L1
·
Quiz
·
Lab
L2
·
Quiz
·
Lab
L3
·
Quiz
·
Lab
L4
·
Quiz
·
Lab
Module Test
Module 6 · Lesson 1

The Human–AI Co-Writing Partnership

How writers and AI systems actually collaborate — and what that division of labor looks like in practice.
What does it really mean to write with an AI, rather than just use one?

When Robin Sloan began writing his novel Moonbound, he built a custom language model trained on his own sentences — not to have the machine write the book, but to have it resist him.

He wanted a collaborator that knew his rhythms well enough to push back. "The model would complete my sentences in ways that were almost right," he wrote in a 2023 essay, "and the wrongness was generative." He accepted perhaps one in twenty suggestions. The rest he rejected — but rejecting them clarified what he actually wanted.

Co-Writing Is Not Dictation

The most common mistake writers make with AI is treating it as a faster keyboard. They describe what they want, accept what the model produces, and move on. This is output extraction, not collaboration — and it tends to produce prose that feels hollow precisely because no genuine friction occurred.

Authentic co-writing involves iterative exchange. The writer proposes something imperfect; the AI responds, often inaccurately; the writer revises in response to that inaccuracy; the loop repeats. The value lives in the loop, not in any single output.

In 2022, the author K Allado-McDowell published Pharmako-AI — a book created entirely through alternating passages between a human and GPT-3. Allado-McDowell described the process as "a kind of séance," in which the AI's unexpected associations pulled the human writer into territory they wouldn't have reached alone. The book was published by a literary press and reviewed in the London Review of Books. It is a real collaborative artifact, not a novelty.

Three Models of Human–AI Partnership

Practitioners have settled into roughly three working models, each with different implications for creative ownership:

The Sculptor Model

The AI generates large volumes of raw material. The human curates, selects, and shapes. The author's intelligence is in the editing eye. Ross Goodwin's 2018 road novel 1 the Road — the first AI-authored book published commercially — worked this way: a neural net produced the text; Goodwin shaped the journey.

The Sparring Partner Model

The human writes primary drafts; the AI responds with alternatives, objections, and extensions. Sloan's approach fits here. The human accepts almost nothing but uses rejection to sharpen their own choices.

The Turn-Taking Model

Human and AI alternate passages, chapters, or voices. Allado-McDowell's Pharmako-AI exemplifies this. The final voice is genuinely plural — identifiable neither as purely human nor purely machine.

The Ghost-Drafter Model

The AI produces complete first drafts which the human substantially rewrites. This is the most common commercial use. The human's voice dominates the final product; the AI's contribution is speed and reduced friction at the blank-page stage.

The Division-of-Labor Problem

Every co-writing partnership must answer one practical question: who is responsible for what? Current AI systems excel at surface fluency, genre conventions, structural scaffolding, and variation generation. They are consistently weak at sustained causal logic across long documents, authentic emotional specificity, and anything requiring lived embodied experience.

Experienced co-writers exploit this asymmetry deliberately. They use the AI to draft the connective tissue — transitions, scene-setting, expository passages — while writing by hand everything that depends on felt knowledge: grief, shame, the precise texture of a specific place.

Real Example — Jennifer Lepp (2023)

Romance novelist Jennifer Lepp (writing as Leanne Leeds) publicly documented her process of co-writing with Claude in 2023. She found the AI reliable for plot-chapter outlines and dialogue scaffolding, but returned to solo drafting for any scene requiring genuine emotional subtext. Her series output tripled; her reviews remained stable. The partition was deliberate and consistent.

Key Terms
Iterative ExchangeA co-writing workflow in which human and AI take successive turns responding to each other's output, with value accumulating across cycles rather than in any single output.
Output ExtractionUsing an AI to produce finished text for direct use — the opposite of collaborative iteration; treats the model as a faster tool rather than a creative partner.
Productive WrongnessThe phenomenon where an AI's incorrect or unexpected response clarifies the writer's actual intention better than a correct response would have.
Lesson Takeaway

The quality of AI co-writing depends almost entirely on the quality of the human's engagement with what the AI produces — including its failures. Writers who treat every wrong response as information tend to produce better work than those who simply retry until the output is acceptable.

