LLM as Reading Companion

mechanism

An LLM works best as a responsive margin note: it supplies missing context, unknots confusing passages, and helps pursue questions while the book remains the primary encounter.

The useful reading assistant may be the one that never reads the book for you. Instead, ChatGPT sits beside the page, ready to recover a forgotten reference or explain the sentence that stopped you.

E1

Keep the book in the driver’s seat

Reading creates questions at unpredictable moments: a term is unfamiliar, an argument skips a step, or earlier context has faded. A general-purpose LLM can answer at that exact point of friction. This shortens the detour between confusion and renewed attention without requiring the author to anticipate every reader’s gaps.

The division of labour matters. The text supplies the argument, voice, and evidence; the model supplies responsive scaffolding around them. Used this way, it helps you remain inside the encounter long enough to form sharper follow-up questions rather than abandoning the difficult passage.

E1

A margin note is not the book

The mechanism breaks when the assistant becomes a substitute for the text. Its explanation may smooth away ambiguity, emphasis, or style that matters—and may itself be wrong. Keep its answers provisional: useful for orientation, but answerable to the passage in front of you.

Open a companion thread

Tomorrow, keep one chat alongside your book. When you stall, paste only the relevant passage and ask one precise question: “What context am I missing?”, “Which step in this argument is implicit?”, or “Explain this term without summarising the chapter.” Then return immediately to the text and test the answer against it.

E1

Episodes that teach this