AI Control Profiles: A Taxonomy for Creative Direction
A framework for measuring and disclosing the degree of human creative control in AI-assisted writing.
THE MACHINE ROOM IS OPEN.
Research, workflows, white papers, experiments, and field notes from the edge of AI-assisted authorship and publishing.
Explore the Lab
Research
Long-form analysis on AI authorship, creative control, and publishing futures.
Explore →Workflows
Repeatable processes, prompt templates, and editorial systems for AI collaboration.
Explore →White Papers
Formal publications on standards, disclosure frameworks, and industry infrastructure.
Explore →Field Notes
Subscribe for updates on research, workflows, and observations from the field.
Explore →Videos
Recorded talks, tutorials, and visual explorations of AI-assisted authorship.
Explore →Experiments
Lab tests, case studies, and results from ongoing research.
Explore →Latest Research
Validation, Not Editing: The New Author Workflow
How to work with AI as a validator and iterative collaborator rather than a text editor.
Workflow Library
This site documents repeatable author workflows: AI-assisted publishing systems, prompt chains, editorial processes, and production experiments.
Prompt Engineering Checklist for Fiction
A step-by-step guide to crafting effective prompts for AI-assisted narrative generation.
White Papers
Formal publications on AI in publishing, industry standards, and disclosure frameworks.
The Novel Writing Machine
An agentic system for long-form fiction organized like a Formula 1 race weekend. Fifteen specialized roles protect author judgment at the moments where it actually matters.
Names Carry Weight
A character’s name is never a neutral label. Holding a prompt constant and changing only the name reorganizes setting, backstory, cadence — even a character’s gender. A craft-and-ethics look at how names import bias into AI-assisted fiction, and how to take back control.
Reinforcement Logic
For years the rule was “never tell the AI what not to do.” The models changed; the rule didn’t. This paper introduces reinforcement logic — pairing an allowed state with its forbidden opposite so a single intention is reinforced from both sides — and proves it with a live three-condition experiment.
The Em Dash Problem
The em dash became the great tell of AI prose. The reason isn’t stylistic — it’s structural. A short, shareable case study in how tokenization makes a single punctuation mark load-bearing, and what that reveals about prompt engineering.
Why a Single Word Changes Everything
Prompt engineering gets dismissed as folklore because nobody can see what a model does with their words. This flagship paper opens the black box for authors — tokens, token IDs, embeddings, and the weighted vector field where meaning lives — then proves with a single-word experiment that one changed word reorganizes an entire scene.
Video Log
Recorded talks, tutorials, and visual explorations of AI-assisted authorship.
Prompt Engineering IS Back!
The full Future Fiction Academy session on how language models actually read your prompts — tokens, embeddings, names, and the em dash. The source session behind the four-part white paper series.