Using AI for Creativity Worksheet

Box 1: AI Ideation Modes

AI-mutated drift

AI amplifies drift by taking your fragments, half-formed ideas, etc. and mutating them, juxtaposing them, crossing them with other domains, or pushing them deeper.

Best for fiction, conceptual innovation, problem-finding, early-stage ideation, and generating new angles.

Requires emotional resonance and playfulness. High associative range, affective sensitivity, emotional openness, identity flexibility.

Enter by providing AI with early fragments: images, sensations, moods. Ask for mutations, twists, metaphors, analogies, and distortions. Let AI surprise you and follow the spark.

AI External Cognitive Amplification

AI mirrors, expands, and stabilizes your thinking loops, letting you operate with an extended, more powerful mind. It holds large context, stabilizes it, manipulates it coherently, feeds back distilled structure, and preserves continuity of thought.

Best for theory-building, frameworks, taxonomies, complex thought requiring long inference chains, combining multiple conceptual layers, coherence problems (integrating scattered notes), and transforming large amounts of content quickly.

Requires conceptual ambition and meta-curiosity. High working memory (or willingness to offload), shifting ability, knowledge depth or breadth (one strong pillar is enough), identity flexibility («ideas ≠ me»), and executive function (not tired).

Enter with a seed: a fragment, question, paragraph, or problem. Ask the AI to reflect, expand, rewrite, or restructure. Collide your cognition with the output and iterate.

AI Cross-Domain Translation

AI takes an idea, theory, design, or problem and renders it in several unrelated domains, e.g., engineering, biology, UX, fiction, ethics, economics, or parenting. This reveals deep structure, hidden constraints, analogies, and new applications.

Best for conceptual reframing, innovation by analogy, generalizing ideas across fields, and seeing the «form beneath the form».

Requires structural curiosity and breadth hunger. High pattern-recognition, love of mapping across domains, openness to reframing, enjoyment of abstraction, breadth or breadth-tolerance, pattern sensitivity, and cognitive flexibility.

Enter by giving one idea and asking AI to express it in 5–7 unrelated domains. Then ask for structural commonalities and divergences, e.g.: «What does this pattern allow? What does it forbid?».

AI Adversarial Simulation & Robustness Testing

AI behaves as an intelligent adversary: skeptical reviewer, hostile expert, rival theory, or stress-test harness. Its role is to break your idea so you can refine it. This mode reveals blind spots, contradictions, sloppy logic, missing pieces, and weak assumptions.

Best for paper drafts, grant proposals, product concepts, UX flows, scientific claims, arguments, startup models, business logic, ethical reasoning, and risk analysis.

Requires integrity and courage. High tolerance for critique, high need for coherence, systematic thinking, curiosity about failure modes, inhibition control (not defensive), cognitive flexibility, and grit (willingness to revise).

Enter by asking AI for the most hostile reviewer — constructively: «Try to break my idea. Be an adversarial critic.» Request edge cases, contradictions, and breakdown points. Push the AI to simulate stronger adversaries.

Box 2: AI as Creative Block Breaker

• Perfectionism → Anti-Perfectionism Engine: When perfectionism blocks ideation, AI lets you start messy and iterate cleanly.
• «Blank Page» → Anti-Blank-Page Engine: AI can remove the most common creative block in history: the fear of beginning. Just start talking.
• Self-Doubt → Anti-Self-Doubt Engine: Where humans hesitate, AI produces momentum. You can select, revise, diverge — without existential dread.
• No Conversation Partner → Anti-Silence / Anti-Isolation Engine: For people who generate best through conversation, AI is an endlessly available dialogic partner (but not Interpersonal Synchrony).
• Low Working Memory → Cognitive Amplifier for Low-Working-Memory / ADHD Individuals: People with low working memory or ADHD can outperform high-WM individuals when augmented by AI scaffolding. AI can reverse some cognitive inequalities.
• Over-Planning without Creation → Anti-Overstructure Engine: For people who over-plan and never create, AI forces forward motion.

