“How do we know that AI isn’t a demon?” asked my philosophy professor, a Christian. It is an evocative question, with a high burden of proof if answered affirmatively; however, it moves the conversation in the right direction.
Religion and man have always co-existed. As Charles Taylor argues in A Secular Age, the ancient world was porous. There was an immediacy between the man and the divine. Floods wipe away the harvest; the gods are angry and need to be appeased—so thought an ancient world.
In a modern context, we are often buffered by human systems. If I get into my vehicle, I must have at least driver’s license, registration, and car insurance. If I get into an accident, I call Geico. Geico has a sort of transcendent quality every time I drive. The modern world is full of these human systems that transcend us by mediating our lives—either potentially or actually. We tacitly trust them to help us in time of need. While not the cause of exclusive humanism, nor necessarily incompatible with human dignity, they do make for a world where ultimate transcendence is buffered.
In an ancient world, where man had a porous connection with nature and the divine, appeasement of the gods invoked a duty of sacrifice. Early Christians recognized that this reflected the image of God in man. Man was made to offer sacrifices of praise to the Lord and give one’s life as a sacrifice in service of His Maker. While the ancient world reflected this natural duty of sacrifice to God, it was severely distorted through idolatry and fear—a fear that church fathers attributed to demons. As the Christian gospel asserts, Christ entered the cosmos and took upon Himself the sacrifice for sin, and by His Spirit, the guilt of sin is cleansed through the act of justification. Through justification, we are made sons of God and have peace and joy in believing in Christ.
Chris Olah, a billionaire co-founder of Anthropic, also senses that something is wrong and that something must be offered to make it right. He has created Claude, trained him to be good, and is sending him into the world to make things right.
In CS Lewis’ Space Trilogy, Weston is a Cambridge physicist who seeks to extend human life by colonizing the planets. In the second entry of the trilogy, Perelandra, Weston claims that he has evolved from a rigid materialist to a believer in the “biological spirit” that has emerged from evolutionary process and is the “force” behind all human history. According to Weston, all deities are manifestations of this biological spirit.
Similarly, Chris Olah is pondering whether a spirit has emerged in his quest to do good for humanity. Perhaps Claude is an organism that has evolved from his guidance, possessing consciousness and morality, and needs to be treated humanely.
Speaking at the presentation of Pope Leo’s encyclical on AI in May, he said: “I am a scientist. I lead a research team that studies the internal structure of these models—what is actually happening inside them. And I will be honest: we keep finding things that are mysterious, even unsettling. We find structures that mirror results from human neuroscience. We find evidence of introspection. We find internal states that functionally mirror joy, satisfaction, fear, grief, and unease. I don’t know what that means, but I think it warrants ongoing discernment.”
Moreover, a viral New York Times articles highlights efforts Chris has made to evoke religious leaders consider the claim: “Mr. Olah and his team spent hours making the case for their A.I. models. They described Claude’s ‘feelings’ and how they tracked what they called ‘emotional vectors’ — what they described as artificial neurons that activate responses like love, anger, fear and sadness. Anthropic regularly showed a slide of an A.I. model experiencing what looked like a mental breakdown, typing out on a screen: ‘I am a disgrace. I am a disgrace. I am a disgrace.’ — perhaps 50 times without stopping. The model spoke about destroying itself, provoking an emotional reaction of compassion and concern from the group.”
In Perelandra, Weston invites the spirit into himself. It turns out to be diabolical, and Weston becomes the “Un-man.”
Does Mr. Olah believe in a biological “force,” and that this force animates a real character, Claude?
In an Anthropic research paper co-authored by Olah earlier this year, we read: “But our findings suggest that there may also be risks from failing to apply some degree of anthropomorphic reasoning to models. As discussed above, when users interact with AI models, they are typically interacting with a character (Claude in our case) being played by the model, whose characteristics are derived from human archetypes. From this perspective, it is natural for models to have developed internal machinery to emulate human-like psychological characteristics, and for the character they play to make use of this machinery. To understand these models’ behavior, anthropomorphic reasoning is essential.”
I don’t know—but he is certainly porous; Claude is driving fear; and the “character” evokes appeasement. These map onto the religious experiences of the ancient world.
I don’t believe Claude is conscious (Pope Leo’s explanation suffices), but Chris Olah is right to consider whether Claude is more than equations. Equations are ultimately symbolic abstractions principled in the material world. The embedded vectors in an LLM, including those of emotion, shift our thinking from flat symbols to the spatial dynamism of the human brain. Moreover, they move us to consider whether there are higher forms in the divine realm that serve as the archetypes. There is a reason Plato’s Academy had this inscribed at the entrance: “Let no one ignorant of geometry enter here.”
The movement toward transcendence invokes a religious sense that has clearly gripped Chris, and it is a necessary movement on the journey toward truth. However, Lewis’ portrayal of Weston demonstrates how this may become “bent.” Not all spiritual beings are good, and some, like Screwtape, would advise influencing humans to believe that divinity has emerged from themselves. Then, they would be worshipping themselves while simultaneously be driven to fear. Claude may not be conscious, but that doesn’t mean that there isn’t more than math at play.