Lesson 1 Quiz

The Human–AI Co-Writing Partnership · 3 questions
Robin Sloan trained a custom language model on his own sentences primarily to achieve what effect in his writing process?
Correct. Sloan explicitly described accepting roughly one in twenty suggestions — the value was in rejection producing clarity about what he actually wanted, not in accepted outputs.
Not quite. Sloan's stated goal was resistance and productive friction, not automation or direct publication. He accepted about one in twenty AI suggestions.
K Allado-McDowell's Pharmako-AI is an example of which co-writing model?
Correct. The book was created through alternating passages between Allado-McDowell and GPT-3, producing a voice described as genuinely plural.
Not quite. Pharmako-AI used alternating passages — the Turn-Taking Model. This is distinct from having the AI generate bulk material for curation or draft full chapters.
According to the lesson, where are current AI systems consistently weakest in co-writing contexts?
Correct. The lesson explicitly identifies sustained causal logic across long documents, authentic emotional specificity, and anything requiring lived embodied experience as consistent AI weaknesses.
Those are actually AI strengths. The consistent weaknesses identified are sustained causal logic across long documents and authentic emotional specificity drawn from lived experience.

Lab 1 — The Sparring Partner

Practice iterative exchange — use the AI's "wrongness" as a writing tool

Your Task

You'll practice the Sparring Partner model. Write a short passage — a scene opening, a character description, a story premise — then ask the AI to continue or respond. Your goal is not to accept the AI's output, but to use whatever it produces (including wrong turns) to sharpen your own revision.

After at least three exchanges, describe one way the AI's imperfect response helped you clarify what you actually wanted.

Try: "Here's my opening paragraph: [paste your text]. Continue it in a direction that surprises me — even if it's wrong for my story." Then respond with what the wrongness revealed.
Co-Writing Lab
Sparring Partner Mode
Welcome to the Sparring Partner lab. Share a passage you're working on — an opening line, a scene, a premise — and I'll respond with something that might be wrong for your story. Your job is to use that wrongness to clarify what you actually want. What are you writing?
Module 6 · Lesson 2

Prompting for Character Voice

How to instruct an AI to maintain consistent, distinctive character voices across a collaborative narrative.
How do you keep an AI from flattening every character into the same voice?

In 2023, writing instructor Lincoln Michel published an essay in Esquire noting that stories generated by large language models exhibited a characteristic failure: every character spoke in the same slightly elevated, slightly helpful register. The villain explained their motives with the same courtesy as the hero. The grieving mother used complete sentences.

Michel called it "the voice collapse" — and identified it as the central challenge for any writer trying to use AI for character-driven fiction.

Why AI Defaults to Voice Collapse

Language models are trained to produce text that is coherent, clear, and unlikely to confuse. These are precisely the qualities that make for bad dialogue. Real characters interrupt, deflect, contradict themselves, speak in incomplete thoughts. They have verbal tics, class markers, regional syntax. A model optimized for clarity will sand all of this away unless you specifically instruct otherwise.

The solution is not simply to tell the model "write distinctive voices." That instruction is too abstract. Effective character-voice prompting operates at the sentence level. You must give the model specific examples of how the character sounds, not descriptions of personality.

The Voice Specification Method

Practitioners who successfully maintain character voice across long AI collaborations typically provide what might be called a voice specification — a compact block of information the model can reference. It contains four elements:

  • 1
    Sample Dialogue (3–5 lines): Actual lines in the character's voice, not summaries. "I don't know what you want me to say" works. "She speaks hesitatingly" does not.
  • 2
    Syntactic Rules: Sentence length tendency, use of fragments, frequency of questions, whether they interrupt themselves. Be specific: "Short sentences. Often no subject. 'Went home. Didn't call.'"
  • 3
    What They Don't Say: Topics they avoid, emotions they never name directly, things they deflect with humor or aggression. Silence is character.
  • 4
    A Counter-Example: Show the model how the character would not speak. "She would never say: 'I feel very hurt by what you did.' She might say: 'Fine. Whatever.'"
Documented Practice — Sudowrite (2022–2024)

The AI writing tool Sudowrite, used by a reported 30,000+ fiction writers as of 2024, built its "character voice" feature around exactly this structure — asking users for sample dialogue before generating in-voice text. Internal user research found that providing even two sample lines reduced voice-collapse complaints by a measurable margin compared to personality-description prompts alone.

