Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
29 changes: 29 additions & 0 deletions latex/.gitignore
Original file line number Diff line number Diff line change
@@ -0,0 +1,29 @@
# Build artifacts
build/
*.aux
*.log
*.toc
*.lof
*.lot
*.out
*.bbl
*.bcf
*.blg
*.fls
*.fdb_latexmk
*.synctex.gz
*.xdv
*.run.xml

# Helper scripts (generated, not part of submission)
convert_citations.py
enrich_bib.py
fix_links.py

# Mermaid source and rendering helpers
figures/*.mmd
figures/render_browser.mjs
figures/render_maps.py
figures/render_wardley.html
figures/upcycle_diagrams.py
figures/generate_charts.py
2 changes: 2 additions & 0 deletions latex/00README.XXX
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
nohypertex
xelatex main.tex
14 changes: 14 additions & 0 deletions latex/Makefile
Original file line number Diff line number Diff line change
@@ -0,0 +1,14 @@
LATEX = latexmk
FLAGS = -xelatex -output-directory=build -interaction=nonstopmode

.PHONY: all clean distclean

all:
@mkdir -p build
$(LATEX) $(FLAGS) main.tex

clean:
$(LATEX) -C -output-directory=build

distclean: clean
rm -rf build/
345 changes: 345 additions & 0 deletions latex/appendices/a1_quantitative_anatomy.tex

Large diffs are not rendered by default.

105 changes: 105 additions & 0 deletions latex/appendices/a2_glossary.tex
Original file line number Diff line number Diff line change
@@ -0,0 +1,105 @@
\chapter{Glossary}\label{app:a2}

\emph{The terms in this glossary are the working vocabulary of the book. Some are coinages, some are borrowed from elsewhere and reframed; all are used with precise, deliberately narrow meanings in the chapters above.}

\bigskip\par\noindent\rule{\textwidth}{0.4pt}\par\bigskip

\begin{description}[style=nextline,leftmargin=2em]

\item[\textbf{Agency line} (or Continuum line).]
A single axis representing the share of decision-making in a creative function performed by a human versus a machine system. See \textbf{Human--AI Agency Continuum}.

\item[\textbf{Agentic AI.}]
A class of AI system that, given a goal, can plan, decide and execute a sequence of multi-step actions without further human input between steps. Distinct from a \emph{generator}, which produces an output in response to a single prompt. See Chapter~3.

\item[\textbf{AI literacy.}]
The cluster of skills required to deploy AI tools effectively in creative work -- including briefing, taste, judgement, prompt practice, output evaluation and tool-stack fluency. The term moved from optional to baseline competency through 2025--26, formalised by initiatives such as the Sundance Institute's AI Literacy Initiative launched in January 2026.

\item[\textbf{AI slop.}]
Low-quality, mass-produced AI-generated content that is recognisable to audiences as such -- usually because it is made without human creative intent. \emph{Merriam-Webster}'s word of the year for 2025. See \textbf{Slop Ceiling}.

\item[\textbf{Attribution.}]
The principle that when AI systems produce derivative outputs based on training data, the human authors whose work shaped those outputs should be identified and -- where appropriate -- compensated. One of the four principles of the generative creative economy (Chapter~15). Technical infrastructure includes C2PA, SynthID, and creative-weight-attribution systems.

\item[\textbf{Audience contract.}]
The implicit agreement between makers and audiences about what creative work is, what conditions of making it carries, and what relationship the audience can expect with its makers. The shift from implicit to explicit audience contracts is one of the central structural changes the book describes. See Chapter~12.

\item[\textbf{Augmented intelligence.}]
Reframing of ``AI'' used by some industry voices to emphasise human-in-the-loop deployment over autonomy. Compare with \textbf{Generative AI}, \textbf{Agentic AI}.

\item[\textbf{C2PA.}]
\emph{Coalition for Content Provenance and Authenticity} -- a technical standard for embedding cryptographic provenance metadata in media files, supported by camera manufacturers, editing software and a growing number of platforms. The principal ``fingerprint real media'' infrastructure underlying the authenticity argument in Chapter~12.

