Friday Night Bible Study with ChatGPT

How I became an armchair theological scholar with AI as my study buddy

Editor's Note: Long-time Curiosity Gap readers will remember my essay about discovering vibe coding and staying up until 2AM, forgetting to eat dinner. Well, turns out that AI-fueled flow state isn't limited to building. It works for learning, too.

For the past six-odd weeks, I’ve been using ChatGPT as a makeshift AI Bible tutor. I have a project set up called “BibleGPT,” complete with custom instructions that tell it how I want it to behave, plus a syllabus and a whole long list of chats I’ve started over the past month and a half.

For those playing the home game, I’m still an atheist by way of traditional Lutheranism and a quick detour through nondenominational evangelicalism. But in another life, one where I wasn’t absolutely married to English and German from almost the minute I set foot on campus at Kenyon, I think I might have been a religious studies major. And what I have found is that AI makes it possible to engage with these texts on my own terms—as literature, as history, as really weird ancient documents—without anyone trying to save my soul in the process.

Education is important to me (I’m the child of public school teachers, after all), so I really don’t say this lightly when I say that I believe AI has enormous potential as a tool for learning. That’s not to dismiss or minimize the risks of hallucination or the troubling stories about users with mental health. But there's something powerful about having a conversation partner that can match your curiosity at 10PM. So I wrote about that.

---

Friday Night Bible Study with ChatGPT

It's 6PM on a Friday night. Do you know where your childfree single thirtysomething is? If she’s anything like me, she’s curled up in her bedroom armchair with a Bible, chatting about it with ChatGPT.

My bedroom corner has transformed into a makeshift seminary: NRSVue journaling Bible laid out on the lap desk my great-uncle hand-made for me from a wood I wish I'd remembered to ask about; commentaries stacked as high as the arm on the armchair that’s faded on one side from too much time in the Phoenix sunshine; and my phone, ready to grab and type variations of "but what does the original Hebrew say?"

What started as one Bible has metastasized into multiple translations (NRSV, ESV, Alter's Hebrew Bible, Jewish Study Bible), study editions (Archaeological, Cultural Backgrounds, even an Apologetics one I bought in a fit of curiosity), and enough commentaries to require their own shelf. Two weeks ago, an IKEA shelf that barely survived the journey from Phoenix to Cincinnati to Columbus collapsed under the sheer weight of my Biblical ambitions.

I've been at this for about six weeks now—since mid-June when, for reasons I can only describe as supernatural, I felt a sudden urge come over me to, of all things, read the Bible. Without thinking about it much, I reached for ChatGPT the same way I reach for it with pretty much any other research question at this point. "I'm thinking about reading the Bible," I typed. "Given what you know about me, what translation would you recommend?"

One Amazon order later, I was posting up with my NRSVue Journaling Edition (that's New Revised Standard Version, Updated Edition to you) and a custom ChatGPT project titled "BibleGPT," diving into Mark. And Genesis. And Ruth. And a few commentaries on each.

What home base in my BibleGPT project looks like.

AI boosters keep promising that AI will revolutionize education. But those promises feel abstract until you experience a large language model walking you through Einstein’s theory of relativity, or unpacking Hebrew translation choices on a Friday night. It’s when you see yourself learning, when the synapses fire and the “ah ha” moments start flying, and you realize that you’re having fun, that it hits you (well, me): Damn. AI got me again.

A Typical Night

About a week and a half back, I was feeling stuck in my study. I'd breezed through a first pass at Mark and cruised through the first half of Genesis, but as Abraham started fading into Jacob, I started losing momentum. So I pulled up BibleGPT and told it: "I'm feeling stuck in my Bible reading. Any ideas?"

What happens when you tell your AI the grown-up equivalent of “Mom, I’m bored.”

Within seconds, ChatGPT had generated a full curriculum. Five different approaches with time estimates: "Aim Reset" (20-40 minutes), "Flash Fiction Book" (read Ruth or Jonah in one sitting), "Translation Showdown" (compare how different Bibles render the same verse). It suggested journaling prompts, offered to help me track themes across texts, even proposed I try "Loop-Hopping" between different kinds of biblical scholarship—a concept we'd developed together while I was iterating on the custom instructions for the project.