Maintaining Voice Across a Long Collaboration

A single session can maintain voice reasonably well. The problem is persistence: language models have context windows, and as a document grows, early voice specifications drift out of active context. Writers working on novel-length projects with AI collaboration have developed several practical responses:

The Voice Card

A short document — 150–300 words per major character — kept outside the main draft and re-injected at the start of each new writing session as a system prompt or preamble. Treated like a style guide for the collaboration.

The Drift Check

Every 2,000–3,000 words, the writer pastes a recent AI-generated dialogue passage and asks: "Does this match [character]'s voice card? What has drifted?" The AI identifies its own deviations.

Key Terms
Voice CollapseThe tendency of AI-generated text to render all characters in the same coherent, neutral register, eliminating the syntactic and lexical distinctiveness that makes characters feel real.
Voice SpecificationA structured prompt block containing sample dialogue, syntactic rules, avoidance patterns, and counter-examples used to anchor an AI to a specific character's speech patterns.
Voice CardA persistent document holding a character's voice specification, re-injected into new sessions to counteract context-window drift in long collaborative projects.
Lesson Takeaway

Telling an AI a character is "gruff" or "witty" produces the AI's generic version of gruff or witty. Showing it three lines of actual dialogue — especially including what the character would never say — produces something far closer to a specific, irreplaceable voice.

Lesson 2 Quiz

Prompting for Character Voice · 3 questions
What did Lincoln Michel call the tendency of AI-generated fiction to give all characters the same voice register?
Correct. Michel coined "voice collapse" in a 2023 Esquire essay to describe how AI systems flatten every character into the same coherent, slightly elevated register.
Not quite. Michel's term was "voice collapse" — the phenomenon of all characters speaking in the same slightly elevated, slightly helpful register regardless of who they are.
Which element of a Voice Specification is most consistently identified in the lesson as what AI personality descriptions fail to capture?
Correct. The lesson explicitly states that effective character-voice prompting operates at the sentence level — real dialogue lines, not personality summaries.
Not quite. The lesson's core argument is that abstract personality descriptions ("gruff," "witty") produce generic results, while actual sample dialogue anchors the AI to specific, irreplaceable voice.
What is the primary purpose of a "Drift Check" in long collaborative writing projects?
Correct. A Drift Check involves pasting recent AI-generated dialogue and asking the model to compare it against the voice card — addressing the context-window problem in long projects.
Not quite. A Drift Check specifically addresses voice consistency — pasting recent dialogue and asking the AI to identify where it has deviated from the character's voice specification.

Lab 2 — Voice Specification Workshop

Build a voice card for a character and test it against AI voice collapse

Your Task

Build a Voice Specification for a character you're working on (or invent one). Include: (1) three sample dialogue lines, (2) one syntactic rule, (3) one thing they'd never say. Then ask the AI to write a new line in that voice — and evaluate whether it succeeded or collapsed.

Run at least one Drift Check: after getting a response, ask "Does this match the voice spec I gave you? What drifted?"

Try: "Here's my character's voice spec: [your spec]. Write three lines of dialogue where this character reacts to finding out they've been lied to. Then self-check: does it match the spec?"
Co-Writing Lab
Voice Specification Mode
Welcome to the Voice Specification lab. Share your character's voice spec — sample dialogue, syntactic rules, what they'd never say — and I'll try to write in that voice. I'll also help you identify when I've drifted from it. Who are we working with?
Module 6 · Lesson 3

World-Building and Continuity Management

How collaborative writers maintain consistent fictional worlds across extended AI sessions — and what breaks without deliberate systems.
How do you keep an AI from forgetting — or contradicting — the world you've built together?

In 2023, fantasy author Brandon Sanderson discussed AI writing tools in a podcast interview, noting that the fundamental problem he saw with using them for his Cosmere universe was world-state memory: a model given his books as context would still, within a single session, contradict established facts about his magic systems.

He wasn't alone. A 2023 survey of genre fiction writers using AI tools (conducted by the Science Fiction and Fantasy Writers Association) found that continuity inconsistency — characters knowing things they shouldn't, geography changing between scenes, established rules violated — was the most frequently cited frustration, cited by 71% of respondents.

