I have struggled to unify my thoughts about the latest developments in Generative AI. There is however one image that I keep coming back to that might encapsulate all that is happening.
This is of course not an AI generated image. It is a 1920, DADA era sculpture by Raoul Hausmann, entitled The Spirit of Our Time. In spite of that primitive, hodgepodge, DADA style, it captures an unease that I expect remains relatable today. It is that of a spiritual man mechanized, reduced to measurements, dimensions, and constituent parts.
Already in 1920, was realized the potent dream (or nightmare) of the coming era: the merging of the human with the machine. Man once measured and fully categorized, can be made into mechanism, and therefore made controllable, predictable, and finally immortal. But this hybridity creates a problem of ambiguity. In the transitional period the line between human and machine becomes blurred, and with it the line between fact and fiction.
Fear of acute deception drives much of the conversation in present day A.I. Safety circles. How can generative AI be used without giving power to manipulators?
It is a problem, certainly. But they miss the more fundamental erosion of the bedrock of perception that is occurring. Now all media is suspect — not merely as before, at the narrative level — now the most fundamental aspects of visual perception can mislead us.
Foot Washing, the 2024 Super Bowl ad, the photographic work of Julia Fullerton-Batten, was mistaken by many as Generative AI art. Photography will be forced to change itself in order to distinguish itself from AI images, or face the mistrust of its audience. All other visual mediums will follow.
Ted Giola, a writer here on Substack across diverse topics between music and culture, has noted how he is being impersonated on Amazon to sell AI generated books based on his writings.
The ambiguities are only beginning.
Tellingly, images of human-robot hybrids proliferate in Generative AI. Here is a cursory survey of my search results when image searching “Midjourney Cyborg.” There are already an almost infinite number of images without much to distinguish them from one another beyond the mere rearrangement of features.
One cannot help but notice the proliferation of sexualized feminine robots. Gone are the days of the impassive terminator, and the cold, sterile, unfeeling war machine. Now the cyborg is sexy, or so the pictures tell you. Yet why is sexualizing a robot not a taboo? Few, it seems, can conjure up any moral outrage beyond a resigned sigh of disgust.
There is a feeling of inevitability about the whole situation. Perhaps it is the lack of clear boundaries, a moral ambiguity between porn, AI porn, AI digital assistants, and embodied AI relationships. These are all simulacra of relationships of varying degree, and it requires little imagination to see that boundaries between these questionable activities will grow only more vague as time goes on.
But there is something else going on in these images. Eroticism is not their main purpose. Without knowing it, without grasping what is happening, spontaneously we produce images of the principalities that rule over us. These images reveal to us the character of our era: the merging of the mechanical with the divine feminine.
A few years after Hausmann assembled his mechanical head, in 1927, fellow Austrian, Fritz Lang would unify these images in the film Metropolis, in which he realized one of cinema’s earliest and most memorable images of the Machine-Human:
Fitz Lang recognized the fundamentally religious and eschatological nature of what was occurring. AI, like the atom bomb of the last century, appears to our society as a harbinger of the end times. In both cases, extinction itself is not necessarily the deepest fear, but a placeholder for what we cannot see that lies beyond the door.
Moloch — that old name has found a referent again. What we might call in other language, a demon spirit of ruthless efficiency that demands endless sacrifices of our humanity in order to extend its body into the world.
with full knowledge and understanding they themselves offered up their own children, and those who had no children would buy little ones from poor people and cut their throats as if they were so many lambs or young birds ; meanwhile the mother stood by without a tear or moan ; but should she utter a single moan or let fall a single tear, she had to forfeit the money, and her child was sacrificed nevertheless ; and the whole area before the statue was filled with a loud noise of flutes and drums so that the cries of wailing should not reach the ears of the people.
Now the world presents to us yet another embodiment of the old demon, in the large-language-model that feeds on, what we might euphemistically call data, but what we we should really call the entire body of written and visual human achievement.
I hear people telling me “you better keep up, you better master AI, as a tool, or you’ll be left behind,” and I feel the hands shoving me up the steps.
Already ten years ago, Scott Alexander of Slate Star Codex, in an article that has become quietly influential, used the same name to describe the proliferation of lose-lose scenarios that pervade our society — the need to beat the competition at a games we know it would be better not to play at all except but to maintain a slight competitive edge. Scott, a rationalist atheist, didn’t realize he was describing a kind of spiritual causation, yet he stumbled into the power of that three millenia old name as the most apt metaphor for the decision trees that we are all forced into play out.
He was describing the spirit of our time.