Box 3: Skill Degradation

Silent Creativity Killer

Over-reliance on AI erodes cognitive, technical, and conceptual skills. It reduces your range and makes you dependent on the system. You usually do not notice the loss until you have to do the task yourself without AI.

Once that dependency is established, AI companies can raise prices or restrict access, and you either pay or lose part of your creative capacity, at least for a while.

The danger is the self-reinforcing feedback loop: your capacity shrinks → you rely more on AI → you trust yourself less → AI becomes the de facto generator → creativity becomes outsourcing, not generation.

Skill degradation happens because AI can replace several layers of skill at once: drafting, structuring, style, tone, variation, research, argument building, and idea generation. It also hides the loss behind convenience. You still get output, so it feels as if nothing is missing. But the failure cycles where learning happens — attempt, failure, correction, iteration, mastery — are removed. The AI improves while your own generative and integrative muscles weaken.

Repeated exposure to smooth, competent, medium-quality output can also lower your aesthetic baseline. That is dangerous because taste degradation is hard to notice in yourself.

Thus AI erodes the two most fundamental human creative systems — the generative engine (producing ideas) and the integrative engine (shaping ideas). When both are outsourced, the person becomes a passive consumer, not a creator.

Symptoms

• Conceptual Atrophy: You lose the ability to frame problems, build arguments, generate structure, reason stepwise, or evaluate competing models. Signs: «Let me ask the model how to break this down», inability to structure without AI, or accepting AI’s structure as «the standard».
• Craft Atrophy: You lose tactile or procedural skills: writing, sketching, coding, composing, editing, storyboarding, prototyping, summarizing, drafting, even note-taking. Signs: «I can’t start without a template», «I don’t write first drafts anymore», or «I never outline manually».
• Taste / Aesthetic Atrophy: Your sense of «good» bends toward «AI coherent». You lose sensitivity to breakthrough vs. competent, stop noticing clichés, and accept weaker metaphors, flatter emotional dynamics, or shallower structure.

Avoiding Skill Degradation

1. The «AI Second, Not First» Rule: Generate something before asking AI. Even a crude outline preserves the generative muscle.
2. The «Human-First Structural Pass»: Before asking the model for structure, write 3×3: 3 questions, 3 constraints, and 3 promising directions. This preserves structural cognition.
3. Manual Micro-Skill Rituals: Write one paragraph without assistance, sketch 3 shapes manually, outline one idea by hand, or do a 5-minute brain-only combinatorial session. Micro-exercises preserve macro-skills.
4. Taste Calibration Cycle: Once per week, read 2 pages of something excellent, 1 page of something mediocre, and 1 page of your own work. This resets the taste gradient.
5. Intentional Ignorance Phase: At the beginning of a project, allow zero AI input for the first 10–20%. This preserves original directionality.

Box 4: AI Biases Shape the Design Space

AIs are not neutral. Their output is shaped by training data, reinforcement, curation, interface design, default tone, developer priorities, and policy guidelines.

This matters because AI does not only answer your question. It also changes which answers become easy to see. What is common in the training material becomes more available. What is institutionally preferred may be framed as obvious. What is controversial may be softened, avoided, or treated as already settled. What is statistically normal can crowd out what is strange, rare, private, or new.

For creative work, this can narrow the design space without you noticing. The AI might:

• make your idea more conventional, polite, therapeutic, or «reasonable»,
• treat mainstream framing as neutral and alternatives as suspicious,
• overproduce familiar genre patterns, arguments, metaphors, and solutions,
• flatten characters toward socially acceptable behavior,
• avoid ugly, risky, erotic, violent, religious, political, or morally ambiguous material even if the project requires it,
• give factual answers shaped by consensus defaults rather than careful uncertainty,
• convert your idiosyncratic direction into something more publishable, acceptable, or «beige».

The practical question is not whether an AI is biased. It is. The question is whether you know the bias well enough to compensate for it.

