I wrote an AI textbook — how long until AI can do it better?
AI researcher Nathan Lambert argues LLMs remain weak at long-form technical writing, questioning whether models can autonomously organize scientific knowledge for breakthroughs.
Nathan Lambert describes writing a post-training textbook, Reinforcement Learning from Human Feedback, and finds today's LLMs weak at organizing long-form technical content despite becoming superhuman at coding and math. He notes GPT 5.5 Pro found deep typos across a 200-300 page manuscript while Claude models proved more useful as editors. He argues that compressing knowledge through writing is a prerequisite for autonomous scientific insight and tempers expectations for near-term AI-driven open science.
- LLMs stagnated in long-form non-fiction writing while improving rapidly at coding and math
- GPT 5.5 Pro found deep typos in a 200-300 page manuscript but muddled chapter-level organization
- Claude models proved more useful as editors with better taste and task understanding
- Author links knowledge compression to scientific insight, tempering autonomous AI science claims
- Vendors
- OpenAIAnthropicMoonshot AI
- Organizations
- OpenAIAnthropicMoonshot AI
- AI models
- GPT-4.5Kimi K2GPT-5.5 Pro
From the original article
11 min read at interconnects.ai
Reflections on AI's writing ability and how AI models get more capable.
By Nathan Lambert · interconnects.ai
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.