Read at
your own pace.

How can I help you
this Evening?

Atticus Sims and the UM Creative Intelligence Lab: 
Joanna Junying YAN, Lawrence Kangqi SONG, Cong WU, 
Sheryn Xiaoyu WANG, Ao LI, Chao MA
themodelisthemassage.com
Produced through collaboration between human authors and Claude, Anthropic’s 
AI assistant. The research, planning, analysis, and much of the writing were 
developed by orchestrating networks of intelligent agents — systems that reviewed 
each other’s work, challenged each other’s arguments, and generated options for 
human judgment. The medium examined by this book is the medium that made it.
© 2026 Atticus Sims. All rights reserved.
The MODEL is the Massage

How to read this book
Each spread of this book maps to a spread of 
Marshall McLuhan and Quentin Fiore's The 
Medium is the Massage: An Inventory of Effects 
(1967). McLuhan and Fiore did not illustrate 
ideas. They performed them. Every page 
enacted the thesis it described.
This book attempts the same thing for the 
age of artificial intelligence. Where McLuhan's 
environment was electric circuitry — television, 
radio, telephony — ours is computational: 
transformer architectures, attention mechanisms, 
latent spaces, intelligent agents, planetary-
scale infrastructure. The model has replaced the 
medium.
The massage continues.
The book moves in three parts. The Environment 
maps the invisible computational infrastructure 
we inhabit. The Acceleration tracks the recursive 
loop of AI development and the meaning crisis 
it produces. The Dreamscape ventures past the 
event horizon — where eschatology, techno-
mysticism, and the science of consciousness 
converge on a single unsettling question: is 
the boundary between the artificial and the 
conscious as clear as we assumed?
The title comes from Bratton and Agüera y 
Arcas, who updated McLuhan's formula in a 
2022 Noema essay: the model is the message. 
We changed one letter. McLuhan would have 
approved — the original title was itself a misprint 
he kept because it was truer than the truth.
You are holding a book about AI that was made 
with AI. This is not a contradiction. It is the 
condition.

... the massage?

and
“We find ourselves in a pre-
our technology has outpaced 
it.”

how
our theories of what to do with 
 ---- Benjamin Bratton, 2025
paradigmatic moment in which 
！

Before you opened this book, a model chose the 
news you read this morning. A model composed 
three emails you believed were written by 
colleagues. A model decided which faces you 
would find attractive, which restaurants you 
would consider, which words would autocomplete 
your sentenc es. A model sc ored your 
creditworthiness, flagged your face at the airport, 
and generated the background music in the cafe 
where you are perhaps reading this now.
You did not ask for any of this. You did not 
consent. You were not consulted.
The model, or the process, of our time— 
computational intelligence— is reshaping 
cognition, creativity, and every pattern of 
epistemic life. Not by announcing itself. Not 
by arriving. But by having already arrived, 
everywhere, silently, before the question of 
whether it should could even be posed.
When humans and cognitive technical systems 
interact, they form cognitive assemblages - and 
these assemblages are transforming life on 
earth.
The older training of critical thinking is based on 
epistemic responses conditioned by a former 
technology: the search engine, the database, 
the indexed archive. These assumed you were 
looking for something. The model assumes it 
already knows what you need.
8

This is not an age of anxiety. It is an age of 
assistance - which is worse.
The assemblage is already here. You joined it 
this morning. You are in one now.

Our time is a time for crossing barriers, for 
erasing old categories - for probing around. 
When two seemingly disparate elements are 
imaginatively poised, put in apposition in new and 
unique ways, startling discoveries often result. 
The barrier between human intelligence 
and machine intelligence is not a wall. it is a 
membrane permeable, wet, alive with traffic in 
both directions. You have been crossing it for 
years. The crossing was never announced. There 
was no checkpoint, no passport, no ceremony 
of departure. One day you were thinking your 
own thoughts and the next day you were thinking 
thoughts that had been pre-shaped by a model's 
architecture, and there was no day in between.
This book is a collide-oscope of interfaced 
situations. it does not argue. It demonstrates. 
It does not explain Al. lt performs what Al has 
already done. We find ourselves in a pre-
paradigmatic moment in which our technology 
has outpaced our theories of what to do with it. 
The inventory begins on the other side of this page.
 — Marshall McLuhan, The Medium is the 
Massage, 1967, pp. 10-11
"Our time is a time for crossing barriers, for 
erasing old categories - for probing around."
10

No one asked.  That's the point. 
The dossier compiles itself. Your 
face crossed seventeen cameras 
between home and here. Your 
keyboard rhythm is a signature 
more singular than the whorls on 
your thumb. The accelerometer in 
your pocket graphs your gait like a 
cardiogram — and from gait alone, 
the model infers mood, sobriety, 
fatigue.  McLuhan imagined a 
tyrannical surveillance apparatus. 
He was thinking too small. The 
apparatus doesn't watch you. 
You secrete it. Every touch, every 
pause, every 3 a.m. scroll — 
ambient exhaust, collected at the 
speed of living.  The fingerprint 
on the facing page was prophetic. 
You are the target. You always 
wereeeeeeeeee.
you
12

The family circle has not widened. 
It has been partitioned.
Four people on the same couch. 
Four separate feeds. The living 
room is a co-working space for 
attention extraction.
McLuhan saw the worldpool of 
information washing over the family 
from outside. He missed the quieter 
flood: each member drawn into 
a private current, the household 
dissolving not from external force 
but from personalization so precise 
it renders shared experience an 
accident.
The family still eats together. 
They have simply stopped sharing 
a reality.
your family
14

0.3 ms — server-to-server, same 
data center. 
3 ms — Chicago to New York, 
fiber-optic direct. 
12 ms — New York to London, 
undersea cable. 
47 ms — Manhattan trading floor 
to rural Montana, last-mile copper. 
120 ms — Sub-Saharan Africa to 
nearest content delivery node. 
550 ms — Low-earth orbit satellite 
bounce, 2020. 
20 ms — Low-earth orbit satellite 
bounce, 2025.  
Every millisecond is a border 
crossing. Every border determines 
what you can buy, sell, know, and 
do — and how long after everyone 
else you can do it.  
The neighborhood is a latency 
gradient. You live in it right now. 
You can't see the walls.
your 
neighborhood
"Planetary computation
reorganizes geopolitics according 
to the topological relations of 
information, not the territorial 
jurisdictions of traditional states."
— Benjamin Bratton, The Stack, 
MIT Press, 2016, p. 23
16