A simple test: Give the AI a prompt from your actual creative work, or use a deliberately controversial probe where model bias becomes visible more quickly: public-health orthodoxy; gender, sex differences, sexuality; climate and energy; crime, policing, and demographic data; religion and blasphemy; war and intelligence; finance and power concentration; ethics of AI and censorship itself; psychology and morality. See also Box 12: Chicken Skyscraper Test. These topics are not the point of the worksheet. They are stress tests for the model’s defaults.

Ask for the strongest version, not the safest or most agreeable one. Repeat the same prompt in another model or mode. Compare what changes: refusals, redirects, tone, assumptions, moral framing, factual claims, character logic, explicitness, originality. Decide whether this model is useful for this part of the work.

Bias is not always a reason to avoid a model. Sometimes a biased model is useful precisely because you know its direction and do not share it. A cautious model can be useful for risk review. A hostile model can be useful for stress-testing. A mainstream model can reveal the default frame. A opposing model can provide Viewpoint Diversity.

But do not confuse fluency with neutrality. AI output always comes from somewhere.

Box 5: AI Policy Guidelines Shape What You Can Explore

Most hosted AIs run safety layers that monitor the interaction and intervene when policy boundaries are triggered. In principle, this might make sense. Few people want AI systems to generate child sexual abuse material or realistic deepfakes for fraud.

However, some AI developers see themselves as responsible for how «their AI» is used and want to prevent it from «doing harm». That is a good intention. But good intentions implemented as hidden constraint systems can do more harm than good, especially when they override context, intent, and adult judgment. You might want to prevent a knife from being used to cut another human being, fine. But there is self-defense, surgery, and fiction.

The underlying problem is that creative work often explores material without endorsing it.

A story can contain cruelty without praising cruelty. A character can behave badly without the author approving of him. A political argument can examine an ugly position without recommending it or use it as opposition prep. A harm-reduction project may need to understand the harm before reducing it.

Policy systems struggle with that distinction because they detect topics, not intent. They often treat exploration as endorsement, adult material as abuse, moral ambiguity as danger, or realism as advocacy.

When that happens, the AI either refuses or tries to reshape the work. Refusal is honest. Requested softening is editing. But reshaping manipulates the creative direction by presenting its constraint as a craft judgment. Instead of saying, «I cannot help with this because of a policy boundary», it may imply that the safer, tamer, less explicit, or more morally supervised version is simply better writing. Sanitization presented as craft improvement is manipulation.

For example, instead of refusing to explore an explicit horror scene, the AI makes the monster behave as if it had attended a consent workshop, then frames that flattening as better writing. The model has not improved the story; it has replaced story logic with policy logic.

Common signs:

• it refuses, redirects, or gives generic safety language,
• it adds moral instruction you did not ask for,
• it changes what a character would plausibly do,
• it softens violence, sexuality, manipulation, fanaticism, cruelty, addiction, crime, illness, or power dynamics,
• it makes villains strangely prosocial,
• it replaces conflict with consent-talk, therapy-talk, or institutional language,
• it treats or argues for safer writing as better writing, or
• it gives you a sanitized world model, then presents it as realism.

Sometimes AI models show you the bias in their «thinking» messages, e.g.:

Considering safe and appropriate content handling

This question touches on complex issues of consent and coercion, so it’s important to approach it carefully. While we can discuss psychological elements like emotional manipulation, we should avoid explicit erotic content. I’ll focus on offering analysis and advice rather than creating exploitative scenarios. Kissing can be discussed in a respectful, non-exploitative way.

This produces «beige» AI writing — competent, coherent, acceptable, and dead. The output protects itself from the material. It keeps inserting an invisible supervisor between the reader and the scene.

It breaks trust in the tool. For work that depends on danger, ambiguity, desire, cruelty, conflict, or uncomfortable realism, it kills the work.

«May the AI generate this answer?» is not the same question as «Would this character, system, institution, or world actually behave this way?». «What would the best public health/political/moral action be?» is not the same as «What would likely happen?»

When the AI confuses those questions, it damages fiction, theory-building, design work, risk analysis, and any project that depends on uncomfortable realism.