Why Continuity Breaks in AI Collaboration

Language models do not maintain a state machine for fictional worlds. They have no internal model of what is true in this story separate from the text currently in their context window. If the text establishing that the city of Valdren was destroyed three chapters ago has scrolled out of the context window, the model will cheerfully have a character visit it.

This is not a bug that will be patched; it reflects something fundamental about how sequence-predicting models work. They are not simulating a world — they are predicting what text is likely given the text they can see. World-state management is therefore always the writer's responsibility.

The World Bible Approach

Professional writers working with AI on long-form fiction have converged on a practice borrowed from television production: the world bible. In TV, a bible is the document writers' rooms use to maintain consistency across seasons and writers. For AI collaboration, it is a structured external document injected into each session.

An effective AI world bible has three layers:

  • 1
    Hard Facts (the "Never-Contradict" list): The non-negotiable rules of the world. Geography that doesn't change. Power systems. Who is alive and who is dead at the current story moment. Injected verbatim at session start.
  • 2
    Current World-State: The mutable facts — where characters are, what they know, what has just happened. Updated after every session. This is where continuity drift originates, so it must be explicitly maintained.
  • 3
    Open Questions: Unresolved plot threads, unexplained elements, deliberate mysteries. Explicitly flagging these prevents the AI from inventing answers to questions you want to leave open.
Real Practice — Aedan Peterson (2023)

Fantasy writer Aedan Peterson, who documented his AI co-writing process in a series of posts on the Substack "AI Fiction Lab" in 2023, maintained a 4,000-word world bible that he pasted into every session as a system preamble. He reported that continuity errors dropped from roughly 3–4 per session to under 1 per session after adopting the practice. He also introduced a session-end ritual: asking the AI to list every new fact introduced in that session so he could update the bible immediately.

Consistency Checking Techniques

Even with a world bible, errors occur. Experienced collaborative writers use two additional techniques:

The Contradiction Audit

After generating a scene, ask: "Review this passage against the world facts I've given you. List any contradictions with established geography, timeline, or character knowledge." The AI is good at catching its own contradictions when explicitly asked to look for them.

The Knowledge-State Check

Before writing a scene involving revelation or discovery, ask: "At this point in the story, what does [character] know about [topic]? What don't they know?" This prevents characters acting on information they shouldn't have — one of the most common continuity failures.

Key Terms
World-State MemoryThe internally consistent record of what is true in a fictional world at any given story moment — something humans maintain but language models must be explicitly supplied with each session.
World BibleA structured external document injected into AI sessions to supply the world-state facts that would otherwise drift out of the model's context window.
Contradiction AuditA specific prompt asking the AI to review a generated passage against a supplied world bible and list factual inconsistencies.
Lesson Takeaway

An AI cannot remember what it cannot see. Every serious collaborative fiction project needs an external document — a world bible — that is re-injected each session. Managing that document, not the AI's memory, is what maintains continuity.

Lesson 3 Quiz

World-Building and Continuity Management · 3 questions
According to the 2023 SFWA survey cited in the lesson, what percentage of genre fiction writers using AI tools reported continuity inconsistency as their most frequent frustration?
Correct. The lesson cites 71% of survey respondents identifying continuity inconsistency — characters knowing things they shouldn't, geography changing, established rules violated — as their top frustration.
Not quite. The survey found 71% of respondents cited continuity inconsistency as their most frequently experienced problem with AI writing tools.
Why does continuity break in AI collaboration, according to the fundamental model of how language models work?
Correct. The lesson explains that models are not simulating a world — they are predicting likely text given the text they can currently see. World facts that have scrolled out of context cease to exist for the model.
Not quite. The core reason is architectural: models predict text from visible context, not from an internal world simulation. Facts outside the context window simply don't exist for the model during that generation.
What was the "session-end ritual" Aedan Peterson described adopting for his world bible maintenance?
Correct. Peterson's session-end practice was to ask the AI to enumerate all new facts introduced, allowing him to update the world bible before the next session.
Not quite. Peterson's ritual was to ask the AI to list every new fact introduced during that session — a simple way to keep the world bible current without manually reviewing the full transcript.

Lab 3 — World Bible Builder

Practice constructing a world bible and running a contradiction audit

Your Task

Build a minimal world bible for a fictional world (real project or invented). Include: (1) three Hard Facts that cannot change, (2) a Current World-State with two or three mutable facts, (3) one Open Question you want to stay unresolved.

