Interconnects uses the experience of writing an AI textbook to examine a narrower question than whether models can produce fluent prose: when can they create useful, structured nonfiction on their own?
The essay argues that large language models are already helpful for filler text, editing, and routine drafting, but that serious explanatory writing requires choices about emphasis, context, and what a reader needs next. Those choices are not just matters of style. They shape whether a technical topic becomes understandable.
The piece is especially relevant as models are increasingly marketed as tools rather than chat assistants. A textbook is a demanding test because it must hold a consistent arc, introduce concepts in the right order, and avoid plausible but shallow explanations.
The practical takeaway is not that AI has no role in long-form technical writing. It is that the hardest work remains editorial: deciding what matters, what to omit, and how to make a reader build durable understanding rather than skim a polished summary.