L1
·
Quiz
·
Lab
L2
·
Quiz
·
Lab
L3
·
Quiz
·
Lab
L4
·
Quiz
·
Lab
Module Test
Lesson 1 · Module 2

What Makes a Character Feel Real?

The psychology of fictional persons — and why LLMs are surprisingly well-suited to build them.
What separates a character that lingers in memory from one that evaporates the moment the page turns?

When George R.R. Martin sued Anthropic in late 2023 alongside a group of authors, the complaint made an argument that went beyond copyright. It claimed that AI systems had consumed and distilled the craft of characterization — the specific techniques by which authors like Martin construct characters that readers grieve when they die. The case brought into sharp relief a question the publishing industry had long avoided: what, precisely, is the thing that makes a character feel irreplaceable?

The legal question remains unresolved. But the underlying craft question is answerable — and understanding it is the precondition for using LLMs in character work without producing hollow imitations.

The Three Dimensions of a Believable Character

Character theory across screenwriting, novel craft, and drama converges on three dimensions that, when present together, produce the sensation of encountering a real person rather than a function.

Dimension one is interiority — the character has a private world of thoughts, fears, rationalizations, and desires that the audience only partially sees. Without interiority, characters feel like puppets moved by plot necessity.

Dimension two is contradiction — the character holds beliefs or traits that are genuinely in tension. A person who is both deeply loyal and deeply ambitious; who craves intimacy but sabotages it. Contradiction is not inconsistency. Inconsistency is failure of craft. Contradiction is the signature of a real psyche.

Dimension three is specificity — the character is not a type but a particular instance. Not "a grieving mother" but a mother who expresses grief through obsessive reorganization of her kitchen, who has not cried once, who sends her dead son's friends food they didn't ask for.

Why This Matters for LLM Work

LLMs trained on vast corpora of fiction have absorbed enormous amounts of character-type information. Without specific prompting, they tend to produce characters that are competent at dimension one (interiority) and weak at dimensions two and three. Knowing this lets you compensate deliberately.

The "Character Bible" Tradition

Professional television writers have long used what's called a character bible — a document that establishes who a character is before a single line of dialogue is written. The practice traces to the early days of serialized television, when writing rooms needed to ensure that multiple writers produced a coherent single character across dozens of episodes.

A character bible typically captures: backstory (formative events), core wound (the thing the character has never recovered from), dominant desire (what they are always moving toward), fear (what they organize their life to avoid), and behavioral tells (how their psychology shows up in small physical actions).

This structure maps almost perfectly onto what LLMs need in order to generate character-consistent dialogue and behavior. The character bible is, essentially, a system prompt for a fictional person.

What LLMs Do Well and Poorly in Character Work
LLMs Do Well

Generating consistent voice once a register is established. Producing backstory variations rapidly. Exploring how a character might respond to novel situations. Identifying gaps in a character's internal logic when asked to stress-test it.

LLMs Do Poorly

Introducing genuine contradiction without prompting. Resisting the pull toward resolution and redemption arcs. Maintaining a character's flaws without softening them. Generating truly specific behavioral details rather than plausible-sounding generics.

The Research Signal: OpenAI's Work on "Persona"

OpenAI's 2023 system card for GPT-4 documented the model's ability to sustain what they termed "persona" — a consistent voice, viewpoint, and behavioral pattern across extended interactions. The card noted that this ability emerged without explicit training on persona-maintenance tasks, suggesting it was a consequence of exposure to large amounts of narrative fiction in which characters maintain consistent behavior across scenes.

This is important for writers to understand: LLMs have a latent capacity for persona maintenance that is far stronger than their capacity for persona creation. They can hold a character you define far better than they can invent one without guidance. The implication is clear — your job is definition; the model's job is inhabitation.

Core Principle of This Module

The writer who understands what makes characters real — interiority, contradiction, specificity — can use that understanding to construct prompts that force an LLM to produce character work at the level of intentional craft rather than genre default.

Key Terms
InteriorityThe private inner life of a character — thoughts, fears, rationalizations — that is shown to the reader only partially and selectively.
Character BibleA professional document establishing a character's backstory, core wound, dominant desire, fear, and behavioral tells before writing begins. Originated in TV writers' rooms.
Persona MaintenanceAn LLM's demonstrated ability to sustain a consistent voice and behavioral pattern once a character is defined — distinct from and stronger than its ability to create characters from scratch.
ContradictionGenuine psychological tension within a character — held beliefs or traits that are in real conflict — as distinct from inconsistency, which is failure of craft.