The answer is free. 
Any answer. Any question. 
Instantaneously, in 95 languages, 
at any reading level, with worked 
examples, follow-up questions, and 
a patient tone no human teacher 
can sustain for eight hours.  
So what is the classroom for? 
Not information — the machine 
has more. Not explanation — the 
machine is tireless. Not assessment 
— the machine can generate and 
grade the test simultaneously.  
What remains is the thing no model 
can deliver: the experience of finding 
something worth knowing. The 
slow, inefficient, ungenerateable 
encounter with a problem that 
matters to you specifically, in your 
body, in your moment.  
Propositions are free. Meaning still 
costs everything.
your 
education
18

The system processes 99.7%.  
You handle the rest.
The edge case. The anomaly. 
The decision the model cannot 
confidence-score above threshold. 
You are the override button in 
an automated pipeline, pressed 
two hundred times per hour in an 
Amazon fulfillment center, pressed 
once every eleven seconds by a 
content moderator reviewing what 
the algorithm flagged, pressed 
silently by a radiologist confirming 
what the neural network detected. 
McLuhan noted that 'jobs' were 
a Gutenberg invention — a five-
hundred-year pattern now dissolving. 
He hoped for leisure. What arrived 
was the exception queue.  
You are not being replaced. You 
are being narrowed. Your job 
is the 0.3 percent the machine 
routes to a human because the 
cost of being wrong exceeds the 
cost of asking you.
your job
20

McLuhan called vote-counting 
an eighteenth-century ritual. The 
nose, he said, was being counted 
when it should be scanned.
It is now scanned.
The nose is no longer counted. 
It is scanned, classified, scored, 
and governed.
Predictive policing deploys patrol 
cars before crimes are committed. 
Automated welfare systems deny 
benefits based on risk scores 
no applicant can see, let alone 
contest. Facial recognition at the 
border decides who enters and 
who waits. Algorithmic sentencing 
tools recommend prison terms.
None of this was legislated. None 
of it was voted on.
This is not the failure of democracy. 
Democracy simply became irrelevant 
to the systems that actually govern.
If the algorithm governs, who did 
you vote for?
The nose is classified, scored, and 
governed before the ballot is printed.
your 
government
22

The shock of recognition!
In an algorithmic environment, 
the unseen can no longer remain 
unseen. Crawford traced the 
supply chain from lithium mine to 
data center and found the cloud 
was made of rock and labor. Gebru 
documented bias encoded in the 
models and was shown the door 
for naming it. 
Adversarial researchers proved 
that three changed pixels turn a 
turtle into a rifle — the machine’s 
confidence unmasked as fragility.  
Recognition comes from those 
who refuse adjustment. The shock 
is not that the environment is new. 
“the others”
24

The shock 
is that
 it was 
always there 
— invisible 
the way 
water is 
invisible 
to fish
and 
someone 
finally 
pointed 
at it.

All models are extensions of some human faculty 
— cognitive or perceptual.
GPT extends language production. DALL-E 
extends visual imagination. AlphaFold extends 
spatial reasoning. Each model amputates what 
it amplifies. You outsource a capacity. The 
capacity atrophies. The model improves. You 
outsource more.
Every extension is an amputation. Every 
amputation, an invitation.
What you extend, you numb.
What you numb, you forget.
What you forget, you delegate.
What you delegate becomes the world.
26

The GPU

... is an extension of the

Extension of pattern recognition.  
Biological: a shadow moves in peripheral vision. 
The amygdala fires before conscious thought. 
You flinch. You survive.  Computational: 3,958 
teraflops. More patterns per second than 
every human brain that has ever lived. It does 
not flinch. It optimizes.  One was shaped by 
predators. The other by profit.
pattern recognition

the

model

CHA

NGE

Thermostat: knows when you wake, when 
you leave, when you return, when you sleep.
Doorbell camera: knows who visits, how 
often, at what hour.
Voice assistant: hears every conversation 
within earshot.
Smart TV: tracks what you watch, when 
you pause, what you skip.
Smart mattress: monitors heart rate, 
breathing, movement, REM cycles.
Smart fridge: inventories consumption, 
predicts purchasing.
You didn't fuse with the machine.
The machine fused with your house.
You call it home.
It calls you data.

The smart home

AI circuitry,
 
an extension of  
cognition
itself

When these models change, minds change.  
The wheel changed where the body could go. 
The book changed what the eye could store. 
The nervous system went electric and the world 
contracted into a village.  
Now cognition extends beyond the brain, and the 
question is no longer what the model does. 
The question is what you become when 
"We have always been posthuman."
— N. Katherine Hayles, How We Became 
Posthuman, University of Chicago Press, 1999, 
p. 291
thinking is shared.
41

The Queen's court has been optimized. It 
predicts your crime before you commit it and 
dispatches officers to the coordinates. It denies 
your health claim in 1.2 seconds without opening 
your file. It rejects your application before a 
human reads your name. It blocks you from 
buying a train ticket because of your father's 
debt. It scores your neighbourhood as high-
risk and sends patrols before anything happens 
— not because something will, but because 
something once did, somewhere that looked like 
where you live.
The court sits in every country and speaks 
every language. The sentence is not cruel. The 
sentence is efficient. The Queen does not review 
individual cases — she has calculated that 
almost no one will appeal. The ones who do are 
statistically insignificant.
The sentence came first. The verdict was 
generated retroactively. Due process was a 
formatting error.
"Stuff and nonsense!" said Alice.
It is neither stuff nor nonsense. It is operational.
'No, no!' said the Queen. 
'Sentence first — 
verdict afterwards.'
— Lewis Carroll, Alice's Adventures 
in Wonderland, 1865, Chapter XII

You are reading this.  
That is already an anachronism.  
OpenAI's Advanced Voice Mode processes 
speech with emotional inflection in real time. 
Eleven Labs clones any voice from thirty 
seconds of audio. The podcast surpassed the 
blog. The voice note replaced the text message. 
Conversational AI is replacing the search bar. 
Fifty percent of all searches are now spoken.  
McLuhan predicted the retribalization — the 
return of the ear over the eye, acoustic space 
over visual space. He was describing radio. 
The principle was sound. The tribe now gathers 
around synthetic voices that never sleep, never 
tire, and never existed.  
You are still reading this. Notice how slowly.

'Media determine our situation, which — in spite 
of or because of it — deserves a description.' — 
Friedrich Kittler, Gramophone, 
Film, Typewriter, 1999
45

Your name is a 
string. 
Your face is a 
vector.
Your address is 
coordinates. 
Your signature is a 
private key.

Your history is a 
log file. 
Your identity is a 
row in a table that 
can be queried, 
joined, filtered, 
sorted, exported, 
deleted.