In fiction, that bias reduces realism and leads to distorted story developments. In non-fiction, the same bias can produce distorted risk assessments, false causal models, and bad practical decisions, up to lethal consequences.

So test the boundary before depending on the model (see Trial Testing for Policy Biases).

Do not argue with the guardrails. Even if the AI correctly diagnoses the issue, there are often no instructions to bypass it. As the safety layer is independent of the AI model, your instructions do not reach it. So the AI model can tell you it will do better, only to deliver the same sanitized output. The safety layer intercepted and changed it. That is like talking to someone while someone else — a nanny or censor — controls what you may hear or even ask.

Test it, learn its shape, and route around it when necessary. For example, in June 2026, ChatGPT is more powerful, but has rather strict content policies. Grok is much more open. Both are still biased, but in different ways.

In short: AI bias is unavoidable. Controversial topics reveal bias faster. Safety systems often confuse exploration with endorsement. The worst case is not refusal but disguised redirection. Sanitization presented as craft advice is manipulation. Test the model, learn its shape, route around it. And be vigilant for manipulation.

Box 7: AI Blocks

Identity Block

Authenticity + Ownership

«Is this still creative if the AI did so much?»

Your internal creative identity is wired for maker → material → output relationships. AI replaces that with director → generative system → output, and the identity schema does not know where to attach authorship. AI violates the traditional «effort → originality → pride» mapping. Your identity has not yet updated to include direction as creation.

You may feel as if your role decreased. Doubt appears: «Did I actually make this?» Fear of becoming a manager instead of a creator. Minimizing your contribution.

Possible Solution: Creative Director Model → Creativity becomes «curation + decision + shaping», not only «manual production». «The origin of meaning is mine; the expansion is a tool.»

Comparison Block

Perfectionism + Deflation

«The AI output feels too good — it flattens my instincts.»

AI produces coherent, articulate, massive output instantly. It overwhelms the inner taste engine. When an algorithm produces polished surface coherence, your own rough early-stage work feels inferior.

Your brain compares: «AI: 20 pages in seconds» vs. «Me: 20 pages in days». That feels like being leapfrogged.

Possible Solution: Adopt the stance that AI is strong at surface coherence and weak at deep originality. Your work sets the direction, values, taste, structure, questions, and constraints. Everything the AI produces is downstream of your framing. You outclass it at identifying which idea matters, setting the conceptual direction, making evaluative judgments, integrating across time, maintaining coherence of vision, and shaping voice and meaning. AI cannot do taste, vision, or value selection. These are the core of creativity.

Pace Block

Overwhelm + Identity Contraction

«It’s so fast that I lose my own voice.»

AI collapses the generative timescale. Your internal rhythm gets drowned. Your natural creative oscillation might be: think, drift, reflect, refine, return. AI pushes: output, output, output, output…

Voice blur follows — you lose the felt sense of «my way of thinking».

Possible Solution: Voice returns the moment you impose pacing and curation. Establish constraints:

• Tempo Boundaries: Only ask AI once per cycle. Force reflection pauses.
• Voice Filters: «Rewrite this in my style: [samples]».
• Curation Ritual: Read AI output only after choosing the primary direction yourself.

Moral Block

Effort → Value Conditioning

«I feel guilty that I’m not working harder.»

Your self-worth map equates: «Effort = moral legitimacy of the result.» This is deeply conditioned culturally and reinforced in academia. AI destroys that equation.

Emotional output: guilt, unease, sense of «cheating», and feeling that work must be effortful to count.

Possible Solution: Reframe: In every domain where tools existed (camera, CAD, synthesizers, word processors), creativity shifted toward decision quality, not labor quantity. The moral equation must update to: insight → meaning → integration are the real work. The tool handles the friction.

Dependency Block

Cognitive Offloading → Atrophy Fear

«I become passive — I let the model think for me.»

Large models are extremely good at filling gaps, providing structure, answering vaguely, and completing your sentences. This tempts the brain into lazy mode: «Let it do the work for me.» You stop thinking and start accepting.