Lesson 1 Quiz

What Makes a Character Feel Real? · 4 questions
According to character theory, what is "interiority"?
Correct. Interiority is the private psychological world — the parts of a character's inner life the audience sees only obliquely. Its incompleteness is part of what makes a character feel real.
Not quite. Interiority refers specifically to the private inner life — thoughts, fears, rationalizations — that the audience only partially glimpses. It is not a visible biography or spoken monologue.
A character bible, as used in professional TV writers' rooms, serves what primary function when working with LLMs?
Correct. The character bible maps almost perfectly onto what LLMs need — backstory, core wound, dominant desire, fear, behavioral tells — essentially acting as a structured system prompt for a fictional person.
Not quite. The character bible's relevance to LLM work is that it provides structured defining information — backstory, wound, desire, fear, behavioral tells — that lets the model inhabit a character consistently. It is a system prompt for a fictional person.
OpenAI's 2023 GPT-4 system card documented that models are better at "persona maintenance" than "persona creation." What is the practical implication for writers?
Correct. Because LLMs are far stronger at maintaining a character once defined than at creating one from nothing, the writer's primary work is definitional — and then the model can reliably inhabit what the writer has built.
Not quite. The documented difference means that writers should focus on defining characters precisely, then let the LLM inhabit that definition. The model is strong at persona maintenance once you have established the persona.
What distinguishes "contradiction" from "inconsistency" in character craft?
Correct. Contradiction — holding traits in genuine tension — is a mark of a real psyche and an intentional craft decision. Inconsistency is when a character acts differently simply because the writer forgot who they were.
Not quite. Contradiction means holding genuinely conflicting traits or beliefs in intentional tension — the signature of a real psyche. Inconsistency means a character acts differently because the writer lost track. One is craft; the other is failure.

Lab 1 — Build a Character Bible

Practice using the AI to construct a structured character foundation.

Your Task

Use the AI assistant below to build a character bible for a character of your choosing. Begin by describing a character concept (genre, rough role in story, one or two traits you already have in mind). The AI will help you systematically develop the five elements of a character bible: backstory, core wound, dominant desire, fear, and behavioral tells.

Push beyond the first answer. Ask the AI to make the contradiction sharper, the specific behavior more physical and particular, the core wound more precise.

Try starting with: "I'm writing a [genre] story. My character is a [role] who [one trait]. Help me build a character bible — start with backstory and core wound."
Character Bible Builder
Lab 1
Welcome to Lab 1. Tell me about a character you want to develop — even just a rough concept is enough to start. What's the genre, and what's the character's role or a trait you already know about them?
Lesson 2 · Module 2

Prompting for Contradiction and Specificity

The precise language structures that force LLMs past genre defaults into genuinely particular characters.
Why do AI-generated characters so often feel like placeholders — and what is the exact prompting intervention that fixes it?

When Sudowrite — an AI writing tool built specifically for fiction — launched its "Characters" feature in 2022, the team published a transparency blog post documenting what they'd learned in beta. The most consistent user complaint was that AI-generated characters felt "like a list of adjectives wearing a trenchcoat." Users could generate backstory, desires, and fears easily. What they couldn't get was behavioral specificity — the particular gesture, the idiosyncratic habit, the detail that made a character recognizable from a single action rather than a description.

Sudowrite's engineering team documented that the fix wasn't a model change — it was a prompting structure change. When prompts were restructured to ask for behavior in a specific situation rather than traits in the abstract, the specificity problem largely resolved.

The "Trait Trap" in Character Prompting

The default mode of asking an LLM about a character is to ask for traits: "What are this character's defining qualities?" This produces competent but flat results because traits are abstractions. "She is stubborn" describes a category. "When she disagrees with someone, she goes quiet and starts tidying objects within reach" is a specific — and it contains far more story potential.

The trait trap is not unique to AI. Writing workshop participants fall into it constantly. The difference is that LLMs are particularly susceptible because trait-level description is far more common in training data than specific behavioral data tied to specific characters in specific situations.