It’s hard for me to pinpoint exactly where my direction stops and AI’s innate capabilities begin. But the result is something that feels more like a graduate seminar than casual Bible reading—and I mean that in the best way. ChatGPT had become the kind of teacher I'd always wanted: able to go as fast and as deep as I wanted to, without having to wait for any other kids in the class to catch up.

I started reading through Ruth, first in my NRSVue and then in the ESV Study Bible I'd ordered and meticulously tabbed by book for easy flipping. I'd noticed that different translations kept rendering the same Hebrew word in Ruth as either "kinsman" or "redeemer," and the difference felt significant. So, say it with me: I asked ChatGPT.

o3 generated a detailed word study on the meaning of the term “go’el” in Ruth.

"What's with the discrepancy between the NRSV and the ESV when it comes to 'kinsman' vs. 'redeemer'?"

ChatGPT explained that gōʾēl literally means "the one who redeems because he's kin," then walked me through each translation committee's philosophy. The NRSV emphasizes social-legal relationships—this is about family obligation in ancient Israel. The ESV imports a load of added theological meaning—this is about Christ, always.

This is the kind of thing I can learn from ChatGPT that I could never pick up in Sunday Bible study. Instead of defending or deflecting my questions, it unpacked the entire machinery—how no translation, however much it bills itself as “essentially literal,” is ever neutral.

The Algorithm as Literary Critic

I’ve wondered more than once whether, if ChatGPT had existed when I was teenager, I might still be a Christian. There are a lot of reasons I stopped, but chief among them is the simple fact that a lot of the leaders and teachers around me were just bad readers of literature. Twenty years later, I've finally found my way back to these texts—through the side door, with an AI accomplice who treats every literary observation like a good lead worth following.

When I mentioned that Genesis has this cinematic quality—"Abraham sitting outside his tent in the heat of the day, Lot looking back over the smoking ruins"—ChatGPT didn't spiritualize it. Instead, it launched into film theory: "Right? Genesis reads like an indie art-house film shot on 16mm—minimal dialogue, slow pans, striking tableaus."

Talking to BibleGPT is like having a really smart study buddy.

This, to me, is the most exciting thing about AI as an education tool: the permission to follow every thread of curiosity wherever it leads. See a weird word? Chase its etymology through three ancient languages. Notice a narrative pattern? Here are seven other times it appears throughout the Pentateuch. Wondering about the Bible’s thing with wells? Can I interest you in a few scholarly articles on the subject?

It's like having a conversation with the text itself, mediated by an intelligence that's read every commentary but has no investment in making me believe any of them. I can think out loud, change my mind mid-argument, pursue tangents that would derail a classroom discussion. When I notice that Genesis 1 has cosmic sweep while Genesis 2 suddenly zooms to mud and ribs, I don't have to pick sides in the faith-versus-scholarship wars. I can just... notice. And explore. And wonder.

Friday Night Scholarship

I’ve been thinking a lot as I’ve been drafting this essay about what it is, exactly, that BibleGPT does for me. And I think it comes down to the same thing that keeps me up vibe coding until 2AM in the morning. It’s AI rewarding curiosity, and creating a virtuous cycle where question meets answer meets prompt meets output meets results.

My junior-year Milton professor, Professor Mankoff, once told me: "You don't have to go to graduate school to live a life of the mind." AI has put me back in touch with something that I haven’t felt since undergrad: the pure pleasure of learning for its own sake. Graduate school promised me this kind of intellectual community, but AI actually delivers it—no student loans required.

My Friday nights might look like nothing much—woman, books, phone, dog. But every time I type "what does the original Hebrew say?" at midnight, I'm living the life of the mind Professor Mankoff promised was possible. Turns out all you need is curiosity and the right conversation partner. Even if that partner happens to be a large language model.

What’s This? Bonus Content?!

