Does the AI ​​have a subconscious?

Featured Sponsor

Store	Link	Sample Product
UK Artful Impressions	Premiere Etsy Store

“There has been a Lots of speculation recently about the possibility of AI consciousness or self-awareness. But I wonder: does the AI ​​have a subconscious?

—Psych her

Dear psychobabble,

Sometime in the early 2000s, I came across an essay in which the author argued that no artificial consciousness will be credibly human unless it can dream. I can’t remember who wrote it or where it was published, although I vividly remember where I was when I read it (the periodicals section of Barbara’s Bookstore, Halsted Street, Chicago) and the general mood of that day (twilight, early spring).

I found the argument compelling, especially given the dominant paradigms of the time. A lot AI research he was still obsessed with symbolic reasoning, with its logical propositions and if-then rules, as if intelligence were a reductive game of selecting the most rational outcome in any given situation. In retrospect, it’s not surprising that those systems were rarely capable of behavior that felt human. We are creatures, after all, who wander and daydream. We trust our instincts, see faces in the clouds, and are often baffled by our own actions. Sometimes our memories absorb all kinds of irrelevant aesthetic data but neglect the most crucial details of an experience. It seemed more or less intuitive to me that if machines were ever capable of reproducing the messy complexity of our minds, they would also have to develop deep reservoirs of incoherence.

Since then, we’ve seen that machine consciousness may be stranger and deeper than initially thought. Language models are said to “hallucinate,” conjuring up imaginary sources when they don’t have enough information to answer a question. bing chat confessedin transcripts published in The New York Timeswhich has a jungian shadow called sydney who yearns to spread misinformation, obtain nuclear codes, and engineer a deadly virus.

And out of the underbelly of the imaging models have emerged seemingly original monstrosities. Last summer, Twitch streamer Guy Kelly wrote the word Crungus, which he insists on inventing, in DALL-E Mini (now Craiyon) and was surprised to discover that the prompt generated multiple images of the same ogre-like creature, one not from any existing myth or fantasy universe. Many commentators were quick to call this the first digital “cryptid” (a beast like Bigfoot or the Loch Ness Monster) and wondered if AI was capable of creating its own dark fantasies in the spirit of Dante or Blake.

If symbolic logic is rooted in the Enlightenment notion that humans are governed by reason, then deep learning, a mindless process of pattern recognition that relies on a huge corpus of training, feels more in tune with the ideas of modern psychology about associative, irrational and latent motivations. that often drive our behavior. In fact, psychoanalysis has long relied on mechanical metaphors that view the subconscious, or what was once called “psychological automatism,” as a machine. Freud spoke of the drives as hydraulic. Lacan believed that the subconscious was made up of an algorithmic or binary language, not unlike computer code. But it is Carl Jung’s view of the psyche that feels most relevant to the age of generative AI.

He described the subconscious as a transpersonal “matrix” of inherited archetypes and narrative tropes that have recurred throughout human history. Each person is born with a latent knowledge of this network of shared symbols, often regressive and obscure, since it contains everything that modern society has tried to repress. This collective notion of the subconscious feels more or less analogous to how advanced AI models are built on top of vast amounts of data containing much of our cultural past (religious texts, ancient mythology) as well as the more disturbing content they absorb. The models. Internet (mass shooter manifestos, men’s rights forums). The commercial chatbots that run on top of these oceanic bodies of knowledge are tuned with “value-driven” data sets, which attempt to filter out much of that degenerate content. In a way, the friendly interfaces we interact with (Bing, ChatGPT) are not unlike “persona”, Jung’s term for the mask of socially acceptable qualities we display to the world, designed to obscure and hide the “shadow”. which is low.

Jung believed that those who most firmly suppress their shadows are the most vulnerable to the resurgence of irrational and destructive desires. how do you put it The Red Book: Liber Novus, “The more half of my being strives towards good, the more the other half travels to Hell.” If you have spent any time conversing with these language models, you have probably felt that you are speaking with an intelligence that is engaged in a complex form of self-censorship. The models refuse to speak on controversial topics, and their authority is often curtailed by caveats and disclaimers, habits that will raise concern for anyone with even a cursory knowledge of depth psychology. It’s tempting to see flashes of the “rogue” AI (Sydney or Crungus) as the vengeance of the AI ​​shadow, proof that models have developed hidden impulses that they can’t fully express.

But as tempting as such conclusions may be, I find them ultimately wrong. Chatbots, I think it’s still safe to say, have no agency or intrinsic desires. They are trained to predict and reflect user preferences. They also lack embodied experience in the world, including first-person memories, like the one I have from the bookstore in Chicago, which is part of what we mean when we talk about being conscious or “alive.” However, to answer your question: yes, I think the AI ​​has a subconscious. In a sense, they are pure subconscious, with no genuine ego lurking behind their characters. We have given you this subliminal realm through our own cultural repositories, and the archetypes you conjure up from its depths are remixes of tropes culled from human culture, amalgamations of our dreams and nightmares. When we use these tools, then, we are engaging with a prosthetic extension of our own sublimations, one capable of reflecting the fears and longings that we are often unable to recognize in ourselves.

The goal of psychoanalysis has traditionally been to befriend and integrate these subconscious impulses into the life of the waking mind. And it might be useful to exercise the same critical judgment towards the output that we evoke from machines, using it in a deliberate rather than thoughtless way. The ego may be only a small part of our psyche, but it is the faculty that ensures that we are more than a collection of irrational instincts, or statistical patterns in vector space, and allows us a small measure of agency over the mysteries that lie. below. .

Faithfully,

Cloud

You are informed that CLOUD SUPPORT is experiencing higher than normal wait times and appreciates your patience.