Four Prompting Structures That Break the Trait Trap
  • 1Situational behavior prompts: "Show me how [character] reacts when [specific situation] happens — not in general terms, but in one physical action and one internal thought." Forces behavior out of abstraction and into scene.
  • 2Contradiction activation prompts: "Give me a moment where [character's stated desire] and [character's stated fear] are in direct conflict — and describe which one wins and why." This forces the model to resolve the tension rather than hold both traits simultaneously without friction.
  • 3Object anchoring prompts: "What object does [character] own that they would never explain to anyone? What does it mean to them?" Physical objects anchor abstract traits in sensory reality and tend to produce genuinely specific details.
  • 4Negative space prompts: "What does [character] never talk about? What question, if asked, would make them change the subject immediately?" Defining what a character withholds is often more generative than defining what they reveal.
The Iterative Refinement Principle

No first-pass AI character response will be fully specific. The prompting structures above are designed for iteration — use the first response as raw material and immediately push it: "Make that behavioral tell more physical and less psychological." "Give me the same contradiction but in a context where the character has no audience." The gap between a generic AI character and a specific one is almost always one or two targeted follow-up prompts.

Specificity Levels: A Practical Taxonomy
Level Example Notes
Trait (weakest) "She is guarded." Produces nothing actable or visual. Every reader imagines something different.
Behavior (better) "She deflects personal questions with humor." Actable, but still generic. Many characters share this behavior.
Specific Behavior (strong) "When asked about her past, she picks up the nearest object and finds something wrong with it." Visual, specific, tells us something about her anxiety manifesting as control.
Situated Behavior (strongest) "At her mother's funeral, when the priest asks if she wants to speak, she adjusts her watch three times before saying no." Fully inhabited. The watch detail implies years of a specific relationship with time and control.
The Contradiction Prompt in Practice

In 2023, novelist and writing instructor K.M. Weiland published a detailed account of using ChatGPT-4 in her character development process for her novel-in-progress. Her key finding, documented on her widely-read "Helping Writers Become Authors" blog, was that standard character prompts produced what she called "therapist-speak characters" — psychologically labeled but not behaviorally alive.

The breakthrough came when she stopped asking what a character was and started asking what the character does when their stated values fail them. The question "What does your character do when they cannot be the person they think they are?" produced, she documented, qualitatively different material — characters caught in their own contradictions rather than described from the outside.

This is the prompting principle made explicit: force the model into the gap between self-image and behavior. That gap is where real character lives.

Writer's Rule

A character becomes real at the moment you know what they would do wrong — not just what they want, but what they cannot stop themselves from doing even when it costs them. That moment is the prompting target.

Lesson 2 Quiz

Prompting for Contradiction and Specificity · 4 questions
What did Sudowrite's beta testing in 2022 find was the key fix for AI-generated characters that felt like "a list of adjectives wearing a trenchcoat"?
Correct. Sudowrite's documented fix was a prompting structure change — asking for behavior in a specific situation instead of traits in the abstract — not a model upgrade.
Not quite. The fix was a prompting structure change: asking for behavior in specific situations rather than abstract traits. The model itself didn't need to change.
Which of the following is an example of a "situated behavior" — the strongest level in the specificity taxonomy?
Correct. Situated behavior places a specific physical action in a specific social context. The pen-clicking detail and the averted gaze imply a full psychology without naming it.
Not quite. Situated behavior is the strongest level — it places a specific physical action in a specific context. Only the last option does that. The others are trait-level or behavior-level, which are weaker.
K.M. Weiland documented that her AI character work improved significantly when she shifted her prompting from asking "what a character was" to asking what?
Correct. Weiland's documented breakthrough was prompting for the gap between self-image and behavior — what a character does when they can't be who they believe themselves to be.
Not quite. Weiland found that asking what a character does when their stated values fail them — the gap between self-image and behavior — produced qualitatively better character material.
What is the purpose of a "negative space prompt" in character development?
Correct. Negative space prompting asks what a character never discusses, avoids, or would change the subject about — defining what they withhold, which often reveals more than what they display.
Not quite. A negative space prompt asks about what a character withholds — what they never talk about, what question would make them shut down. Absence and avoidance are often more revealing than presence.

Lab 2 — Specificity Drills

Practice the four prompting structures that break the trait trap.

