Read a sample: 1. The Conversation Comes First
A book club can begin with a question as simple as, “What did you think?” From there, the conversation may move in directions no one planned. One reader notices a detail everyone else missed. Another dislikes a character the group expected to admire. Someone who struggled through the first half explains why the ending changed their mind—or why it did not.
Those moments are difficult to manufacture. They depend on people bringing their own attention, memories, questions, and disagreements to a shared reading experience. A useful tool can help a group make room for those contributions. It can also, if used without much thought, fill the room before members have had a chance to speak.
That is the distinction at the heart of this book. ChatGPT can suggest discussion questions, help organize a reading schedule, or offer starting points for exploring unfamiliar background. None of those tasks requires the club to hand over its judgment. The members still decide what matters, what is accurate, what they want to discuss, and when a conversation has run its course.
The goal is not to make every club comfortable with every use of AI. It is to help each group make deliberate choices that serve its own conversations.
Before deciding whether to use ChatGPT, it helps to ask what the club wants its meetings to feel like. The answer may seem obvious: the group meets to talk about books. But that description leaves room for many different kinds of gathering.
One club may enjoy close reading, returning to particular scenes and considering how the author constructed them. Another may care most about the personal associations a book brings up. A library group might welcome a mix of regulars and first-time visitors, while a group of longtime friends might value the meeting as much for companionship as for literary analysis. Some clubs want a lively disagreement; others want a gentler space where no one feels pressed to defend an interpretation.
A single club can want several of these things at once. Its purpose can change with the book, the members, and the circumstances of a particular meeting. Naming those preferences is not an exercise in producing a perfect mission statement. It is a way to make later choices easier.
Consider a group whose members say they value hearing different reactions, especially from people who did not enjoy the book. If the facilitator brings a list of questions that all assume the novel was moving and successful, the list works against the group’s purpose—even if it is polished, thoughtful, and quick to produce. The problem is not necessarily that a computer helped write the questions. The problem is that the questions narrow the discussion before it begins.
Now imagine the same facilitator asks members to send in one question each, then uses an AI-generated list as an optional source of additional ideas. The facilitator reviews the suggestions, removes assumptions, and chooses only a few that leave room for different responses. The group’s priorities still shape the meeting.
A good starting question, then, is not “Can AI do this task?” It is “What are we trying to make possible for our members, and would this use help?”
AI-assisted preparation happens before or around the discussion. A person asks a tool for suggestions, reviews what comes back, and decides whether any of it is useful. The tool might help a host brainstorm ways to phrase a question or suggest an order for agenda items. Its output is material for a person to consider, not a set of instructions the group must follow.
AI-led interpretation is different. It occurs when a generated answer becomes the source the group is expected to accept about what a book means, why a character acts, or which reading is correct. That can happen even when no one intends it. A facilitator may read a fluent paragraph aloud and ask, “Does everyone agree?” Members then have to respond to the paragraph rather than begin with what they noticed themselves.
The difference is not always a matter of whether a generated passage is read aloud. A group might choose to compare an AI-generated interpretation with its own, treating the passage as a debatable example. That can be a legitimate activity if members know what they are looking at and want to discuss it. The boundary is crossed when the passage is treated as the answer, or when members feel their readings need its approval.
Imagine a club discussing a novel in which a daughter stops calling her father. A generated question might say, “How does the daughter’s decision to end contact demonstrate that she has finally forgiven him?” The sentence sounds ready for discussion, but it contains two substantial claims: that the contact has ended and that forgiveness explains the decision. Members who read the scene as temporary, angry, or unresolved must first push back against the question itself.
A more open version would be: “How did you understand the daughter’s decision not to call her father? Did your view change as you read?” That question does not require the group to reach agreement. It invites readers to bring their evidence and uncertainty into the room.
A person could write either question, and a tool could suggest either one. The important work is the review. Someone needs to notice the assumption, compare it with the book, and ask whether the wording gives members space…