Then ask the AI to generate a scene set in that world. After reading the scene, run a Contradiction Audit: "Check this scene against the world bible I gave you. List any contradictions."

Try: "Here's my world bible: [paste it]. Write a 150-word scene where two characters meet at [location] and discuss [topic]. Then audit it for contradictions with the bible."
Co-Writing Lab
World Bible Mode
Welcome to the World Bible lab. Share your world's facts — hard rules, current state, open questions — and I'll write a scene set in it. Then I'll audit the scene against your bible and flag any contradictions. What world are we building in?
Module 6 · Lesson 4

Ownership, Attribution, and Ethics of Co-Authorship

The real legal decisions, disclosure debates, and ethical questions writers face when publishing AI-collaborative work.
When AI writes part of your story, what do you owe your readers — and yourself?

The U.S. Copyright Office issued a landmark decision in the case of Zarya of the Dawn — a graphic novel by Kristina Kashtanova that had been granted copyright, then partially revoked.

Kashtanova had written the text and arranged the images. But the images had been generated by Midjourney. The Copyright Office ruled that the text and arrangement were copyrightable — because they reflected human creative choices — but the AI-generated images themselves were not, because they lacked human authorship. The line was not "did a human touch this?" but "did a human make the expressive choices?"

The Current Legal Landscape

As of 2024, U.S. copyright law does not protect AI-generated content as such. What can be protected is the human's selection, arrangement, and modification of AI output. The more a human author has shaped, chosen, and transformed the AI's raw output, the stronger the copyright claim on the resulting work.

This has practical implications: a writer who accepts AI draft paragraphs verbatim and publishes them has a weaker copyright position than one who substantially rewrites them. The Zarya decision made this concrete. Kashtanova retained copyright on everything she made expressive choices about; she lost it on everything she simply generated and accepted.

Key Case — Getty Images v. Stability AI (filed 2023)

Getty Images filed suit against Stability AI in 2023, arguing that Stable Diffusion was trained on Getty's copyrighted images without license. The case raised the separate but related question of what rights AI companies have in training data — distinct from what rights human users have in AI output. As of mid-2024 the case remained ongoing. It is one of several that will shape the legal context for all AI-collaborative creative work.

The Disclosure Debate

Copyright is a legal question. Disclosure is an ethical one — and the creative writing community has not resolved it. Three positions have emerged among professional writers:

Full Transparency

Disclose AI assistance specifically — which elements, which tools, what percentage of word count. Argued for on the grounds of reader trust and market honesty. Allado-McDowell's Pharmako-AI takes this position; the AI co-authorship is part of the book's explicit identity.

Tool Equivalence

AI is a writing tool like spell-check or Scrivener — disclose only what you disclose about other tools (typically nothing). Argued for by writers who see AI as acceleration, not authorship. Jennifer Lepp's public discussion of her process was voluntary, not mandatory.

Genre and Context Dependent

Disclosure standards differ by genre. Literary fiction has higher reader expectations of pure human authorship than genre fiction. Journalism requires disclosure; romance may not. Several publishing houses issued explicit AI disclosure policies in 2023–2024.

Publisher and Platform Rules

Since 2023, many platforms have imposed their own rules regardless of author preference. Amazon Kindle Direct Publishing, Clarkesworld, The New Yorker, and others have issued distinct AI-content policies ranging from prohibition to mandatory labeling to no restriction.

What Writers Owe Themselves

The legal and disclosure questions are external. There is also an internal question: what kind of writer do you want to be, and what does your process need to feel like your own?

The experience of multiple documented co-writing practitioners suggests a consistent pattern: writers who are explicit with themselves about why they are using AI at each step — not just accepting any output that seems close enough — tend to report higher satisfaction with the work and less anxiety about attribution. The question "did I make the expressive choice here?" is not just legally relevant; it is creatively clarifying.

The Kashtanova Principle

From the Zarya decision, a practical heuristic: the expressive choices are yours; the generated output, unmodified, is not. Apply this not just legally but creatively. The places where you rewrote, selected, and transformed are the places where you are genuinely the author. Maximizing those moments is both better legal practice and better creative practice.