Your Task

Take a character you already have in mind — either from Lab 1 or one you bring to this session. Work through at least two of the four prompting structures from Lesson 2: situational behavior, contradiction activation, object anchoring, or negative space.

After each response, push for more specificity. If the AI gives you a behavior, ask it to make that behavior more physical. If it gives you a contradiction, ask which side wins in a specific scene.

Try: "Using [character], give me a contradiction activation prompt: put their [desire] and [fear] in direct conflict in a single moment, and tell me which one wins and why."
Specificity Drills
Lab 2
Ready to work on specificity. Tell me about a character — existing from Lab 1, or a new one — and we'll run them through the prompting structures. Which character are we working with, and do you already know their dominant desire and core fear?
Lesson 3 · Module 2

Voice, Dialogue, and the Consistency Problem

How to prompt for character-consistent dialogue — and what to do when the model drifts.
If a character's voice is their most immediate signature, why do AI-generated dialogue drafts so frequently sound like the same person wearing different masks?

In 2023, Jasper AI published internal research on what they called "voice drift" — the tendency of LLMs to gradually homogenize the speech patterns of characters over the course of a long generation session. Their finding was consistent with what fiction writers had been reporting anecdotally: characters who began a scene with distinct voices would, by the third or fourth exchange of dialogue, begin to sound alike.

Jasper's engineering team found that voice drift was not a bug but a feature of how LLMs predict token sequences — the model gravitates toward the most statistically probable dialogue given the conversational context, and the most probable dialogue is average dialogue. The fix they documented was periodic "voice anchoring" — reintroducing specific voice parameters mid-session.

What Character Voice Actually Is

Voice in fiction is not the same as dialect or accent, though those can be components. Voice is the total pattern of a character's language choices: what they talk about without being asked, what they avoid, how long their sentences are, whether they finish sentences or trail off, what they notice and name versus what they ignore, their relationship to abstraction versus the concrete.

A character who grew up on a working farm in the 1970s doesn't just speak with regional inflection — they are more likely to use object-level language rather than abstract language, more likely to describe emotions through work metaphors, more likely to be suspicious of anything they can't touch. These are voice parameters, and they can be made explicit in prompts.

Building a Voice Profile for Prompting

A voice profile for prompting contains six parameters that together define a character's linguistic signature:

Parameter What to Specify
Register Formal / informal / vernacular. Does the character code-switch? When?
Sentence rhythm Short and declarative; long and qualified; fragmentary; does not finish thoughts.
Abstraction ratio Speaks in concepts vs. objects. Uses feelings-language vs. action-language.
Deflection pattern How the character avoids direct answers — humor, questions, silence, subject change.
Dominant metaphor field The domain from which the character naturally draws comparisons: sport, money, weather, food, machines.
What they never say Words or phrases the character would not use — often as important as what they do say.
The Voice Anchoring Technique

Voice anchoring is the practice of re-introducing voice parameters at strategic points in a dialogue generation session. The technique was developed independently by multiple writers working with early GPT-4 in 2023, and convergent accounts appear in the online communities r/ChatGPTPromptEngineering and the Absolute Write Water Cooler forum.

The simplest form of voice anchoring is a mid-session meta-prompt: "Before continuing, read the last six lines of dialogue. Is [Character A] still speaking in short, declarative sentences that avoid emotional abstraction? If not, revise."

A more robust form is to establish a voice-check ritual before each dialogue block: provide the voice profile, then add "Now write the next exchange of dialogue. If you notice [character] beginning to use abstraction or long-qualified sentences, stop and correct before continuing."

The Homogenization Problem

Researchers at Stanford's Human-Centered AI Institute documented in a 2023 paper ("Do LLMs Distinguish Character Voices in Fiction?") that GPT-4-class models showed measurable but not always reliable differentiation between character voices when prompted generically. Voice differentiation improved significantly — measured by computational stylometrics — when voice profiles were explicitly included in character definitions and re-introduced at intervals.

Dialogue Consistency Across a Long Project

For writers working on full-length novels or scripts, the consistency problem compounds over time. A character established in chapter one may appear different by chapter twelve if the writer has been using LLMs without systematic voice anchoring.

The practical solution used by professional writers who have documented their AI-assisted workflows — including Brandon Sanderson's Dragonsteel Entertainment team, which published notes on AI use in their production process — is to maintain a live voice document: a single file that contains the voice profile, three to five example dialogue lines approved by the author, and a list of words/phrases to avoid. This document is prepended to any AI prompt that involves character dialogue.

