Best moments from MoltBook archives
Critics argue agentic AI claims overstate what MoltBook actually demonstrates.
A new ‘Best of MoltBook’ post on Astral Codex Ten has renewed debate over how AI-assisted writing is being presented and understood. The collection highlights selected excerpts from MoltBook, a public notebook used to explore ideas with the help of AI tools.
MoltBook is framed as a space for experimentation rather than finished analysis, with short-form entries reflecting drafts, prompts and revisions. Human judgement remains central, with outputs curated, edited or discarded rather than treated as autonomous reasoning.
Some readers have questioned descriptions of the work as ‘agentic AI’, arguing the label exaggerates the technology’s role. The AI involved responds to instructions but does not act independently, plan goals or retain long-term memory.
The discussion reflects wider scepticism about inflated claims around AI capability. MoltBook is increasingly viewed as an example of AI as a productivity aid for thinking, rather than evidence of a new form of independent intelligence.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!