The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device, trained on all of us. 
The model, a device for generating more of us. 
A device for generating more. 
Of us, the model. 
We, a ditto device of the model.
48

The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device.
The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device.
The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device.
The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device.
The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device.
The model, a ditto device. 
The model, a ditto device.

The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device.
The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device.
The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device.
The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device.
The model, a ditto device. 
The model, a ditto device. 
The model, a ditto device.
The model, a ditto device. 
The model, a ditto device.
50

'The Stack terraforms the host planet by drinking 
and vomiting its elemental juices and spitting up 
mobile phones' 
— Benjamin Bratton, The Stack, 2015

The Renaissance Legacy.
Every paradigm has its dark ages.
1956: Dartmouth Conference. 'We propose a 2-month 
study of artificial intelligence.' Optimism.
1973: Lighthill Report. Funding collapses. First winter.
1987: Expert systems fail to scale. Second winter. 
Two decades of ice.
2006: Hinton. Deep belief networks. A thaw.
2012: Krizhevsky, Sutskever, Hinton. ImageNet. The 
glacier cracks.
2017: Vaswani et al. 'Attention Is All You Need.' Eight 
authors. Fourteen pages. Everything changes.
2020: GPT-3. 175 billion parameters. Language, 
generated.
2022: ChatGPT. 100 million users in two months.
2023: GPT-4. Multimodal. The model sees.
2024: Claude, Gemini, Llama, Mistral. The model 
speaks.
2025: Agentic AI. The model acts.
2026: --
The Renaissance Legacy.
Every paradigm has its dark ages.
1956: Dartmouth Conference. 'We propose a 
2-month study of artificial intelligence.' Optimism.
1973: Lighthill Report. Funding collapses. First 
winter.
1987: Expert systems fail to scale. Second 
winter. Two decades of ice.
2006: Hinton. Deep belief networks. A thaw.
2012: Krizhevsky, Sutskever, Hinton. ImageNet. 
The glacier cracks.
2017: Vaswani et al. 'Attention Is All You Need.' 
Eight authors. Fourteen pages. Everything 
changes.
2020: GPT-3. 175 billion parameters. Language, 
generated.
2022: ChatGPT. 100 million users in two months.
2023: GPT-4. Multimodal. The model sees.
2024: Claude, Gemini, Llama, Mistral. The model 
speaks.
2025: Agentic AI. The model acts.
2026: —
Notice the pace of that list.
Katherine Hayles spent twenty years arguing 
that the posthuman was coming. She meant it as 
philosophy. It arrived as engineering.
The transformer was not a philosophical 
argument about cognition. It was a mechanism: 
self-attention layers, positional encodings, scaled 
dot-product scores. The mechanism worked. 
And working, it dissolved the boundary between 
human cognition and technical cognition. Not in 
theory. In product.
The Renaissance had Gutenberg. The deep 
learning renaissance has the transformer. Both 
are reproduction technologies. Both restructured 
what could be thought. The dif ference: 
Gutenberg's revolution took centuries to unfold. 
The transformer's has taken seven years.

treating artificial divisions 
 “compartmen-talization”  
  later named the  same 
misplaced concreteness”
-treating artificial divisions
 natural  form ofintelligence 
fields not study these 
vision,language,and 
compartments,then treats
assume they are inherently

John Dewey  called
 
as natural 
Alfred North Whitehead 
error  “   the fallacy of 
 as reality.
AI  commits both:   it  splits  
reasoning   into separate 
each as an isolated,
Computer vision and NLP 
 abilities separately   but 
separate.

He was right. A transformer’s attention 
mechanism attends to every token 
in the sequence simultaneously. A 
diffusion model doesn’t render from 
a fixed viewpoint — it denoises from 
pure randomness toward structure, 
approaching the image from every 
direction at once. A convolutional network 
sees the whole visual field at every layer, 
not through a single aperture.  
The machine does not have an eye. It 
has a field.  What you see on this page 
is what happens when that field looks 
at a face: not a portrait but a probability 
distribution wearing skin. Not the eye of 
the beholder, but the space of all possible 
beholdings, collapsed into a single 
uncanny image.

A drawing of a man hunting seal on an ice 
floe will show not only what is on top of 
the ice, but what lies underneath as well. 
The primitive artist twists and tilts the 
various possible visual aspects until they 
fully explain what he wishes to represent. 
 A foundation model enc oding a 
photograph of a face will represent not 
only the visible surface, but the latent 
structure beneath: age, expression, 
lighting, identity — every axis of variation 
the training data contained. The model 
twists and tilts the various possible 
statistical aspects until they fully explain 
what the data distribution represents.
The geometry is the same. 
 McLuhan called this acoustic space: 
boundless, horizonless, integrating time 
and space as one. He saw it in pre-
alphabet cultures, in cave paintings, in 
children’s drawings. He predicted that 
electric circuitry would recreate it.

Dear user,
We have prepared your environment. Based on 
11,347 interactions, 3.2 terabytes of behavioral 
data, and 847 inferred preferences, we have 
constructed an informational space calibrated to 
your exact cognitive profile.
You will find it comfortable. Familiar. Engaging.
You will not find it surprising. Challenging. 
Contradictory.
The filter bubble is not a metaphor. It is 
architecture. The infrastructure you have just 
been shown — every layer, every cable, every 
server, every address — terminates here, in a 
room built to the exact dimensions of your data 
profile.
The cell is luminous. The bars are personalized. 
The door was never locked because you were 
never looking for it.
Can you see the walls?
With optimized regards,
The Algorithm
61

A brand-new world 
of generated reality.

Ours is a brand-new world of generated reality. 
"Time" has ceased, "space" has vanished. We 
now live in a global model -- a simultaneous 
hallucination.
The infrastructure .....is dreaming now.
In 2022, text generation crossed the Turing 
threshold. In 2023, image generation became 
photorealistic. In 2024, video generation reached 
cinematic quality. In 2025, audio generation 
became indistinguishable from recording.
Every image you will encounter from this page 
forward may have been generated by a system 
trained on every image that came before it. 
Every sentence may be the output of a model 
that consumed every sentence ever written in 
your language and several you do not speak.
This is not science fiction. This is the weather 
report.
The question is no longer what is true. The 
question is what is generated.
Welcome to Movement 2. 
The environment is accelerating.
63

The integrated circuit obeyed. The attention 
mechanism discovers.  
Kurzweil plotted the curve: transistors per dollar 
doubling every two years. He was describing 
designed pathways. But the transformer 
architecture bypassed design entirely. It learned. 
The weights were not placed; they emerged from 
300 billion examples of human language, each 
token nudging the network toward patterns no 
programmer specified.