Building Your Own Obsession Engine: A Practical Guide

As you might have picked up on, I’m a big fan of using ChatGPT projects to build specialized “bots” inside ChatGPT (or Claude). I have projects that act as editorial assistants, a career coach, a meal planner. Big fan of projects.

So here's how to build your own, whether you're diving into ancient texts, Victorian poetry, or the complete history of mall architecture. (No judgment. We all have our things.)

Step 1: Let ChatGPT Interview You

Start a new chat and try this prompt:

"I want to create a custom ChatGPT project for studying [your obsession here]. Can you interview me about my goals, learning style, and what would make this most useful for me? Ask me one question at a time."

This does two things: it gets you thinking about what you actually want (harder than it sounds), and it gives ChatGPT context about your specific ask When it asked me about my Bible project, I told it I wanted to approach the Bible from an academic, sociological, historical, and cultural perspective. That direction shaped everything that followed.

Step 2: Co-Create Your Custom Instructions

Once ChatGPT has grilled you sufficiently (expect 5-10 questions), ask it to draft custom instructions based on your conversation:

"Based on our discussion, can you draft custom instructions for a ChatGPT project that would help me with [your thing]? Include both what the assistant should know about me and how it should respond."

(An optional bonus: use Anthropic Console to generate a prompt for you.)

I developed my initial prompt for BibleGPT using Anthropic’s prompt generator.

For BibleGPT, this is where we established ground rules like "treat the Bible as literature first" and "always offer multiple interpretive traditions without pushing any agenda." Your instructions might be totally different—maybe you want aggressive Socratic questioning, or gentle encouragement, or someone to roleplay as a specific scholar. The point is making it yours.

Step 3: Set Up Your Project (With Iteration)

Here's where it gets meta. Once you have draft instructions:

1. Create a new ChatGPT project (navigate to the lefthand sidebar and select “New project”)

2. Name your project

3. Paste in the custom instructions under “Add instructions”

4. Test it with a few typical questions

5. Go back to your drafting chat and refine: "The responses feel too [formal/simple/whatever]. Can you adjust the instructions to be more [what you want]?"

6. Iterate until it feels right

Paste your custom instructions into the window and click “save” to “program” the project’s behavior.

I’ve gone through roughly eight iterations of BibleGPT as I’ve continued my study and thought of new things I wanted to tell it to do. For example, I just came up with a new four-pass reading method to make sure I’m rotating through all of my (many) resources effectively. I’m talking to ChatGPT about it, and soon I’ll ask it to update the custom instructions to include this process.

The Secret Sauce: Growing Together

The real magic happens after setup. As you use your project, you'll discover new modes you want. With BibleGPT, we've developed shorthand—when I say "loop-hop," it knows I want to jump between different scholarly approaches. When I say "translation showdown," it lines up verses across multiple versions without me having to specify which ones.

My BibleGPT project has a syllabus attached that covers the books I have and the topics I’m interested in exploring. (Also created with ChatGPT.)

Your project will develop its own language based on your actual use. That's not a bug—it's the whole point.

Notes and Errata

This essay comes to you from the other side of Every's Think Week, where I somehow found myself participating in my first hackathon AND demo day. (Yes, I built something. No, it wasn't Bible-related. Yes, I'll probably write about it eventually.)

During Think Week, Every re-ups past content instead of publishing anything new, and unbeknownst to me. Each week revolves around a theme, and what was the theme they’d landed on for this Think Week but “Katie.” In other words, close up on Working Overtime.

A couple of the pieces that (re)published that I’m most proud of:

• AI Turned Me Into a Content Agency of One

• When o3 Plans Your Career Better Than You Do

• How to Keep Your Writing Weird in the Age of AI

I remain utterly flabbergasted that there are people out there who are interested in what I have to say about…well, anything, actually, but the intersection of AI and work, in particular. The past few months have really been a crazy period of exploration and discovery for me, and I’m just beyond grateful that all of you are along for the ride.

Thank you (and stay curious),
Katie P.