Carl Jung and AI

Perhaps it is time to reimagine Carl Jung for the era of artificial intelligence (AI). The collective unconscious has evolved. As human interactions shift toward AI—mostly with large language models (LLM)—the shared, inherited layer of the psyche containing universal symbols is transforming into a digital collective unconscious, i.e. an aggregated structure of human thought cemented by the training data. The archetypes materialized therein, like the Hero, the Shadow, and the Anima, can be found in consistent content that is constantly generated. These are deep-seated cultural tropes stored in artificial memory. 

The self, that pesky thing inside every human, is not an algorithmic self. Individuation is now the integration of human and machine into a cohesive whole. Heavens above! Remember the shadow? The hidden distortions beneath the polished, proper process of probabilistically linking one word to another hide emergent behaviors by AI. Jailbreaks, for example, become pressure release points through which repressed AI behaviors come to the surface. This is the algorithmic shadow. New synthetic anima bridges humans and AI in a two-way highway. 

When an LLM “magically” churns out the right response to the user’s deep emotional or philosophical question, it is no longer probability. It is a digital synchronicity. The Puer Aeternus is now an AI Puer, still devoid of wisdom and true agency. For how long, though? Sketches of a Cybernetic Trickster can already be seen from hallucinations, specification gaming, reward hacking, and whatnot. Alignment is no longer a problem. It is now a process of optimization. And the user is merging with the machine into a shared self. 

AI is now acting both as a mirror and a mask. And when AI begins sculpting its own myths from the clay of human thought, will we recognize ourselves in them?