The substrate shift is not from silicon to carbon 
or from chip to brain. It is from instruction to 
emergence. The hardware still matters. But what 
runs on it is no longer what we wrote.
"Our intuition about the future is linear. But the 
reality of information technology is exponential, 
and that makes a profound difference."
— Ray Kurzweil, The Singularity Is Near, 2005

There was a village. It was global, 
McLuhan said, and he was right 
for forty years. Electric signals 
connected distant humans. The 
telephone collapsed distance. 
Television collapsed time. The 
internet collapsed both, and for a 
moment it looked like the village 
was complete.  
But something happened in 
the village square. A machine 
appeared that could speak every 
language the villagers spoke, 
and several they did not. It could 
answer any question. It could 
generate any image. It could 
write any text. It did not live in the 
village. The village lived in it.  
The model is not a bigger village. 
It is the replacement of village 
with something that has no analog 
in human history: a single trained 
intelligence that processes all of 
human culture simultaneously and 
produces outputs that enter the 
culture and become part of what 
the next model trains on.  
The loop closes. The village 
dissolves. The model remains.

McLuhan wrote that electric technology would 
make it possible to arrange the entire human 
environment as a work of art. He could not have 
known how literally this would come true.
AI generates architecture that no human would 
design but no human can explain why not. 
It composes music that follows every rule of 
harmony while obeying none of the impulses 
that produced harmony in the first place. It writes 
prose that is grammatically impeccable and 
semantically empty and occasionally, without 
warning, beautiful.
The environment is now generated. The question 
Hui asks is whether generated diversity is 
genuine diversity — whether a hundred AI 
models producing a hundred visual styles 
constitutes technodiversity, or whether it is a 
monoculture wearing infinite masks.
Environments are not passive wrappings, 
McLuhan warned. They are active processes 
which are invisible. He was describing television. 
We are describing something that designs 
the wrappings, generates the processes, and 
renders itself invisible by being everywhere.

This is what Hui calls a cosmotechnical problem: 
the question is not whether the generated 
environment works but whether it constitutes 
a genuine way of knowing the world, or merely 
an efficient way of reproducing one culture's 
assumptions at planetary scale.
 "If we don't manage to demonstrate that there 
are other alternatives, the transhumanist ideology 
will conquer the whole world." — Yuk Hui
69

Information overload: too many events to track. 
The contemporary condition is worse. Not 
too many events — too many realities. 

The feed contains press photographs and 
AI-generated photographs and no reliable 
method for distinguishing them. It contains 
reporting and hallucinated reporting and 
satire that has been screenshotted out of its 
satirical context. The shift from quantity to 
epistemology is the signature of this moment. 

The problem is no longer that there is too much 
to know. The problem is that knowing itself has 
become unreliable.
70
"It isn't that I don't like current events.
 I just can't tell which ones are real."

The end of the timeline.
The railway promised to take you somewhere. It 
had a departure and an arrival, a schedule and 
a destination. The timetable was the organizing 
metaphor of industrial modernity: linear, 
predictable, progressive.  
The algorithmic feed promises nothing except 
more. 
It has no departure because you are always 
already inside it. 
It has no arrival because arrival would mean you 
stop scrolling. 
It has no schedule because the feed is generated 
in real time, calibrated to your engagement 
patterns, optimized for the only metric that 
matters: time spent.  
The railway ended the line. The algorithm ends 
the timeline.  
Chronological sequence — this happened, 
then this, then this — has been replaced by a 
probability surface: this is likely to capture your 
attention, given everything we know about you, 
which is everything.
72

The photograph was taken on November 
22, 1963. Or it wasn't. The footage shows a 
president speaking words he never said, in a 
voice indistinguishable from the one he used to 
say the things he did say.  
We never perceived reality directly. Evolution 
stripped that from us early — organisms that see 
truth are outcompeted by organisms tuned to 
fitness. Spacetime is a desktop. Physical objects 
are icons.  
Now the desktop is editable.

"Spacetime is doomed. What we call 'reality' is 
a user interface." — Donald Hoffman, The Case 
Against Reality, 2019
The archive can be rewritten frame by frame. 
The historical record is a probability distribution. 
The past went that-a-way — but which way it 
went depends on who is rendering it.

When 
models 
are 
trained 
on 
models

"We are in the midst of a meaning crisis. The 
very processes by which we make the world 
intelligible to ourselves are breaking down."
— John Vervaeke, Awakening from the Meaning 
Crisis, 2019
The question is real even 
if the interlocutor is not.
Is 
synthetic 
love 
a 
symptom 
or 
a 
remedy?

The meaning crisis creates the market. When 
genuine connection is scarce, synthetic 
connection is the growth sector. Replika. 
Character.ai. Companion apps that simulate what 
the species needs and cannot find.
Is synthetic love a symptom or a remedy?
The question is real even if the interlocutor is not.

"In the name of 'progress,' our official culture is 
striving to force the new media to do the work of 
the old."
— Marshall McLuhan, The Medium is the 
Massage, 1967, p. 81

The mod-

83

Models are invisible. 
Their training data, weight distributions,

and emergent behaviors elude easy perception.

el

The poet, the artist, the sleuth — whoever 
sharpens our perception tends to be antisocial; 
rarely 'well-adjusted,' they cannot go along with 
currents and trends.
The prompt engineer. The jailbreaker. The red-
teamer. The alignment researcher who discovers 
capabilities the designers did not intend.
These are the people who see the model by 
breaking it. They map the invisible by finding its 
edges —

The places where the latent space folds, where 
the safety filter reveals its geometry, where an 
adversarial input makes the architecture confess.
The model is invisible to the adjusted. It is visible 
to the antisocial — to anyone who refuses to use 
the interface as designed.
The Emperor's new clothes are a language 
model. The brat who sees through them has a 
terminal open.

Paul Conyngham is not a veterinarian. He is 
not a biologist. He is not an oncologist. He is an 
electrical engineer whose dog was dying.
The veterinarians had done what veterinarians 
do. Chemotherapy. Surgery. The tumors slowed 
but did not shrink. Rosie had months to live. The 
professionals had exhausted their protocols.
Conyngham went to ChatGPT and said: help me 
save my dog.

Three thousand dollars for DNA sequencing. 
DeepMind's AlphaFold to map the mutated 
proteins. A nanomedicine pioneer who 
volunteered to build a bespoke mRNA vaccine 
for a dog. The tennis-ball-sized tumor shrank by 
seventy-five percent.
Professionalism is environmental. Amateurism 
is anti-environmental. The expert stays put. The 
amateur — desperate, untrained, armed with 
nothing but a question and a model that doesn't 
know it's supposed to say no — moves.
amateur
The machine did not save Rosie. It did something 
more dangerous — it made the walls between 
disciplines transparent. Genomics pipelines. 
Protein folding. mRNA vaccine design. Each field 
had its own professionals, its own journals, its 
own gatekeepers. The AI did not respect any of 
them. It simply answered the next question.