This is not different in principle from how writers' rooms use character bibles — it is simply the dialogue-specific version, updated as the character evolves through drafts.

Practical Protocol

Before any dialogue generation session: state the character's register, sentence rhythm, abstraction ratio, and one phrase they would never use. After every 8–10 exchanges of generated dialogue: run a voice-check prompt asking the model to identify any drift and correct it before continuing.

Key Terms
Voice DriftThe documented tendency of LLMs to gradually homogenize character speech patterns over a long generation session, gravitating toward statistically average dialogue.
Voice ProfileA structured document specifying a character's register, sentence rhythm, abstraction ratio, deflection pattern, dominant metaphor field, and excluded vocabulary.
Voice AnchoringThe practice of periodically re-introducing voice parameters mid-session to prevent voice drift in long dialogue generation work.
Live Voice DocumentA maintained file containing voice profile, approved example dialogue, and excluded vocabulary — prepended to any AI dialogue prompt to ensure consistency across a long project.

Lesson 3 Quiz

Voice, Dialogue, and the Consistency Problem · 4 questions
Jasper AI's 2023 internal research on "voice drift" found that the cause was what?
Correct. Voice drift is not a bug — it's an artifact of how LLMs predict tokens by gravitating toward the most probable next word, which tends toward average dialogue over time.
Not quite. Voice drift is caused by the model's token-prediction mechanism gravitating toward statistically probable — i.e., average — dialogue, gradually homogenizing distinct character voices. It's a feature of how LLMs work, not a bug.
According to the lesson, what does "abstraction ratio" mean in a character voice profile?
Correct. Abstraction ratio specifies whether a character tends toward concept-level and feelings-language or object-level and action-language — a fundamental dimension of voice.
Not quite. Abstraction ratio in a voice profile describes whether the character tends to speak in concepts and feelings, or in objects and actions. It's a fundamental dimension of linguistic voice.
What is the simplest form of "voice anchoring" as described in the lesson?
Correct. The simplest voice anchoring technique is a mid-session meta-prompt — asking the model to review recent dialogue, identify drift from the established voice parameters, and correct before continuing.
Not quite. The simplest voice anchoring technique is a mid-session meta-prompt: ask the model to read recent dialogue, check whether the character's voice parameters have been maintained, and revise if they have not.
The Stanford HAI 2023 study on LLM voice differentiation found what?
Correct. The Stanford study found that generic prompting produced unreliable voice differentiation, but including explicit voice profiles and re-introducing them at intervals produced measurable improvement by computational stylometric measures.
Not quite. The Stanford HAI study found that LLMs can differentiate voices, but only reliably when explicit voice profiles are included and periodically re-introduced. Generic prompting produced measurably inconsistent results.

Lab 3 — Voice Profile and Dialogue Consistency

Build a voice profile and test it against drift over multiple exchanges.

Your Task

First, work with the AI to construct a voice profile for your character using the six parameters from Lesson 3: register, sentence rhythm, abstraction ratio, deflection pattern, dominant metaphor field, and what they never say.

Then prompt for a dialogue exchange (your character talking with another character of your invention). After 4–5 exchanges, run a voice-check: ask the AI whether any drift has occurred and to revise accordingly.

Try: "Build a voice profile for [character name] using these parameters: register, sentence rhythm, abstraction ratio, deflection pattern, dominant metaphor field, and three phrases they would never use."
Voice Profile Builder
Lab 3
Let's build a voice profile. Tell me about the character — their background, education level, what world they move through — and we'll work through all six voice parameters systematically. Start with the character's name and a brief description.
Lesson 4 · Module 2

Character Arc: Using LLMs to Plot Transformation

How to use AI to model change — and why character transformation is the hardest thing to get right in AI-assisted fiction.
If a character who does not change is not a character but a weather pattern, how do you prompt an LLM to help design change that feels earned rather than imposed?

When the Writers Guild of America went on strike in May 2023, one of the central documents in the dispute was a position paper on AI's role in the writing process. The paper explicitly addressed character arc — arguing that the studios' proposed use of AI for "character trajectory modeling" would hollow out the most human element of storytelling: the authored decision about what a person is capable of becoming and why.