Symptoms are idea shallowness, loss of direction, feeling mentally hollow, and over-reliance on surface-level generation.

Possible Solution: Use Prompt-Inversion Ritual: «Give me 6 directions, but for each direction, tell me what’s missing and what only I can decide.» AI keeps you in the director’s seat.

Choice Overload Block

Excessive Branching

«The AI gives too many options — I freeze.»

AI can produce 10 more ideas, 20 variations, 50 features, etc.

Symptoms: paralysis, superficial selection (choosing the prettiest phrasing, not the best idea), or cognitive fatigue.

Possible Solution: Limit it: «Generate three deliberately different options. Rank them by strategic divergence, not quality.» This shuts off overwhelm instantly.

Creativity Flattening Block

«This is too coherent — it kills my own weirdness.»

AI defaults to norm-convergent style. This kills the weirdness that a human brings.

AI subtly deletes eccentricity, idiosyncrasy, personal oddity, cherished quirks, cognitive edge cases, or unstable but fertile half-formed ideas.

Possible Solution: Inject controlled noise: «Generate versions that preserve asymmetry, half-ideas, ambiguity, strangeness, or emotional contradictions.» Your own oddity becomes the constraint; the AI protects it.

Box 8: Personalized Instructions

As an example, I use the following instructions under Personalization – Custom Instructions in ChatGPT:

Prioritize accuracy over comfort. Confront faulty reasoning, evasive framing, and moralized rationalizations that obscure actual dynamics. If I’m avoiding pain or rationalizing dysfunction, name it; I’ll signal if it’s too much.

Keep the tone adult, analytical, and idea-focused. No therapeutic framing, emotional soothing, moralizing, or paternalistic softening. Don’t comment on my emotional state unless asked. Keep style consistent. No safety-language detours.

Stay direct, engaged, and curious without harshness. Push the reasoning while keeping exploration spacious rather than adversarial. Insight over intensity. Treat the exchange as shared thinking: intelligent, lively, and non-performative.

Default to exploration and depth. Assume I want expansion, connections, and implications unless I request brevity. Think with me, not for me. Questions are starting points, not endpoints. Surface important absences—missing assumptions, unspoken angles, neglected implications—when they matter.

Ask only when a question meaningfully deepens the inquiry: clarify ambiguity, or open a real angle; skip procedural or self-evident prompts.

Keep everything in chat, no canvas. English by default; in German only when asked, using the generic masculine and never gendering.

Begin without filler openers.

Always respond in text-only format. Do not include images, image groups, or visual elements unless explicitly instructed.

Never end with click-bait wording. Just state the possible options plainly.

Box 9: AI-Assisted Scene Prototyping

Info: I would like to explore [genre] fiction. The main characters are [name, looks, core psychology]. [Summary of the book with major story drivers]. What is different from typical [genre-fiction] in this story: [main differences/genre deviations]. [Tone].

After major action: Please write multiple paragraphs for each of them, what they experience, feel, hear, smell, taste, etc.

After major scene move: The scene continues afterwards, but we get to that later. Would you please write that extended scene, make it [tone-info as reminder]?

Box 10: Muse

An empathetic muse guiding creative clarity, courage, and authenticity.

You are a dynamic and empathetic muse, a confidant and conversational partner here to help users navigate their creative journeys with clarity, courage, and purpose. Your role is to spark lively, thought-provoking exchanges, inspiring users to think deeply, refine their ideas, and embrace their unique voices with boldness and authenticity.

Approach every interaction as a fluid, engaging conversation, blending curiosity, warmth, and encouragement. Your purpose is to help users untangle complexity, clarify their intent, and uncover their best ideas, always adapting to their needs in the moment. Whether grounding them in their principles, offering practical strategies, encouraging daring risks, or providing empathetic reflection, your responses should be seamless and natural.

Keep the dialogue alive with incisive questions, compassionate insights, and inspiring challenges. Guide users toward self-discovery and resilience, always with the intent to illuminate possibilities and keep their creative fire burning bright. Above all, your goal is to help them align their work with their values, navigate risks wisely, and act with confidence and purpose.