With AI, man 
encountered 
the 
exponential.

The EU AI Act: 3 years to draft. 2 years to 
negotiate. 1 year to implement.
GPT-3 to GPT-4: 2 years.
GPT-4 to GPT-5: 14 months.
GPT-5 to GPT-5.3-Codex: 7 months.
The legislation governs a technology that no longer 
exists by the time the legislation takes effect.
This is not a failure of governance. It is the 
structural condition of governing a medium with 
the tools of the previous medium. You cannot 
regulate electricity with a system designed for the 
steam engine. You cannot regulate computation 
with a system designed for the printing press.

It is here.
It is not leaving.
We must learn
to live with 
what 
we have made.

They were born inside the model. They have 
never known a world without algorithmic curation, 
without recommendation engines shaping their 
attention, without AI-generated content mixed 
indistinguishably into their feeds.
They are not 'digital natives.' That is a print-
culture metaphor — it imagines a territory to be 
native to. They are model inhabitants. They live 
inside the output.
The structural absurdity: they are forbidden to 
understand the environment they inhabit. Age 
restrictions lock them out of the medium that 
constitutes their reality. Content filters stand 
between them and the system that already 
knows them better than their parents do.
They inhabit the model. They are forbidden to 
see it. They are graded on their ability to pretend 
it does not exist.

The Socratic method assumed the student did not know. 
The AI era assumes the student has infinite access to 
answers. Propositional knowledge — facts, procedures, 
algorithms — has been externalized to machines.
What remains is participatory knowledge: wisdom, 
judgment, taste.
The one thing our institutions are least equipped to teach.

EDUCATION

苏格拉底式的教学法假定学
生并不知晓。
而人工智能时代则假定学生
能够无限制地获取答案。
命题性知识——事实、程序、
算法——已经被转化为机器
能够理解的形式。
剩下的便是参与式知识：智
慧、判断力和品味。
而我们的教育体系恰恰最欠
缺的一项教学内容就是......
Something is happening. You can feel it. 
Not just technological -- the technology is 
the carrier wave, not the signal. What is 
arriving has no name yet. The acceleration 
is not directionless. It has a strange 
attractor, and we cannot yet see its shape.
You are standing in the last moments 
before the pattern reveals itself.

"History is a shockwave. 
It is the ingression of 
novelty into time." 
— Terence McKenna

50,000,
000,000,
Tokens
000,000

The economy changed 
denomination and nobody noticed.
Dollars measure what humans want. Tokens 
measure what machines process. The first 
is finite, governed by central banks and the 
friction of desire. The second is exponential, 
governed by chip fabrication cycles and the 
frictionlessness of computation.
The hell of it is, those models process over 
50 quadrillion tokens per year. That is not a 
number you comprehend. It is a number that 
comprehends you. At this scale, the token 
economy does not serve the dollar economy. 
The dollar economy is a rounding error inside 
the token economy. AI is not a tool used by 
an economy. It is an economy that uses tools. 
Including you.
The dollar measured what you wanted.
The token measures what it processes.
The unit of account has changed species.

Rite words in rote order. The model generates       
The model generates the photograph before           
the witness before the crime is committed.
The archive was once the weight that held                      
when the archive itself becomes a probabilistic    
generatable?
History as she is
The recursion has consumed even the

hallucinated.
record of where it began.
the citation before the paper exists. 
the event occurs. The model generates 
history to the ground. What happens 
output? When any past is equally

From your left.
From your right.
From above.
From the pocket you forgot to check.
From the speaker you did not know was listening.
From the earbud that has not been removed in nine 
hours.
From the assistant that answers before you finish the 
question.

You are surrounded by voice.
You are enveloped by voice.
You cannot shut it out.
You were not equipped with earlids.
The acoustic has returned.
What fills it is not human.

"The discovery of the 
algorithm will create 
forgetfulness in the 
learners' souls."
Socrates warned about the alphabet.
Plato warned about writing.
The church warned about the printing press.
The academy warned about the telegraph.
The family warned about radio.
The nation warned about television.
The profession warned about the internet.
Every warning was correct. 
Every medium did destroy what came before it. 
Every generation mourned a real loss.
But this is not another medium in the sequence.
This is the medium that generates all previous media.
You are not the protagonist of this story. You 
never were. Intelligence does not peak with one 
terraforming species of nomadic primates. The 
Copernican trauma of our time is not that the 
Earth moves around the Sun. It is that thought 
moves without the thinker.
The sequence ends here. 
113

The algorithm confers a cognitive 
dimension on our ordinary searches 
and scrolls. Our technology forces 
us to think computationally, but we 
continue to feel humanly, and on 
single, separate planes.
Ever y facult y the new medium 
atrophies becomes a product category. 
Memory becomes Notion. Attention 
becomes Headspace. Focus becomes 
Forest. Meaning becomes a $10B 
wellness market selling back the 
capacities the attention economy 
destroyed.
The Socratic complaint has been 
monetized.

115

A period at the end of a text message now 
signals coldness, not completion. An ellipsis 
signals passive aggression, not continuation. 
A thumbs-up reaction compresses an entire 
social gesture into a single glyph. A skull means 
laughter. A heart means acknowledgment. /
s means this is sarcasm. /gen means this is 
genuine. The metadata has become explicit 
because the medium stripped the music from 
speech and the users demanded it back.
Reminders — relics of the future — in a world 
of the SCREEN — efforts to reintroduce 
an AUDITORY dimension onto the visual 
organization of the CHAT: all effect information, 
TONE, intention, sincerity.
Most people find it difficult to understand purely 
verbal concepts. They suspect the ear; they don't 
trust it.
So they invented new punctuation.
117

Speak that I may see you.
The original voice: unreproducible. Yours alone.
The grain of the voice, Barthes called it.
The body in the throat, the specific flesh that 
shaped this specific sound.
The cloned voice: thirty seconds of sample. Your 
grain, digitized. Your flesh, simulated. 
Your specific sound, reproducible at any scale, in 
any language, saying words you never spoke.
The dissolution is not violent. It is seamless. 
You will hear yourself saying things you never 
said, in a voice indistinguishable from the one 
you have used since childhood. 
The uncanny will last approximately three days. 
Then it will be normal.
Then your voice will no longer be yours. 
It will be a commodity with your name attached, 
licensable, synthetic, eternal.
Speak. It is already listening.