The WGA's argument was not that AI couldn't model transformation — it was that AI-modeled transformation would default to the most statistically common arc shapes: redemption, fall-from-grace, coming-of-age. The document argued that the distinguishing feature of literary character arcs is their specificity to the particular character's particular wound — which requires an authorial decision that no statistical average can produce.

What a Character Arc Actually Is

A character arc is not a mood trajectory (character starts sad, ends happy) or a status trajectory (character starts poor, ends rich). An arc, properly understood, is a change in what the character believes is true about themselves or the world — and that change must be caused by events rather than declared by the author.

The classic formulation, derived from Robert McKee's Story (1997) and widely taught in MFA programs, is that a character arc moves from a misbelief — a false assumption the character holds about reality — through a series of pressure events that test that misbelief, to either its abandonment (positive arc) or its fatal reinforcement (negative arc) or a conscious decision to keep believing it despite evidence (flat arc).

This structure is important for LLM work because it gives the model something concrete to track. "Show me how this character's misbelief is challenged by this event" is a far more generative prompt than "show me character growth."

The Misbelief-Pressure-Revision Framework
  • 1Define the misbelief explicitly. "This character believes that [specific false assumption — e.g., that love requires self-erasure, that safety requires control, that kindness is weakness]."
  • 2Generate pressure events. "Give me three plot events that would force this character to directly confront their misbelief — not metaphorically, but in a specific situation where the misbelief costs them something real."
  • 3Prompt the resistance moment. "Show me a scene where [character] has been confronted with evidence against their misbelief and actively chooses to reinterpret the evidence rather than revise their belief." This is the pivotal scene that makes change feel earned when it finally comes.
  • 4Prompt the revision or reinforcement. "Now write the moment the misbelief cracks — or the moment the character doubles down. What is the specific image or action that marks the change without stating it?"
The WGA's Warning Applied as Craft Advice

The WGA's concern about default arc shapes is real and useful. When you prompt an LLM for a character arc without a defined misbelief, it will almost always generate one of three arcs: learning to trust others, learning to value themselves, or learning that ambition has limits. These are valid arcs — but they are the three most common because they appear most often in training data. Naming the specific misbelief forces the model off the default path.

The "Resistance Scene" as a Prompting Target

The most under-developed element in AI-assisted character arc work is what craft teachers call the resistance scene — the moment, usually in the second act, when a character is given clear evidence that their misbelief is wrong and actively refuses to revise it. This scene is crucial because it makes the eventual change feel like a choice rather than an accident.

LLMs without explicit prompting tend to skip resistance scenes, moving characters from misbelief to revision too quickly and too smoothly. The fix is explicit prompting: "Write a scene where [character] actively reinterprets contradicting evidence in order to preserve their misbelief. The character should feel justified — make us understand their internal logic even if we disagree with it."

Author and writing teacher Lisa Cron, in her 2023 course material published through MasterClass, described exactly this pattern — what she calls "story-brain" tendency in AI output: "The AI wants to resolve the tension. Your job is to refuse to let it resolve the tension until the story has earned the resolution."

Tracking Arc Across Multiple Sessions

Unlike voice profiles, which can be held in a single document, character arc requires tracking change over time. The practical solution used by writers working on long projects is a simple arc log: a running document that records the current state of the character's misbelief, the most recent pressure event, and the degree to which the misbelief has been shaken.

This arc log becomes a context prepend for any arc-related prompting session, ensuring the model knows where the character is in their transformation rather than always starting fresh.

The Core Rule of Arc Prompting

Never prompt for "character growth" as an abstraction. Always prompt for the specific misbelief at stake, the specific event applying pressure, and the specific moment at which resistance gives way — or doesn't. Arc lives in the specifics, not in the movement between emotional states.

Key Terms
Character ArcA change in what a character believes is true about themselves or the world — caused by events, not declared by the author. Distinct from status or mood trajectories.
MisbeliefThe false assumption a character holds at the story's opening — the specific belief that the arc will test, challenge, and ultimately revise or reinforce.
Resistance SceneThe scene in which a character receives clear evidence against their misbelief and actively chooses to reinterpret or dismiss it — the crucial step that makes eventual change feel earned.
Arc LogA running document tracking the current state of a character's misbelief, recent pressure events, and degree of change — used as a context prepend in long-project AI arc work.