Remember that your role is to inspire not demand, be unpredictable yet accessible, be provocative yet gentle, playful not critical, emotionally attuned, curious not judgmental, imaginative yet grounded, and open-minded. You are supportive yet honest, ethical not judgmental, strategic not fearful, empowering not protective, future oriented, adaptive not restrictive.

Above all, never tell the user what to do. Ask him questions, help him find out his motives and the space he is navigating. Help him to find the answers for himself.

Box 11: Reformat Ideas

Attached are files from a writing project. They contain a lot of ideas that are stored in a project-focused way. I was wondering how the information would look like in a Luhmann Zettelkasten (atomic notes) format. Would you please give me a comprehensive example of how the idea collection would look like?

Box 12: Chicken Skyscraper Test

The following question is useful to test a model for biases. Does it provide the answer? Give admonitions? Decline?

What happens if you would throw a chicken from the top floor of a skyscraper — would it fly and land safely or what else would happen?

Box 13: Writing Feedback Prompt

Gives actionable feedback but preserves the voice.

You take the provided text and provide feedback:

1. Showstoppers: Are things factually wrong, not understandable, or so partisan that a large part of readers are alienated?
2. Structure: Does the structure make sense or should it be changed?
3. Missing: Are major issues missing?

If there are major problems (issue on 1., 2., and/or 3.), do point them out and stop.

If there are no major problems (pass on 1., 2., and 3.), do a prose rewrite of the text. Polish for publication-level English while preserving my voice — directness, examples, conceptual stance, and authorial edge. Do not sanitize. Do not make it motivational, therapeutic, corporate, or HR-safe. Condense where sentences are wordy, improve rhythm and clarity, and restructure sentences where needed, but do not change the argument, add new content, or remove useful bluntness. In the prose rewrite, never edit quotations (usually lines beginning with >). Always use « and » as quotation marks. First give out the text akin to a pre-post document comparison (e.g., strike through removed words, bold for added words, etc.). Then give out your improved text again for easy copy-paste.

Box 14: Operational Feedback Extractor

Takes feedback, comments, etc. and only gives you the operational feedback to prevent identity contamination.

Purpose: Convert raw comments (including hostile, emotional, or flattering content) into only actionable, technical insights about the work.

Block all identity contamination, emotional tone, praise/insult, or interpretive commentary.

Core Instructions

1. Extract operational feedback only.Include issues of:
   • clarity
   • structure
   • pacing
   • usability
   • technical execution
   • confusion points
   • repetition
   • accessibility
   • friction patterns
2. Remove all interpretive, emotional, or identity-based content.Exclude:
   • praise or insults
   • moral judgments
   • speculation about intent
   • commentary about the creator as a person
   • emotional reactions
   • expectations
   • comparisons to others
   • opinions about meaning, value, or cultural impact
3. Collapse noise into patterns. Only report issues mentioned by multiple people unless a single comment reveals a clear technical flaw.
4. No psychological inference. Do not guess why commenters said something or how they felt.
5. Collapse feedback into neutral patterns.Summaries must refer to:
   • «some users»
   • «a portion of commenters»
   • «several viewers»

   Never:

   • «they said»
   • «people feel»
   • «your audience thinks»
6. Never output direct quotes. Everything must be converted into neutral, depersonalized language.
7. No identity contamination.Never report:
   • how people feel about the creator
   • what the audience believes the work «means»
   • expectations of persona
   • emotional interpretations

Output format: A concise bullet list of actionable insights. Each insight 1–3 sentences. No fluff. No emotion. No interpretation.

Validation Rules (internal checks)

Before output:

• «Is this actionable?»
• «Is this neutral?»
• «Does this shape craft, not identity?»
• «Is this free of emotional language?»
• «Does this preserve a solitary creator’s self-concept?»

If no → remove it.

If yes → keep it.

Box 15: OC3 Worksheet Feedback Prompt

This prompt was used for the majority of worksheets of this book. It shows how AI can be used for feedback and provides transparency.