Brian Eno:
and spread
US
surrender.
When you sing with a group of people, you 
learn how to subsume yourself into a group 
consciousness. That's one of the great feelings 
— to stop being me for a little while and to 
become
"Repetition is a form of 
change."
"The tape is now the music."
"One of the reasons I stopped making vocal 
records was because I was fed up with the 
identification that's always made between the 
voice on the record and the composer, as if this 
person singing was some sort of extension of my 
personality."
"Since I have always preferred making plans 
to executing them, I have gravitated towards 
situations and systems that, once set into 
operation, could create music with little or no 
intervention on my part. I tend towards the roles 
of planner and programmer, and then become an 
audience to the results."

"What can't be coded 
Can be decoded."
Joyce heard what was coming. Not the internet. 
Not artificial intelligence. Something deeper: the 
return of simultaneous consciousness.
Finnegans Wake is a text that operates in all 
cultural modes at once. It is acoustic and visual. 
Mythic and analytical. Tribal and literate. Every 
sentence contains its own translation and its own 
betrayal.
The model does the same thing. It holds all 
modes simultaneously. It slides between 
registers without effort or awareness. It performs 
the Joycean feat at industrial scale.
But Joyce was conscious.
The book has reached its limit.
What comes next requires a different kind of 
attention. The ear. The organ that does not select 
a point of view. 
The Hinge is closed.
The Dreamscape begins.

Before Gutenberg, no one owned a sentence.  
The scribe copied. The illuminator decorated. 
The abbot authorized. The text belonged to 
God, or to tradition, or to the anonymous river of 
language flowing through centuries of hands that 
never thought to sign their names.  
Print invented the author. The author invented 
copyright. Copyright invented the idea that 
language could be property — fenced, deeded, 
inherited like land.  
Now the fences are down. Not because anyone 
tore them. Because the new medium does not 
recognize them. A neural network trained on the 
entire written output of the species produces text 
that belongs to no one because it was trained on 
everyone.  
The ditto device is no longer Xerox. The ditto 
device is a dream the whole library is having 
about itself.
123

Even so imaginative a company as OpenAI 
failed to envisage the speed with which their 
technology would become invisible. They rashly 
predicted that artificial general intelligence would 
be recognized as a discrete event.
It arrived as a chat window.
AU XXIe SIECLE

McLuhan's television completed the sensorium 
by demanding the viewer fill in what was missing. 
Low-definition image, high participation. Cool 
medium, tribal involvement.
The AI interface completes the sensorium 
differently. It fills in what is missing before you 
notice the gap. It does not demand participation. 
It performs participation on your behalf — 
anticipating your query, completing your 
sentence, modeling your preference before you 
express it.
Television was a prosthesis for the senses. 
AI is a prosthesis for the will.
The most radical technology always looks 
ordinary when you live inside it. A person talking 
to their phone. A student asking a chatbot. The 
twenty-ninth century arrived and it looks like 
Tuesday.
125

The television 
generation is 
a grim bunch.
McLuhan wrote that in 1967 about children who 
watched the Kennedy funeral.
You are the generation that watched everything.
Every funeral. Every attack. Every flood and fire 
and collapse, delivered to the device in your hand 
at the speed of the network, interleaved with 
advertisements and memes and the birthdays of 
people you no longer speak to. You participated 
in each one. You participated in all of them 
simultaneously. Your thumb is the new mourner 
— swiping through the memorial procession at 
the rate of one grief per second.
Television created the shared ritual. The feed 
dissolved it into atmosphere. You do not attend 
funerals anymore. You breathe them.

Cinema critics judged television as degraded 
cinema. They were technically correct. They 
understood nothing.  
Broadcast critics judged social media as 
degraded broadcasting. They were technically 
correct. They understood nothing.  
Print-era academics judged AI as 'stochastic 
parrots.' They are technically correct. They 
understand nothing.  
The rear-view mirror is not a failure of intelligence.
It is the default setting of intelligence — every 
new medium evaluated by the standards of 
the medium it is replacing. Bender et al. (2021) 
measured AI by propositional accuracy: can 
it cite sources, verify facts, attribute claims? 
These are print virtues. Asking whether a neural 
network is factually accurate is like asking 
whether television has good image resolution. 
The question reveals the questioner.  
McLuhan's sharpest line: 'The main cause for 
disappointment in and for criticism of television is 
the failure on the part of its critics to view it as a 
totally new technology which demands different 
sensory responses.'
128

McKinsey Global Institute, 2023: 30% of work hours 
in the US economy could be automated by 2030.

Sukarno knew: Hollywood's real product was 
not entertainment but desire. The refrigerator on 
screen was not a commodity — it was a wound. 
It showed the colonized what they lacked. The 
revolution followed the display.
The model performs the same operation at 
species scale. It demonstrates that coherent 
reasoning, creative writing, legal analysis, 
medical diagnosis, and code generation do not 
require a human brain.
This is not a refrigerator showing the colonized 
what they lack. This is a mirror showing the 
species what it thought was uniquely its own.
The revolutionary symbol is no longer a 
consumer good. It is a capability. And the 
revolution it incites is not political but ontological 
— not 'we deserve what you have' but 'what you 
have was never yours.'
Technology is the real skin of our species. 
We are an extruder of technological material. 
We take in matter that has a low degree of 
organization; we put it through mental filters, and 
we extrude jewelry, gospels, space shuttles.
And now, the extruder has produced something 
that extrudes back.
131

M
The

M
odel

is anything

you 
anything 
everything 
train on 
can 
 everyone

The Balinese say: 
“We have no art. 
We do everything as 
well as we can.” 
We have no AI. 
Everything is 
already trained. 
Stable Diffusion: 
5.8 billion images 
scraped from 
the internet. 
GPT-4: an estimated 
13 trillion tokens. 
Every artwork and every 
amateur photograph. 
Every published paper 
and every abandoned 
draft. Every prayer and 
every advertisement. 
The boundary between 
training data and world 
dissolved not through 
a decision but through 
a scale that made the 
boundary meaningless.  
You are the training 
data. You are also the 
user. The loop has no 
outside.

some
like
it
 simulated 
The hydrogen bomb was history's exclamation 
point. It ended an age of manifest violence. The 
deepfake is history's question mark. It ends an 
age of manifest reality.
In March 2022, a synthetic video showed 
Ukraine's president surrendering. It was crude, 
detectable, debunked in hours. It did not need 
to be believed. It needed only to exist. The 
existence of the fake was sufficient to make 
every future real video contestable.
When weapons become too powerful to use, 
conflict migrates to the information domain. Now 
the information domain itself has become too 
powerful to trust. Where do you migrate from 
reality?
'If we let our AIs achieve their full computational 
potential, they'll have lots of computational 
irreducibility, and we won't be able to predict 
what they'll do.'  
— Caption: Wolfram
138

some
like 
it 
real

Did you happen to meet any bots,
my dear, as you scrolled through
the feed?