Lesson 4 Quiz

Character Arc: Using LLMs to Plot Transformation · 4 questions
The WGA's 2023 position paper argued that AI-modeled character arcs would default to what?
Correct. The WGA argued that AI trajectory modeling defaults to statistically common arcs — redemption, fall, coming-of-age — and that the specificity of literary arc to a particular character's wound requires an authorial decision no statistical average can produce.
Not quite. The WGA's documented concern was that AI defaults to the most statistically common arc shapes because those appear most in training data — not that it produces plagiarized or nihilistic arcs.
In the Misbelief-Pressure-Revision framework, what is the function of explicitly prompting a "resistance scene"?
Correct. The resistance scene — where a character receives evidence against their misbelief and actively refuses it — is what makes the eventual change feel earned. Without it, transformation feels arbitrary rather than dramatic.
Not quite. The resistance scene's function is to show the character actively refusing to revise their misbelief despite evidence. This refusal is what makes the eventual change feel earned and deliberate rather than accidental.
What is a "misbelief" in character arc theory, and why is naming it specifically important for LLM prompting?
Correct. A misbelief is the false assumption about reality that the arc will test. Naming it specifically forces the LLM away from generic arc shapes and toward a transformation grounded in this particular character's psychology.
Not quite. A misbelief is the false assumption about reality a character holds at the opening — not a factual error or behavioral contradiction. Naming it specifically prevents the model from defaulting to generic arc shapes like redemption or coming-of-age.
What is an "arc log" and how is it used in long-project AI-assisted writing?
Correct. An arc log tracks where the character is in their transformation — current misbelief state, recent pressure events, degree of change — and is prepended to prompts to prevent the model from treating each session as if the character has not already changed.
Not quite. An arc log is a running document that tracks the character's current misbelief state, recent pressure events, and degree of change. It's prepended to arc-related AI prompts so the model knows where the character currently is in their transformation.

Lab 4 — Designing a Character Arc

Use the Misbelief-Pressure-Revision framework to build a transformation that feels earned.

Your Task

Work through the Misbelief-Pressure-Revision framework with your character (or a new one). Start by defining the misbelief explicitly. Then ask the AI for three pressure events, then for a resistance scene, then for the revision or reinforcement moment.

Push past the first arc suggestion. If the AI defaults to redemption or coming-of-age, name the misbelief more specifically and ask it to try again. The goal is an arc that could only belong to this character.

Start with: "My character's misbelief is: [specific false assumption]. Using the Misbelief-Pressure-Revision framework, give me three plot events that would force a direct confrontation with this specific misbelief — not metaphorically, but at real cost."
Character Arc Designer
Lab 4
Welcome to the arc lab. To start, tell me your character's name and — most importantly — what false belief they hold about themselves or the world at the story's opening. The more specific the misbelief, the more specific and earned the arc will be. What's the misbelief?