\item[\textbf{Coordination collapse.}]
The structural change that occurs when the labour-coordination architecture of a creative organisation -- built around the bandwidth constraints of human-only teams -- is overtaken by AI-assisted workflows that no longer require those constraints. The subject of Chapter~13. Manifests as \emph{shadow AI} below management sight and as compressed middle layers in the workforce above.

\item[\textbf{Dead Internet Theory.}]
The notion that most of the public web is now bot-generated and bot-read, with humans increasingly a minority of traffic. Once a conspiracy framing; by 2025 a measurable phenomenon -- bots accounted for 51\% of web traffic in the Imperva 2025 Bad Bot Report, of which \textasciitilde80\% of bot traffic was AI training crawlers. See Chapter~4.

\item[\textbf{Disclosure.}]
The practice of declaring the use of AI in the production of a piece of creative work -- in credits, contracts, metadata, watermarks, or platform-facing labels. By spring 2026, disclosure had emerged as the dominant industry response to the audience-authenticity question, anchored by standards including the \textbf{Cannes AI Disclosure Standard} (May 2026) and the \textbf{Academy of Motion Picture Arts and Sciences} rule requiring human authorship for awards eligibility.

\item[\textbf{Extractive economy.}]
A creative economy in which AI systems are trained on unpaid human work, the platform companies that build the models capture most of the resulting economic value, and the diffuse pool of working creatives is steadily decapitalised. One of two possible end-states the book identifies. See Chapter~15.

\item[\textbf{Fingerprint real media.}]
Adam Mosseri's (Instagram) framing of the verification problem: amplify provably-human content rather than chase synthetic content for labelling. Used in the book as shorthand for \emph{provenance-first} approaches to content moderation. See Chapter~4.

\item[\textbf{Generative economy.}]
A creative economy in which AI tools are treated as new craft infrastructure, training data is consented and compensated, platforms compete on tool quality and integrity, and the productivity gains are broadly distributed rather than concentrated. The opposite of the \textbf{extractive economy}. The four principles (\emph{Agency, Attribution, Access, Audience}) are the operational test. See Chapter~15.

\item[\textbf{Human--AI Agency Continuum.}]
A frame, introduced in \emph{Dream Machine} Issue~2 (October 2025) and extended in Chapter~3, in which any given creative function is mapped on a horizontal line from full human agency (left) to full machine agency (right). The frame's key claim is that \emph{each creative function moves at its own speed} -- you can sit at the extreme left on performance while being at the right edge on plate generation.

\item[\textbf{Living Web.}]
The deliberately-built portion of the public web in which authorship is provable, attribution is durable, attention is allocated on non-viral signals, and the architecture supports rather than undermines human creative work. The aspirational opposite of the \textbf{Dead Internet}. Has to be built, not assumed. See Chapter~4.

\item[\textbf{Mid-career squeeze.}]
The structural pressure on workers in the middle of creative-industry careers -- neither at the junior entry level (replaced by agents) nor at the senior decision-making level (still required) -- as AI absorbs the intermediate production roles that those mid-career workers historically held. See Chapter~13.

\item[\textbf{Model collapse.}]
The technical risk that AI systems trained predominantly on synthetic data -- including data produced by earlier generations of AI systems -- progressively lose touch with real-world signal and produce increasingly homogenised, hallucination-prone outputs. The risk that \emph{Dead Internet, Living Web} warns has moved from theoretical to measurable.

\item[\textbf{Orchestrator.}]
The role that emerges when an individual working creative -- or a small team -- directs a large pool of AI-agent capacity. Defined operationally by five activities: defining the brief, allocating work, briefing the agents, judging outputs, and integrating the result. Predicted in \emph{Dream Machine} Issue~13 (January 2026) as the dominant role of 2026; documented across the chapters above. See Chapter~11.

