First screen, captured 5 days ago.

On AI writing

What HN thinks

A thread of humans proving they're human by typing two hyphens instead of an em-dash.

AI prose is verbose and empty
LLM writing is padded, hedging and slow to read, burying the point under filler and invented jargon, and summaries miss details.
AI writing from colleagues feels disrespectful
AI-generated PRs, specs and docs from coworkers are hard to care about and break the assumption that the author put in effort; some teams have banned it internally.
Why AI lags at writing
Commenters debate whether writing is unverifiable and thus hard to train, simply under-prioritised by labs versus coding, or limited by forward-only token prediction without real editing.

Consensus: Most agree AI writing is easy to spot, wordy and less pleasant to read than human writing, though they differ on whether that gap is permanent.

Recommended in the comments

Gravitas: Communicate with Confidence, Influence, and Authority (book): Recommended as a phenomenal read on communicating clearly.
fthr ↗ (tool): Shared by a commenter who builds tooling for AI fiction.

Summarised by Claude from the comment thread 3 days ago.

Points: 72
Comments: 61
Posted: 2026-10-20 by dbolkunov
Category: Ai ml

ai-writing llm authorship writing-style stylometry

All captures:

Looks similar

Screenshot of cymerys.com: I'm becoming AI-blind

I'm becoming AI-blind (cymerys.com)

462 points by rcymerys 2 months ago | 468 comments

Personal blog post by Rafal Cymerys describing how he struggles to focus on documents at work, feeling increasingly 'blind' to AI-generated text.

ai-generated-content llm reading-comprehension workplace-communication

Screenshot of shloked.com: Claude’s memory architecture is the opposite of ChatGPT’s

Claude’s memory architecture is the opposite of ChatGPT’s (shloked.com)

408 points by shloked 1 year ago | 216 comments

Blog post analyzing how Claude's memory system works, retrieving from raw conversation history on demand, and contrasting it with ChatGPT's memory design.

claude chatgpt llm-memory anthropic

Screenshot of matthewrocklin.com: My article on why AI is great (or terrible) or how to use it

My article on why AI is great (or terrible) or how to use it (matthewrocklin.com)

151 points by akshayka 9 months ago | 208 comments

Screenshot of storiesby.ai: Why I'm no longer writing stories with AI

Why I'm no longer writing stories with AI (storiesby.ai)

51 points by andreyk 3 years ago | 45 comments

Screenshot of ssp.sh: If you start writing today, there's no way to know if you can write without AI

If you start writing today, there's no way to know if you can write without AI (ssp.sh)

6 points by zazuke 1 month ago | 5 comments

A personal knowledge-base note by Simon Späti arguing that human-read text should be human-written and comparing AI-bloated writing, prompts, and raw writing.

ai-writing llm writing naval-ravikant