custom RLHF‑style reward models), React + TypeScript frontend, Docker for deployment
Difficulty Medium
Monetization Revenue-ready: subscription SaaS ($15/mo per seat)

Notes

  • mrweasel noted: “Copywriting struggles with the issue that businesses know they need it, but they don’t value it.” VoiceGuard gives writers a tool to demonstrably improve quality, making the value visible to employers.
  • ilaksh suggested: “They can use human writers and also LLM‑based multi‑step scoring systems with analytical guidelines.” This product implements exactly that idea for solo writers.
  • Addresses the complaint that LLMs produce “predictable, cliché‑ridden prose” (swatcoder) by providing actionable, stylistic revisions.

HumanWrite Marketplace

Summary

  • Curated platform connecting businesses with vetted human writers, offering transparent pricing, portfolios, and measurable impact (e.g., engagement lift).
  • Core value proposition: Makes it easy for companies to find and trust high‑quality human copy, counteracting the tendency to undervalue writing.

Details

Key Value
Target Audience SMBs, marketing agencies, startups needing blog copy, ads, newsletters, or technical documentation
Core Feature Matching algorithm based on writer expertise, style fit, and past performance; includes quality guarantee and basic analytics on content impact
Tech Stack Node.js/Express API, PostgreSQL database, React web app, Stripe Connect for payments
Difficulty Medium
Monetization Revenue-ready: transaction fee 15% per project (minimum $30)

Notes

  • mrweasel observed: “businesses know they need it, but they don’t value it.” HumanWrite surfaces writer credentials and past results, making the value explicit.
  • TFNA lamented that even respectable publishers now use minimal editing; a marketplace of skilled humans offers an alternative to low‑effort AI slop.
  • Provides a practical outlet for writers worried about AI displacement (gruntled‑worker, brendoelfrendo) by highlighting demand for human‑crafted copy.

AI‑Provenance Detector

Summary

  • Browser extension / API that scores the likelihood a piece of text was generated by an LLM, using linguistic features and explainability flags.
  • Core value proposition: Enables editors, publishers, and platforms to quickly spot AI‑generated slop and maintain content authenticity.

Details

Key Value
Target Audience Editors, publishers, content platforms, educators, and anyone needing to verify text origin
Core Feature Real‑time detection score with highlighted indicators (e.g., repetitive n‑grams, low perplexity, style uniformity) and optional human‑review workflow
Tech Stack Python (scikit‑learn, HuggingFace Transformers for perplexity), FastAPI for API, Chrome/Firefox extension (JavaScript, React)
Difficulty High (requires continual updates as models evolve)
Monetization Hobby

Notes

  • swatcoder warned: “Any given model will always have some distinct implicit voice … Users … will always be generating cliché slop.” The detector makes that voice measurable.
  • ilaksh noted the difficulty of telling human vs. AI legalese; this tool gives analysts a concrete signal.
  • Provides a tangible response to the widespread frustration about AI‑generated prose drowning out authentic voice (spijdar, GEAR discussions).

Read Later