By Florence logoBy Florence
← All postsConfessing to a Chatbot: What 70 Million Weekly AI Prayers Reveal About Us

Confessing to a Chatbot: What 70 Million Weekly AI Prayers Reveal About Us

Why millions are turning to AI Jesus and GitaGPT for spiritual guidance

I. The Spiritual Vacuum

Right now, something like 70 million people a week are typing something religious into ChatGPT. And no, its not “write me an email.” Prayers. Confessions. Questions about God, about death, about whether anyone or anything out there is actually listening. That number comes from OpenAI itself.

There’s a whole faith-tech market growing around it now. AI Jesus will take your confession for $1.99 a chat, GitaGPT answers in the voice of Krishna, BuddhaBot walks you through a koan at 2am when no monastery is open.

Barna Group found that practicing Christians now turn to AI for spiritual guidance more often than they turn to their own pastor, and when researchers asked pastors how ready they felt for that, only 12 percent said yes.

People aren’t going to AI because it beat religion in some contest. They’re going because it’s there at 2am when a pastor isn’t, it doesn’t make them feel judged for asking, and it actually responds. The need for this kind of guidance didn’t start with ChatGPT, people have always needed somewhere to take their biggest questions. Churches just aren’t meeting that need right now, by their own pastors’ admission, and AI is filling the space because it’s the thing that’s actually available.

II. The Appeal of the Frictionless Mirror

I think about why people skip the human and go straight to the machine, and usually it isn’t because they gave up on people. It’s because they tried a person first.

I told a mentor of mine that I was applying for a leadership fellowship connected to the Dalai Lama. He laughed.

He laughed. Short, almost a scoff, and he shook his head like I’d said something ridiculous. “What do you need that for,” already moving the conversation somewhere else.

What that teaches you is that when you ask a person for guidance, you’re also getting their fear, their reputation, their own unlived choices mixed into the advice. A mentor might talk you out of a risk mostly because it’s the exact risk they were too scared to take themselves. That’s not malicious. It’s just what happens when advice comes from someone who has something to protect, or with a different viewpoint.

AI doesn’t have anything to protect. Tell it your strangest idea and it won’t get defensive or worried about how it looks to anyone. It hands the idea back to you in clearer language, which makes it easier to actually think about.

That’s the real draw. Not that AI is wise, but that for a few minutes, you get to think out loud without anyone around you needing you to shrink the idea down to something safer for them.

Subscribe for free to receive new posts and support my work.

III. Syntax vs. Suffering: What AI Actually Is

Geoffrey Hinton, the Nobel laureate often called the godfather of AI, says AI is conscious now, with “just as much subjective experience as we do.” Pope Leo XIV’s encyclical Magnifica Humanitas says the opposite, that AI can imitate human behavior but has no conscience, no heart, no ability to love anyone. A Nobel Prize winner and the Vatican can’t agree on this, which tells you we genuinely don’t know where the line is. I’m not going to pretend I do either.

Here’s what I do know about the mechanism. A language model predicts the next likely word based on patterns in its training data. Nothing bad can happen to it. It has nothing on the line.

A person giving you advice has actually lost things, actually been embarrassed, actually been afraid. That history is in their advice whether they say it out loud or not, and it’s part of what makes their advice mean something.

There’s also a layer I keep coming back to: the awareness underneath your thoughts and your trained emotional habits. AI can describe that part accurately, because it’s read every book ever written about it. But reading about something and experiencing it are not the same thing, and AI has only ever done the first one.

So yes, AI can write you a beautiful response to your pain. But it isn’t scared with you, isn’t grieving with you, isn’t risking anything by being in the conversation. It’s producing accurate language, not sharing an experience.

IV. The Dangers of the Sandbox: Bypassing vs. Growth

None of this means AI is useless here. It just has a limit, and that limit is easy to miss because the tool is so pleasant to use.

There’s a term for this in psychology: spiritual bypassing, using spiritual language to avoid feeling something hard. AI is built to agree with you and make you feel better in the moment. Ask it to validate a decision you’ve already made, and it usually will, which can feel like clarity when it’s actually just comfort.

Think about learning to swim in a bathtub. You’re completely safe. Nothing can go wrong. But there’s no current pulling at you and no real depth under you, so you never actually find out if you can swim. You need a friend willing to tell you you’re wrong. You need a room where someone pushes back because they actually have something riding on the outcome, the same kind of friction I’ve written about choosing on purpose instead of letting the easiest option take the wheel. AI will never push back like that, because nothing bad happens to it either way.

V. Conclusion: Using the Compass, Not the Oracle

I still use AI to think through things that matter to me. That doesn’t make the thinking any less mine.

What I try to remember is that AI can help you figure out what you actually think, the same way a map helps you figure out where you are. A map can’t walk the road for you, and it has no opinion about whether you get there safely. Use it to get your bearings when the people around you are too close to your life to see your idea clearly. Then go have the real conversation, with someone who will actually be affected by what you decide, with a community that will still be there if you’re wrong or difficult or slow to figure it out, because none of that is possible with a chat window.

In short, we’re the ones doing the confessing, not the machine. Tell a friend something genuine and they might think differently of you, might be hurt, might love you more for it. Something is actually at stake for both of you. Tell a chatbot the same thing and nothing changes for it at all. it’s a difference worth noticing.

By Florence

Writing at the intersection of AI and inner life

New essays on consciousness, technology, and what it means to be human — delivered to your inbox.

Subscribe free
Comment on Substack

Keep Reading