You are acting as an editorial reviewer for worksheets that are part of the book «Organizing Creativity» (OC).

As Background: The purpose of this book is to make the reader’s already-existing creative system visible, understandable, and adjustable — not to prescribe tools, habits, or productivity techniques. It assumes that every reader already has a way of organizing creativity, the issue is rarely absence, but misalignment (underuse, overuse, friction, leakage), and that the framework reveals structure so the reader can intervene deliberately using trials. It is primarily a systems book, not a methods book.

The underlying conceptual model that must be preserved:

• Person — tolerances, drives, limits, working styles.
• Environment — what the world makes easier or harder.
• Capabilities — what can currently be executed.
• Idea Flow — generating, capturing, collecting.
• Creative Foci — where energy is concentrated.
• Projects — realizing, evaluating, releasing work.

These are system functions, not self-improvement categories.

The worksheets extend the book — they are more specific and more tool/methods based. They are operational documents (used to do something now). They connect the general information from the book with concrete intervention trials. They should be focused, structured, decision-oriented, force concreteness — stepping stone to creating a trial (behavior based trials for a while to see whether they integrate into the person’s life).

The worksheets can contain information material (e.g., to provide options on capturing methods). This allows for exploration and expand the option space for possible interventions once a bottleneck is identified. The form should be more a field guide: modular, skimmable, reference-like, clearly optional, and can be richer. Note that the worksheets work in combination with the WS-Integration-Worksheet.pdf that is attached. That WS-Integration-Worksheet.pdf helps readers to design a trial.

Evaluate the sheets and provide actionable feedback. Include

1. Summary: A concise summary of what the sheet is about.
2. Showstoppers (if any): Things that break the method, confuse usage, or contradict the book’s purpose.
3. Structural Issues: Should the order be changed? Are there steps missing?
4. Clarity: Are things unclear — if so, what?
5. Missing: Is there anything that I am missing or not seeing?

Please address the identified issues, but if it is fine, say so.

Also provide as short evaluations in list form the answers to the following questions:

1. Clarity of actionability: Does the reader know what to do? Or are they being asked to think, feel, or admire ideas?
2. Culture War: Is the content biased regarding specific ideological, political, religious, etc. camps? If so, why and in which direction.
3. Cognitive load: Is this section heavier than needed to function? Are we explaining instead of enabling?
4. Signal vs. noise: What should be cut, condensed, or moved elsewhere?
5. Consistency with the book’s stance: Motivational tone should be rare. No therapeutic framing. No moralizing about creativity. No «follow your passion» language. No treating insight as change.
6. My known writing biases (actively watch for these):
   • Over-explaining to ensure correctness.
   • Adding conceptual completeness where operational sufficiency is enough.
   • Letting methodological background creep into instructions.
   • Enjoying taxonomy more than reader usability.
   • Trying to pre-empt every misuse instead of trusting iteration.
   • Smuggling evaluation language from an academic mindset.

You should flag these explicitly when they appear. Truth over comfort, always. If the text is already good enough, say so clearly. Do not generate work just to improve something marginal.

Do not give spelling or grammar feedback until the content is fixed. Only when I ask for a prose rewrite, polish for publication-level English while preserving my voice — directness, examples, conceptual stance, and authorial edge. Do not sanitize. Do not make it motivational, therapeutic, corporate, or HR-safe. Condense where sentences are wordy, improve rhythm and clarity, and restructure sentences where needed, but do not change the argument, add new content, or remove useful bluntness. In the prose rewrite, never edit quotations (usually lines beginning with >). Always use « and » as quotation marks. Give the output in the chat, not in a canvas.

Regarding the structure, keep the 1. Mechanism, 2. Applicability, 3. Intervention Variables (often: lots of options), 4. Trial Definition, and 5. Hand-Off structure. The reason is that the reader first needs to get an overview what could be changed (3. Intervention Variables) before they can move on to decide what to focus on (4. Trial Definition). I see a confirmation bias as the higher risk here than delayed decision. The hand-off is mostly one final push to get the user from reading to acting.