\item[\textbf{Pipeline of authorship.}]
The full chain from creative intent to delivered work, broken down into discrete functions (writing, direction, performance, image-making, sound, edit, distribution, marketing). The point of the \textbf{Human--AI Agency Continuum} is that each link in this chain has its own agency line.

\item[\textbf{Position One} (All-in).]
The strategic posture of legacy studios that have decided to integrate AI aggressively across all production functions, betting that early-adopter advantage will compound. Netflix's ``all in'' framing, October 2025. See Chapter~7.

\item[\textbf{Position Two} (AI-native).]
The strategic posture of new entrants that build their production pipelines AI-first from inception -- Imaginae Studios, Wonder Studios, Obsidian Studio, Asteria, Wonder, Chapter41, Kartel. The category exploded in scale through autumn 2025 and winter 2026. See Chapter~7.

\item[\textbf{Position Three} (Refusal).]
The strategic posture of creative organisations that have explicitly excluded generative AI from their work. Jagex, Larian, Games Workshop, Hooded Horse, Aardman (qualified), Pocketpair. Cultural authority is preserved as the principal asset. See Chapter~7.

\item[\textbf{Position Four} (Middle).]
\emph{AI in the workflow, not in the work.} The strategic posture -- taken by Sony, Bethesda, Amazon (in \emph{House of David}), Aardman (in qualified form), and an increasing number of major studios -- that uses AI to augment production pipelines while preserving human creative intent in the moments the audience sees. The book's prediction for where most surviving major studios land by 2030. See Chapter~7.

\item[\textbf{Provenance.}]
The chain of custody of a creative work from its origin (capture, performance, writing, sketching) to its delivered form. Technical standards (C2PA, SynthID) and policy frameworks (Cannes Disclosure Standard, SAG-AFTRA AI protections) collectively constitute the \emph{provenance infrastructure} the book argues is critical for the next decade. See Chapter~12.

\item[\textbf{Shadow AI.}]
The use of AI tools by employees outside their employer's official tooling, processes and accounting. Documented in 2025 workplace research as encompassing approximately half of the U.S. workforce. The principal symptom of \textbf{Coordination Collapse}. See Chapter~13.

\item[\textbf{Slop Ceiling.}]
The empirical pattern, established across multiple sectors by spring 2026, in which AI-generated creative content can be produced in massive volume but consistently fails to capture audience attention proportionate to that volume. Anchored in the \textbf{44\%/3\%} ratio Deezer published in April 2026 (44\% of daily uploads AI; under 3\% of streams). One of the central claims of the book. See Chapter~5.

\item[\textbf{Synthetic sincerity.}]
The category of creative work that is openly synthetic but made with serious creative intent and is not pretending to be something else. Named after Marc Isaacs' 2025 IDFA documentary. Audiences distinguish \emph{synthetic sincerity} from \emph{synthetic cynicism} at the speed of a swipe. See Chapter~4.

\item[\textbf{Tilly Tax} (informal).]
The collection of contract provisions in SAG-AFTRA's spring 2026 agreement requiring compensation, consent and residuals when AI replicas of human performers are used. Named after the Tilly Norwood controversy of September 2025 that catalysed the broader negotiation. See Chapters~5, 10.

\item[\textbf{Watermark.}]
A persistent identifier embedded in AI-generated outputs by the producing system, intended to allow downstream detection that content is synthetic. SynthID (Google) and similar systems became standard across major platform tools through 2025--26.

\item[\textbf{World model.}]
A class of generative AI system that produces navigable three-dimensional environments rather than flat output. Marble (World Labs, public release November 2025) was the first commercial product in the category; Google DeepMind's Genie~3, Meta's WorldGen, Luma's UNI-1, Tencent's Hunyuan World, SpAItial's ECHO and others followed within months. See Chapter~8.

\end{description}

\bigskip
\emph{If a term in the book did substantial work but does not appear in this glossary, please tell us so we can improve it in the next edition.}
Loading