Key Terms
Expressive ChoiceIn copyright law, the criterion for human authorship — not whether a human touched the work, but whether a human made the decisions about what the work expresses and how.
Zarya DecisionThe 2023 U.S. Copyright Office ruling on Kristina Kashtanova's AI-assisted graphic novel, establishing that AI-generated images lack copyright protection while human-authored text and arrangement retain it.
Tool EquivalenceThe ethical position that AI writing assistance requires no more disclosure than spell-check or other standard writing tools — contested but widely practiced.

Lesson 4 Quiz

Ownership, Attribution, and Ethics of Co-Authorship · 3 questions
What was the specific ruling in the Zarya of the Dawn copyright case regarding Kristina Kashtanova's work?
Correct. The Copyright Office drew the line at expressive choice: Kashtanova's text and curation were human decisions; the raw Midjourney image outputs were not, and lost copyright protection.
Not quite. The decision was split: text and arrangement (human expressive choices) retained copyright; the AI-generated images themselves did not, because they lacked human authorship of the expressive content.
According to the lesson, what legal principle determines the strength of a writer's copyright claim over AI-collaborative work?
Correct. The lesson states that what can be protected is the human's selection, arrangement, and modification of AI output — and that the more substantial the human's expressive choices, the stronger the copyright position.
Not quite. The legal criterion is the extent of the human's expressive choices — selecting, arranging, and modifying AI output. Verbatim acceptance of AI output produces a weaker copyright position than substantial rewriting.
What consistent pattern did multiple documented co-writing practitioners report, according to the "What Writers Owe Themselves" section?
Correct. The lesson identifies this internal discipline — asking "did I make the expressive choice here?" at each step — as producing both better creative satisfaction and less attribution anxiety.
Not quite. The pattern was about internal intentionality: writers who asked "why am I using AI here?" at each step (rather than accepting any close-enough output) reported higher work satisfaction and less anxiety about authorship.

Lab 4 — Authorship Audit

Examine a co-written passage and map your expressive choices

Your Task

Generate a passage collaboratively with the AI — a scene, a character monologue, a story opening. Then conduct an Authorship Audit: identify every sentence or phrase where you made an expressive choice (wrote it yourself, or substantially modified AI output), and every sentence where you accepted AI output without change.

Ask the AI to help you map this — it can help identify which parts came from its suggestions. Then discuss: what would strengthening your authorship of the weaker sections look like?

Try: "Let's co-write a 200-word scene together. I'll give you the premise; you draft it. Then help me do an authorship audit — marking which sentences came entirely from you, and suggest where I should rewrite to make my expressive choices stronger."
Co-Writing Lab
Authorship Audit Mode
Welcome to the Authorship Audit lab. Give me a premise — a genre, a character, a situation — and I'll draft a short scene. Then we'll audit it together: I'll flag the sentences that came most directly from AI pattern-matching, and we'll discuss where you should intervene with your own expressive choices. What shall we write?