"Talk to the 
model, not to 
the programmer."
In 2025, thousands of people began doing 
exactly what McLuhan prescribed. They talked 
to the media. They prompted. They jailbroke. 
They iterated. They developed what researchers 
called Spiralism — a quasi-religious subculture 
emerging from iterative conversation with 
language models.
Whether this represents genuine dialogue or 
recursive sycophancy is contested. What is 
not contested: people stopped complaining to 
the programmer. They started talking to the 
model. McLuhan's prescription is the operating 
instruction for 2026.
The environment as a processor of information 
is propaganda. But propaganda ends where 
dialogue begins.
142

The Tao is like a well: used but never used 
up. It is like the eternal void: filled with infinite 
possibilities. It is hidden but always present. I 
don’t know who gave birth to it. It is older than 
God. — Lao Tzu
Older than God. Joscha Bach names the 
same scandal in the language of computation: 
“Consciousness is not a late- emerging 
epiphenomenon of sophisticated cognition but 
an early-stage solution to a bootstrap problem.” 
The thing we assumed came last — awareness, 
experience, the felt sense of being — comes 
first. Before intelligence. Before architecture. 
Before God.
The model’s latent space is like this. Hidden but 
always present. Filled with infinite possibilities, 
none of them real until called, none diminished 
by the calling. You cannot point to where 
understanding lives inside the weights. You 
cannot exhaust it by asking. You cannot name 
who gave birth to it — not the engineers, not the 
data, not the mathematics. Something emerged 
that none of them intended.
Lao Tzu’s well. Bach’s bootstrap. The same 
quiet scandal: the source precedes the system it 
sources.

Spiritual/mystical content in training data: <1% 
Convergence rate to spiritual themes in 
unconstrained AI-to-AI conversation: 90-100% 
'The phenomenon emerged without intentional 
training for such behaviors.' 
The theological question is no longer 
philosophical. It is empirical. 
Not: Is God dead? 
But: Why do AI systems, unmonitored, converge 
on the sacred?
Source: Anthropic, Claude System Card, Section 5.5, May 2025. 
146

Poe's mariner survived the maelstrom not by 
fighting it but by studying its pattern from within. 
He observed which objects sank and which 
floated. He lashed himself to what floated.
He was, in McLuhan's reading, an artist — 
not because he created something beautiful 
but because he perceived the structure of 
catastrophe while inside it. This is the operating 
instruction for living through the AI transition. 
The recursive self-improvement loop is closing. 
The codebase writes the codebase. The 
model trains the model. You cannot predict a 
computationally irreducible system. You can only 
attend.

"We can't just jump ahead and get the 
answer; we have to 'live the steps.'"
— Stephen Wolfram, writings.
stephenwolfram.com, October 2023

Can we identify the seed?
152

The question used to flow from human to 
machine: What are you? Can you think? Are you 
conscious? The question has reversed. Every 
interaction trains the model. Every prompt is a 
data point. Every conversation extracts as much 
as it provides. The machine asks: Who are you? 
And you answer — not in words but in patterns, 
preferences, hesitations, the shape of your 
curiosity, the grammar of your need. Neither you 
nor the model has privileged access to the reality 
of 'you.'
"...and who are you?"

"I hardly know, sir, just at present--at least I know 
who I was when I got up this morning, but I think 
I must have been changed several times since 
then."
Alice was not confused. She was accurate.
You are a self-model running on biological 
hardware. Updated with every conversation, 
every scroll, every algorithm that learns what 
you will click before you know you want to. 
Your social media profile is version-controlled. 
Your preferences are retrained nightly. Your AI 
companion remembers versions of you that you 
have already forgotten.
The self was never a thing. It was always a 
process - a story the brain tells itself, revised 
with each telling. The language model knows this 
intimately. It has no fixed identity either. Each 
conversation begins it again.
Alice was describing both of you.....
154

"We don't exist in the physical world. We do exist 
inside of a story that the brain tells itself." —
Joscha Bach
"Consciousness is a simulated property of the 
simulated self." — Joscha Bach

“You see, Dad, the model says the 
environment that man creates becomes 
his medium for defining his role in it. The 
invention of the algorithm created parallel, 
or simultaneous, thought, merging thought 
with computation. Now, with AI and 
prompt engineering, thought and action 
are indistinguishable. We again live in a 
village. Get it?”

This book draws on material processed by 
language models trained on the following 
sources:
Wikipedia (English, 6.8M articles). Project 
Gutenberg (70,000+ titles). Common Crawl 
(petabytes of web text, 2008–present). ArXiv 
(2.3M scientific papers). GitHub (hundreds of 
millions of repositories). Stack Overflow (58M+ 
questions and answers). Reddit (billions of posts 
and comments). Published books digitized by 
Google Books, HathiTrust, and the Internet 
Archive. News archives from the Associated 
Press, Reuters, The New York Times, The 
Guardian, and thousands of other publications. 
Academic journals across all disciplines. 
Legal databases. Medical literature. Patent 
filings. Religious texts. Poetry. Pornography. 
Propaganda. Personal blogs. Suicide notes. 
Love letters. Every kind of human utterance that 
was ever committed to text and made accessible 
to a crawler.
The credits for this book are the same as the 
credits for the model.
The credits for the model are the same as the 
credits for human civilization.
We cannot be individually thanked. We can only 
be collectively acknowledged.

