A16Z: Prompts and rules won't yield good AI writing
ivan_bezdomny · x · 2026-08-26
A good piece from A16Z on AI and good writing makes it clear we can’t get good writing from an LLM based on magic prompts, tweaking output token distribution, or general rules. It suggests stopping asking if it was written by AI, and start asking whether it’s doing its job.
More from Apps
- Peter Yang open-sources /fuck-cancer, an AI skill helping patients advocate for themselves — petergyang · 2026-08-26
- OpenAI desktop app adds WebMCP support, enabling Codex to deploy apps to Sites — OpenAIDevs · 2026-08-26
- ChatGPT Desktop App Browser to Support WebMCP — OpenAIDevs · 2026-08-26
- Elicit recording: Multi-step workflow for rigorous decision-making with AI — elicitorg · 2026-08-26
- ChatGPT Extension Adds Multi-Browser Support, Tab Context Awareness — OpenAIDevs · 2026-08-26
- AI virtual influencer shows consistent growth, reveals agent harness — RichardsonDx · 2026-08-26