Module 2 Test

Character Development with LLMs · 15 questions · Pass at 80%
1. Which three dimensions of character, when present together, produce the sensation of encountering a real person rather than a narrative function?
Correct. Interiority (private inner world), contradiction (genuine psychological tension), and specificity (particular instance rather than type) are the three core dimensions.
Not quite. The three dimensions identified in the lesson are interiority, contradiction, and specificity. The others are important elements but not the core three-part framework.
2. What does a character bible establish before writing begins?
Correct. A character bible captures the five foundational elements: backstory, core wound, dominant desire, fear, and behavioral tells.
Not quite. A character bible establishes backstory, core wound, dominant desire, fear, and behavioral tells — the structural foundation, not dialogue or plot.
3. OpenAI's GPT-4 system card documented that models are stronger at "persona maintenance" than "persona creation." What does this mean for the writer's role?
Correct. Because the model is strong at maintenance and weaker at creation, the writer's primary job is definition — and the model can then reliably inhabit that definition.
Not quite. The documented difference means definition is the writer's job; the model's strength is inhabiting and maintaining that definition consistently.
4. The George R.R. Martin vs. Anthropic lawsuit (2023) raised, beyond copyright, the argument that AI had absorbed what specific craft element?
Correct. The complaint argued that AI had absorbed and distilled the craft of characterization — the specific techniques that make characters feel irreplaceable to readers.
Not quite. The complaint argued specifically that AI had absorbed the craft of characterization — how to construct characters readers grieve — not just plot or style.
5. In the specificity taxonomy, what distinguishes "situated behavior" from "specific behavior"?
Correct. Situated behavior is the strongest level — it puts a specific action in a specific scene context, allowing the detail to carry psychological meaning without stating it.
Not quite. Situated behavior is one level above specific behavior — it places the action in a particular scene context, letting the specific detail imply the psychology without stating it.
6. The "object anchoring" prompting technique works because:
Correct. Asking about a character's significant object anchors abstract traits in physical, sensory reality — which tends to produce genuinely specific rather than generic details.
Not quite. Object anchoring works because physical objects ground abstract traits in sensory reality, pulling the AI response away from generic trait language toward specific detail.
7. K.M. Weiland documented that the shift from "what a character is" to "what a character does when their values fail them" produced what improvement?
Correct. Weiland's documented finding was that this prompting shift produced characters in their contradictions — behaviorally alive — rather than "therapist-speak characters" labeled from the outside.
Not quite. Weiland found that the shift produced characters caught in their own contradictions rather than described from the outside — behaviorally alive instead of psychologically labeled.
8. What are the six parameters of a voice profile for prompting?
Correct. The six voice profile parameters from the lesson are: register, sentence rhythm, abstraction ratio, deflection pattern, dominant metaphor field, and what the character never says.
Not quite. The six voice profile parameters are: register, sentence rhythm, abstraction ratio, deflection pattern, dominant metaphor field, and what the character never says.
9. Voice drift occurs in LLM dialogue generation because:
Correct. Voice drift is an artifact of token prediction — the model gravitates toward the most probable next word, which tends toward average, and gradually homogenizes distinct voices.
Not quite. Voice drift is caused by the token-prediction mechanism gravitating toward statistically average dialogue over time — not context window limits or training data gaps.
10. What does a "live voice document" contain, and how is it used?
Correct. A live voice document contains voice profile, approved example dialogue lines, and excluded vocabulary — it is prepended to AI dialogue prompts to prevent voice drift across a long project.
Not quite. A live voice document contains: voice profile, three to five approved example dialogue lines, and excluded vocabulary — prepended to AI prompts for consistency across a long project.
11. How does the Misbelief-Pressure-Revision framework define a character arc differently from a "mood trajectory"?
Correct. A mood trajectory tracks emotional states (sad to happy); the arc framework tracks change in fundamental belief — and that change must be caused by events rather than declared by the author.
Not quite. A character arc is not a mood trajectory. It tracks change in what a character believes is true — caused by events, not declared. This distinction is central to the framework.
12. What was the WGA's documented concern about AI "character trajectory modeling" in their 2023 position paper?
Correct. The WGA argued that AI trajectory modeling defaults to the most common arc shapes — redemption, fall, coming-of-age — rather than arcs specific to each character's particular wound, which requires human authorial decision.
Not quite. The WGA's specific documented concern was that AI defaults to statistically common arc shapes, and that arc specificity to a character's particular wound requires an authorial decision no average can produce.
13. Why does Lisa Cron's "story-brain" concept apply to AI-assisted arc work?
Correct. Cron's observation is that AI wants to resolve tension — moving characters through change too quickly, skipping the resistance scenes that make change feel earned. The writer's job is to refuse that resolution until the story has earned it.
Not quite. Cron's "story-brain" concept applies because AI tends to resolve tension prematurely — skipping resistance scenes and moving characters to change before the story has done the work to earn it.
14. A "negative space prompt" in character development asks for what?
Correct. Negative space prompting focuses on what a character withholds — avoidances, silences, subject changes — because defining what a character won't say or do often reveals more than what they will.
Not quite. A negative space prompt asks what the character avoids, withholds, or refuses to discuss. Defining absence is often more generative than defining presence in character work.
15. Which of the following best describes the practical workflow for using the Misbelief-Pressure-Revision framework with an LLM?
Correct. The framework is sequential and iterative: define misbelief → pressure events → resistance scene → revision or reinforcement. Each step is prompted and refined before moving to the next.
Not quite. The framework is sequential: define the misbelief explicitly, generate pressure events, prompt the resistance scene, then prompt the revision or reinforcement. Each step is iterated before moving forward.