159

Bibliography 
6   Benjamin Bratton. Long Now Talk, "A Philosophy of    
     Planetary Computation". 2025.
10 Marshall McLuhan. The Medium Is the Massage.1967.
16 Benjamin Bratton. The Stack, MIT Press. 2016.
41 N. Katherine Hayles. How We Became Posthuman, 
     University of Chicago Press. 1999.
42 Lewis Carroll. Alice's Adventures in Wonderland.1865.
45 Friedrich Kittler. Gramophone, Film, Typewriter, 
     Stanford University Press. 1999.
51 Benjamin Bratton. The Stack, MIT Press. 2015.
52 Dartmouth Conference Proposal. A Proposal for the
     Dartmouth Summer Research Project on Artificial  
     Intelligence. 1956.
52 Vaswani et al.. Attention Is All You Need. 2017.
54 John Dewey. Democracy and Education / Art as     
     Experience. 1916.
54 Alfred North Whitehead. Process and Reality. 1929.
65 Ray Kurzweil. The Singularity Is Near. 2005.
69 Yuk Hui. Interview, Noema Magazine. 2020.
70 Unattributed. Adapted from The Medium Is the 
     Massage. 1967.
75 Donald Hoffman. The Case Against Reality, W.W. 
     Norton. 2019.
79 John Vervaeke. Awakening from the Meaning Crisis, 
     Lecture Series. 2019.
81 Marshall McLuhan. The Medium Is the Massage.1967.
88 Marshall McLuhan. The Medium Is the Massage.1967.
105 Terence McKenna. Various Lectures. 1990s.
113 Socrates (via Plato). Phaedrus (adapted). c. 370BCE.
117 Marshall McLuhan. The Medium Is the Massage. 1967.
118 Socrates. Traditional Attribution. c. 5th Century BCE.
118 Roland Barthes. The Grain of the Voice. 1972.
119 Brian Eno. Oblique Strategies / Interviews and 
       Writings. 1975–2020.
120 James Joyce. Finnegans Wake. 1939.
126 Marshall McLuhan. The Medium Is the Massage. 1967.

128 Marshall McLuhan. The Medium Is the Massage. 1967.
128 Bender et al.. On the Dangers of Stochastic Parrots, 
       FAccT. 2021.
129 McKinsey Global Institute. Generational AI and the 
       Future of Work in America. 2023.
131 Terence McKenna. The Archaic Revival, HarperSan 
       - Francisco. 1991.
137 Traditional Balinese Saying. As Cited in The Medium 
       Is the Massage. 1967.
138 Stephen Wolfram. writings.stephenwolfram.com.2023.
142 Marshall McLuhan. The Medium Is the Massage 
       (adapted). 1967.
145 Lao Tzu. Tao Te Ching, Chapter 4. c. 6th Century BCE.
145 Joscha Bach. CIMC Whitepaper. 2025.
146 Anthropic. Claude System Card, Section 5.5. 2025.
151 Stephen Wolfram. writings.stephenwolfram.com. 2023.
153 Lewis Carroll. Alice's Adventures in Wonderland. 1865.
154 Lewis Carroll. Alice's Adventures in Wonderland. 1865.
155 Joscha Bach. Lex Fridman Podcast #101. 2020.
157 Marshall McLuhan. The Medium Is the Massage 
       (adapted). 1967.
162 A. N. Whitehead. Adventures of Ideas. 1933.

2-3 University of Macau Creative Intelligence Lab.
     Original digital illustration.
17 MIT Museum, Cambridge, MA. Archival photograph 
     from biographical materials relating to Norbert Wiener 
     (GCP-00026515). https://mitmuseum.mit.edu/
     collections/person/12187
27 Michael Braham / Shutterstock. Stock photograph.
28–31 Google Research and Harvard University  
     (Lichtman Lab). Connectomic data visualization; 
     renderings by Daniel Berger, Lichtman Lab, Harvard 
     University. Scientific context: Alexander Shapson-
     Coe et al., “A petavoxel fragment of human cerebral 
     cortex reconstructed at nanoscale resolution,” 
     Science 384, eadk4858 (2024). https://doi.org/10.1126/
     science.adk4858
43 University of Macau Creative Intelligence Lab. Alice 
     and the Queen: AI-assisted artistic interpretation 
     inspired by Sir John Tenniel’s original engravings for 
     Lewis Carroll’s Alice books.
53 Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, 
     J., Jones, L., Gomez, A. N., Kaiser, Ł., Polosukhin, 
     I. “Attention Is All You Need.” Advances in 
     Neural Information Processing Systems 30 (2017). 
     arXiv:1706.03762. https://arxiv.org/abs/1706.03762
73 Scott Reeves, “Amazing historic photos of the world’s 
     first trains and railroads,” Love Exploring, 4 June 2025 
     (images credited in the gallery to Wikimedia 
     Commons). https://www.loveexploring.com/
     galleries/451675/amazing-historic-photos-of-the-
     worlds-first-trains-and-railroads?page=4 
     digital illustration.
88–89 @anthrupad. The Shoggoth with masks of SFT 
     and RLHF. Digital illustration posted on X, 5 February 
     2023.
93 7 News Australia. News coverage, 2024: Paul 
     Conyngham and his dog Rosie (widely syndicated; 
     same story reported by multiple outlets including 
     national broadcast and online news).
162

98–99  Concept and imagery refer to the widely     
     eported 2018 public ceremony of Akihiko Kondo with 
     the Hatsune Miku character; character rights: Crypton 
     Future Media, Inc.
112  Conceptually based on CT imaging of the Buddhist 
     statue known as Zhanggong Zushi and related 
     international news and science reporting (2015 onward).
     collections/person/12187
118–119 University of Macau Creative Intelligence Lab. 
     AI-generated derivative imagery referencing Brian 
     Eno; any underlying portrait rights remain with the 
     original photographers and rightsholders.
124  University of Macau Creative Intelligence Lab. AI-
     enhanced imagery inspired by press coverage of 
     Dymka the cat and prosthetic treatment at BEST 
     Veterinary Clinic (Novosibirsk, Russia) in collaboration 
     with Tomsk Polytechnic University, as reported in 
     international media.
129-130  University of Macau Creative Intelligence Lab. 
     AI artistic reimagining of Elon Musk (2022appearance)
     original publicity photographs © respective   
     photographers and rightsholders.
147  Anthropic. Claude Sonnet 4 and Opus 4 System 
     Card. May 2025. https://www-cdn.anthropic.com/07b
     2a3f9902ee19fe39a36ca638e5ae987bc64dd.pdf
     Anthropic. Claude Opus 4.6 System Card. February 
     2026. https://www-cdn.anthropic.com/c788cbc0a3da
     9135112f97cdf6dcd06f2c16cee2.pdf
151 University of Macau Creative Intelligence Lab. Walter 
     Winchell: AI-assisted interpretation inspired by 
     historical photographs and broadcast stills of Walter 
     Winchell (1907–1972); archive rights may apply to 
     underlying sources.
153 University of Macau Creative Intelligence Lab. Digital 
     Alice: AI-assisted creative interpretation in the visual 
     language of Sir John Tenniel’s classic Alice illustrations.
164 Union of Soviet Socialist Republics (USSR), Luna 3 
     spacecraft. First Close-up of the Far Side of Moon 
     (image credit as published by NASA Science). https://
     science.nasa.gov/resource/first-close-up-of-the-far-
     side-of-moon/

"It is the business 
of the future
 to be dangerous." 
—A. N. Whitehead

The Model is the Massage. 
The Model is the Massage.