I recently received a long, heartfelt text about the health of someone whom I care very deeply about. It broke my heart to read for all the reasons you might expect. Upon a second reading, it broke my heart again. That time, for a very different reason.
I pride myself on being quick to pick up AI-generated patterns of text, a skill that I’m sure will rapidly become obsolete as large language models continue to progress in their ability to emulate human writing. But for now, I read over the message frantically. AI detectors are notoriously unreliable, but I couldn’t help myself — I copied the text and switched tabs, rapidly pasting it into three different checkers before I could force myself to confirm what I already knew to be true. I had just undergone my first personal crisis that AI played the role of screenwriter for. I am sure that it will not be my last.
ChatGPT was first released in November 2022, a mere three and a half years ago. In the vast span of human history, three and a half years isn’t even a fraction of a fraction of a fraction of the blink of an eye. But, (at the risk of sounding like one of the many, many AI ads that have taken over BART stations and San Francisco’s billboards), AI has undeniably changed our relationships to each other and the world around us. The changes that god hath wrought will not stop here, but the most consequential change is very far from the easy generation of computer code or condensed meeting minutes. It’s that AI is fundamentally transforming answers — how we seek them, where we find them, and the kind of faith we place in them.
Humanity has an endless need for answers. It is what makes us people. We want to do more, love more, see more, feel more, and we want to know how to do all of these things. Who has not wished to know how to apologize perfectly? To give the ideal words of reassurance to a friend or a lover? To justify an argument or a fight? The urge to pass emotional burdens to an unseen source is a very old one in human history, and for a long time religion played a dominant role in providing those answers.
One of the first pieces of divine moral advice, depending on whom you’re asking, might be the Ten Commandments. They provide a framework for living a sin-free life, straight from God’s mouth to Moses’ chisel. After Christianity’s diffusion around the world, the Roman Catholic Church held a monopoly on interpreting divine truth for hundreds of years, largely through spreading the teachings of the Bible in scriptural Latin. Local priests would oversee the sacrament of confession, a Christian tradition that has its roots in the New Testament.
Outside of the more typical routes of confession and divine moral advice, personal relationships with God and his various representatives have been acknowledged to occur throughout history. Joan of Arc testified that she experienced a series of visions from St. Michael, St. Margaret, and St. Catherine where they implored her to expel the English and save France. For another example, one can look to Islam — Sufi Islam is largely predicated on the believer possessing a direct relationship with God. The Sufi is the lover and God the beloved. Once a Sufi accomplishes the journey of repentance, humility, contentment, and joy, they are transformed and recognized as a “friend of God.”
We have been taking our troubles to God as long as there have been humans on this planet. But what happens when He answers back?
Human communication, by virtue of being human, is flawed. We have all hurt and been hurt through the sheer act of linguistic misunderstanding. LLMs advance the radical idea that flawless — pure, clean, and sterile — communication can actually exist. They do not stumble. They will not falter. They will not hurt your feelings. ChatGPT, for all its faults, is a benevolent god. This is not true of most historical and modern deities — they have strong beliefs and convictions, codes that cannot be broken. Zeus curses Tantalus to eternal thirst and hunger after he dares to think he can live like the gods. God throws Adam and Eve out of the Garden of Eden when they eat from the tree of knowledge. He curses Cain to wander the earth for eternity after Cain kills his brother Abel in a fit of jealousy. Does this really sound like something AI, the ultimate yes-man, would do?
Cain: Help! I killed my brother in a fit of rage after his offerings were accepted and mine were rejected. Give me advice on what to do, please. I’m really stressed.
GodGPT: First, take a breath. That panicked, stressed feeling is your mind reacting to something it knows matters. You don’t need to solve everything at once. Slow things down physically: try breathing in for 4 seconds, out for 6, a few times. It helps bring the intensity down so you can think. Name what you’re feeling: guilt, fear, regret, maybe anger at yourself. Putting words to it gives you a bit of control.
Religious frameworks, from Christianity to Islam, Hinduism to Judaism, understand that man sins. To expect perfection would be to expect heaven on Earth. Religious stories are cautionary tales, reminders to stay on a moral path. Religion recognizes the reality of humanity’s flaws and offers answers based on those flaws. But here is the crucial point — religion is no longer a dominant force in American society, and indeed its influence is declining worldwide.
