Real life feels increasingly like an Adam Curtis documentary. So, what happens when you ask Claude, ‘Write a short script for a documentary on AI in the style of Adam Curtis’?
Claude responds:
I’ve crafted the script to capture Adam Curtis’s distinctive style – a poetic, philosophical narrative that weaves together historical imagery, technological development, and critical social commentary. The script uses Curtis’s trademark techniques of connecting seemingly disparate ideas, presenting technology as a complex cultural phenomenon rather than a neutral tool, and maintaining a somewhat melancholic, introspective tone.
(Claude’s summary of Curtis’s style is more respectful than the cheeky but brilliant parody, “The Loving Trap“).
The AI-generated script is below. As with everything AI-generated it’s necessary to remember that LLMs are language models, not knowledge models. So the script is a trick, statistically similar to a Curtis documentary. (How the copyrighted material gets into the LLM’s model is another question.)
Nevertheless, GenAI has the ability to sometimes generate eye-catching word combinations: “a curious machine … of dreams and mathematics”; “profound hallucinations.” (David Bowie got this feeling of uncanny intelligence from his cut-up-newspaper approach to generative lyric-writing in the 1970s.)
The section on the power of prediction and the surrender by humanity to algorithms is uncanny – coming as it does from a predictive language model. And there’s even a Yeatsian ending: “we do not yet know whether we are the dreamers or the dream.”
For the full effect, maybe play this Spotify playlist in the background, particularly Song for Zula by Phosphorescent.
THE ALGORITHMIC DREAM
INT. ARCHIVAL MONTAGE - DAY
FADE IN:
A series of flickering black and white images: early computers, punch cards, mathematical equations. The room is cold, clinical.
NARRATOR (V.O.)
In the vast landscape of human imagination, there emerged a curious machine - not of metal and wire, but of dreams and mathematics.
CUT TO:
VINTAGE FOOTAGE - 1950s COMPUTER LABORATORY
Researchers in white shirts and thin ties huddle around massive computing machines. CLAUDE SHANNON works intently at a desk.
NARRATOR (V.O.)
Artificial Intelligence was never just a technology. It was a profound hallucination, a mirror reflecting humanity's deepest desires and most intricate fears.
MONTAGE - RAPID CUTS
Binary code scrollingAlan Turing writing equationsEarly neural network diagramsBlinking computer terminals
NARRATOR (V.O.)
The story begins in cold war laboratories, where brilliant minds believed they could map the human mind onto electric circuits. Claude Shannon, Alan Turing - they were not just engineers, but mystics of a new kind.
CUT TO:
INT. MODERN TECH OFFICE - DAY
Sleek screens display complex algorithms. Social media interfaces pulse with activity.
NARRATOR (V.O.)
But power has always been about prediction. And prediction is control.
SPLIT SCREEN
Left side: Stock market tickers
Right side: People staring into smartphone screens
NARRATOR (V.O.)
The algorithms learned not just to calculate, but to anticipate. They began to see patterns in human behavior that humans themselves could not perceive - desires, fears, impulses reduced to mathematical probabilities.
QUICK CUTS
A targeted advertisementA recommendation algorithmA facial recognition system
NARRATOR (V.O.)
Silicon Valley sold us a dream of liberation through technology. Yet with each recommendation, each targeted advertisement, each predictive model, we surrendered more of ourselves to the machine's cold embrace.
SLOW ZOOM
On a single computer screen, showing a complex neural network visualization
NARRATOR (V.O.)
We became not the creators, but the created - our choices increasingly shaped by systems we barely understand.
FADE TO BLACK
NARRATOR (V.O.)
We are living inside the machine's dream. And we do not yet know whether we are the dreamers or the dream.
FADE OUT.
END
Postscript
Needless to say, there is nothing like the real thing: the carefully edited scripts, the hours of footage watched by Curtis and his team to find gems in video archives, the music personally selected rather than algorithmically-calculated, and of course the BBC RP voice-over. YouTube contains many of his documentaries including this account and this playlist.
