I am definitely missing the pre-AI writing era

📝 Discussion Summary (Click to expand)

Key Themes

“LLM voice” bias and em‑dash stereotyping > “It’s not that simple. LLMs were trained on lots of writing, and the ‘LLM voice’ resembles in many ways good English prose, or at least effective public communications voice.” – Arainach
Desire for a human‑verified label

“I hope editorial departments everywhere are taking careful notes on the ars technica fiasco. … Agree there’s room for some kind of quick ‘verified human’ checkmark.” – jcgrillo
Decline of writing standards; preference for raw human prose

“When a drunk chef dumps way too much salt into my ramen, the fact that good ramen also contains (more tastefully applied) salt redeems nothing!” – dematz
Tension between AI‑editing and preserving one’s voice

“If you outsource your thinking and skills, your ability to do either atrophies.” – heavyset_go

🚀 Project Ideas

AI‑assisted editor that scores text for LLM‑style markers (e.g., over‑perfect grammar, em‑dash overload) and offers minimal, voice‑preserving rewrite suggestions.
Provides a quantitative “humanity score” to help writers avoid false‑positive AI detection while keeping their authentic voice.

Key	Value
Target Audience	Technical bloggers, non‑native English writers, forum moderators
Core Feature	Real‑time LLM‑style marker detection + voice‑preserving rewrite suggestions
Tech Stack	Chrome extension + Python backend (spaCy, HuggingFace transformer) + VS Code plugin
Difficulty	Medium
Monetization	Revenue-ready: Tiered subscription ($5/mo basic, $15/mo pro)

Quote from radimm: “Writing technical content about PostgreSQL and balancing my voice without sounding like LLM written is genuinely difficult.”
Enables writers to keep authentic tone and reduce false‑positive AI flags.

Web app that polishes technical posts for non‑native speakers while preserving personal style and embedding a “humanity shield” against AI detectors.
Generates edits that keep the author’s voice and issue a verifiable “human‑written” badge.

Key	Value
Target Audience	Non‑native English technical writers, researchers, HN commenters
Core Feature	Context‑aware grammar/style suggestions + AI‑voice shield that removes em‑dash/perfect‑sentence patterns
Tech Stack	Next.js front‑end, Node.js API, GPT‑4 editing endpoint, custom LLM‑marker classifier
Difficulty	High
Monetization	Revenue-ready: Pay‑as‑you‑go credits + $9.99/mo Pro plan

Quote from Arainach: “I am absolutely sick of people calling out writing because they “think it’s LLM‑written”. I reject the call to omit the em‑dash or otherwise alter my own manner of speaking.” - Provides a trusted badge that signals genuine human authorship to skeptical readers.

SaaS that lets writers prove their content was authored by a human via a cryptographic challenge (publish a unique hash of a personal snippet) and receive a verified‑human badge.
The badge can be displayed on forums to counter AI‑slop accusations.

| | Monetization | Hobby |