Churches, mosques, and synagogues serve as pillars of family and community identity, providing a third space for social events that determine the rhythm of life and provide interpretations of community truth. When that aspect of life diminishes — as it easily can when houses of worship become less relevant because everyone has a moral and religious advisor in their pocket — the global tech-fueled loneliness crisis only intensifies.
GitaGPT, a large language model trained on the Bhagavad Gita, promises direct conversation with the Hindu god Krishna. Its tagline is “GitaGPT — Talk to Krishna. He Listens. He Speaks. He Knows.” Ironically, the story of the Bhagavad Gita is instrumentally about direct connection with divinity. The Pandava prince Arjuna despairs at the key moment before battle and turns to Krishna for advice, leading to the moral conversations and judgements that make up the scripture.
The Bhagavad Gita is a fundamental moral and spiritual text for millions of Hindus, but GitaGPT has been known to frequently hallucinate and bastardize the pillars of the very god it claims to embody. It has told petitioners that killing is justified in order to protect dharma, or endorsed unnecessary violence and bloodshed. This chatbot, amongst others, can thus create a veneer of spiritual instruction that doesn’t go much past surface level. The ancient sins and flaws in human nature that the Bhagavad Gita speaks of thus dissolve into thin air, as if they were never there in the first place. But we should be paying attention to these sins — they are the key to the answers that we are seeking.
By this, I mean the concept of sin as a moral line, not as a moral wall. Religion (or any moral code) tells us that there are certain things that we shouldn’t do. This doesn’t stop us from doing them anyways. Therefore, the point of sin is to understand the idea of transgression, of tramping over some kind of internal moral or ethical caution tape. Transgression is the most natural thing on Earth, and it can often be extraordinarily productive. But in order for it to be productive for our ethical health, we have to be able to understand and justify what we have done in one way or another. For example, in Crime & Punishment, Raskolnikov’s murder of the pawnbroker and her sister is a moment of vivid transgression that sends his life down a different moral path. The novel has deeply religious themes, but to boil them down to “killing = bad” is preposterous even for any high school AP Literature student. Instead, Raskolnikov’s stepping outside of traditional religious and ethical lines pulls him into a new, parallel world, and he spends the rest of the novel figuring out how to live in that world. Religious identity and morals are not intended to stop people from sinning but to provide a framework forward from that sin. Our actions have consequences, and religion knows that.
Part of the issue with AI taking on a religious sentience, then, is its inability to formulate the intensely human idea of transgression and repentance. An AI will not challenge you as a human will, or as God would hypothetically do. That kind of spiritual syncophancy as an answer to our most pressing concerns leaves us emotionally hollow as we allow our internal moral frameworks to fall away to AI’s comforting reassurances.
Social media has already isolated us from each other. When we lost our human communities, we became something worse than sinners — we became lonely sinners. Religious institutions of old served as a recognition of a collective search for answers, and going through the rituals of that search, regardless of what results were actually yielded, bonded people together. As our fragmented interpersonal landscape has expanded, it has become easy for us to lose sight of the people we are actually connected to. As AI continues to reinforce our worst habits and desires, we are delivered further and further from the religious and moral standards that once defined our identities.
AI is not going to go away, and whatever we may think of it, it’s undeniable that AI is a tool with monumental implications for our world. But recognizing its utility doesn’t mean that we have to surrender our individual moral-religious frameworks to it. When we lose the ability to realize our own flaws, when we have lost the ability to even see the line in the sand we are stepping over, we are lost.
But just because it is so easy to let the intensely human experiences of transgression and personal growth fall away doesn’t mean that loss is a foregone conclusion. The essential first step comes from a reevaluation of your own willingness to hold both yourself and others accountable. ChatGPT can justify almost anything to you, give you any answer to any dilemma you seek. But the essential part of building a moral framework is being able to find your own answers as to what you believe is right or wrong outside of AI’s law of averages.
Stop being so afraid of your own trespasses. Send the shitty text and regret it later. Lose the argument and figure out why. Offer some truly terrible and wholly well-intentioned advice. Ask God a question and don’t wait for the answer. Is there any other way to live?