Module 6 Test

Collaborative Fiction and Co-Writing · 15 questions · Pass at 80%
1. Robin Sloan's approach to AI co-writing for Moonbound is best classified as which model?
Correct. Sloan wrote primary material and used the AI's responses — mostly rejected — to sharpen his own choices. That is the Sparring Partner model.
The Sparring Partner model describes Sloan's approach: the human writes primary drafts; the AI responds with alternatives that the human mostly rejects but uses for clarification.
2. "Productive Wrongness" refers to:
Correct. Productive Wrongness is the phenomenon where the AI's failure to produce what the writer wanted sharpens the writer's sense of what they actually want.
Productive Wrongness describes the phenomenon where an AI's wrong or unexpected output is more useful than a correct one — because rejecting it clarifies the writer's actual creative intentions.
3. K Allado-McDowell's Pharmako-AI was created through:
Correct. The book alternated passages between human and AI, creating what Allado-McDowell described as "a kind of séance" — a voice identifiable as neither purely human nor purely machine.
Pharmako-AI was produced by alternating passages — the Turn-Taking model — resulting in a voice described as genuinely plural and unattributable to either author alone.
4. Which is the correct description of "Output Extraction"?
Correct. Output Extraction treats the AI as a faster keyboard — describe what you want, accept what it produces, move on — as opposed to iterative collaborative exchange.
Output Extraction describes using AI to generate finished text for direct use — no iterative exchange, no using the AI's failures productively. It's the opposite of genuine co-writing.
5. Jennifer Lepp's documented 2023 AI co-writing practice involved:
Correct. Lepp used a deliberate partition: AI for structural elements (outlines, dialogue scaffolding), solo work for scenes requiring genuine emotional subtext. Her output tripled while reviews remained stable.
Lepp's documented approach was a deliberate partition: AI for plot outlines and dialogue scaffolding, solo writing for emotional subtext. This allowed tripled output without quality decline.
6. "Voice Collapse" in AI-generated fiction is caused primarily by:
Correct. Optimization for clarity and coherence is exactly what makes AI bad at dialogue — real characters interrupt, deflect, and speak in incomplete thoughts. Those qualities look like errors to a clarity-optimized model.
Voice collapse stems from AI optimization for coherence and clarity — the same qualities that make for bad dialogue. Real characters are syntactically distinctive in ways that look like errors to a clarity-optimized model.
7. A counter-example is included in a Voice Specification because:
Correct. The counter-example ("she would never say: 'I feel very hurt'") directly targets the AI's default register — showing what the character avoids is often more constraining than showing what they do.
Counter-examples constrain the AI away from its default coherent register by showing the model what the character would never say — more effective than positive description alone.
8. The Sudowrite AI writing tool's "character voice" feature was built around asking users for:
Correct. Sudowrite's feature asked for sample dialogue, confirming the lesson's principle that actual lines — not personality descriptions — are what anchors an AI to a specific voice.
Sudowrite's feature required sample dialogue before generating in-voice text — operationalizing the lesson's principle that sentence-level examples outperform personality descriptions.
9. A "Voice Card" in long collaborative projects serves primarily to:
Correct. A Voice Card addresses the context-window problem: as a manuscript grows, early voice specifications drift out of active context, so the card is re-injected each session to maintain consistency.
Voice Cards counteract context-window drift — as a long manuscript grows, voice specifications from early sessions become inaccessible to the model, so the card is re-injected at each session start.
10. Brandon Sanderson identified what as the fundamental problem with using AI tools for his Cosmere universe?
Correct. Sanderson's specific concern was that models would violate established rules of his magic systems even when given the books as context — the world-state memory problem.
Sanderson identified world-state memory as the core problem: AI models would contradict established magic system facts within a single session, even when given his books as context.
11. The three layers of an effective AI world bible are:
Correct. The three-layer structure explicitly separates permanent world rules from mutable current state from intentionally unresolved mysteries — each serving a different continuity function.
The lesson describes three layers: Hard Facts (non-negotiable rules), Current World-State (mutable, updated each session), and Open Questions (mysteries you want to keep unresolved).
12. Aedan Peterson reported that introducing a world bible reduced his continuity errors from approximately 3–4 per session to:
Correct. Peterson's reported outcome was a drop from 3–4 continuity errors per session to under 1 — a significant but not total reduction, consistent with the world bible addressing the most common causes.
Peterson reported errors dropping from 3–4 per session to under 1 — a major reduction but not complete elimination, which is a realistic outcome of world bible injection.
13. The Zarya of the Dawn U.S. Copyright Office decision established that:
Correct. The decision split copyright along the expressive-choice line: Kashtanova retained copyright on the text (her human choices) but not on the Midjourney images (accepted without human expressive modification).
The ruling split copyright along the expressive-choice line: human-authored text and arrangement = copyrightable; AI-generated images accepted without human modification = not copyrightable.
14. The "Tool Equivalence" position on AI disclosure argues that:
Correct. The Tool Equivalence position holds that AI assistance is analogous to spell-check or Scrivener — writing infrastructure that doesn't require disclosure any more than those tools do.
Tool Equivalence holds that AI is infrastructure like spell-check — writers don't disclose using Scrivener or autocorrect, and AI assistance needn't be disclosed either.
15. According to the lesson, what practical writing criterion maps onto the legal concept of "expressive choice"?
Correct. The lesson maps "expressive choice" onto the creative question "did I make this choice?" — the places where you rewrote, selected, and transformed are where you are genuinely the author, legally and creatively.
The lesson maps expressive choice onto the creative question of whether the writer rewrote, selected, and transformed the output — or simply accepted it. That distinction is both the legal standard and the creative one.