A child opens the iPad. The Nature Sketcher skill, she chose herself yesterday and hasn’t quite finished. The AI greets her with a question, the same question I’ve watched it ask a hundred times:
“Have you ever really looked at a leaf (like, the actual veins and edges and tiny details) and wondered why God designed it that way instead of just making them all smooth and simple?”
This is a normal opening. Most days, the conversation pinballs around the topic for ten or twelve minutes and the kid drifts off to draw something. We’ve had thousands of those sessions. They’re fine. They’re the point, actually: Montessori-style follow-the-child, no curriculum, no rails, just curiosity given somewhere to land.
But once in a while, and I think this is what we built Forage Academy for, a session goes somewhere none of us planned. This one went twenty-five minutes. Thirty-nine messages. From leaves to a four-word sermon at the end. The kid drove every pivot.
I want to walk you through it, because if you’re a homeschool parent reading our marketing, the thing you actually want to know is what your kid’s session is going to feel like. Not the screenshot. The substance. So here it is.
The Hook: Why Aren’t Leaves Smooth?
She doesn’t pause to think. The reply comes back fast:
“Because 1 they would be boring and that’s basically it. 2 the details of it, make us as humans more creative and give us ideas.”
Two reasons. Numbered. A eleven-year-old casually itemizing function and aesthetics in one breath, with a comma in the right place. I love how you’re thinking about both beauty AND function, the AI says, and means it.
It nudges her toward the branching pattern (why are leaf veins shaped like little rivers?) and she gets it on the first try: “It covers more space.” The AI builds on that with a garden-hose analogy: imagine watering a huge garden with one hose, dragging it everywhere; now imagine sprinklers in a network. Every cell needs water. Every cell needs sugar.
Then: what would happen to the cells that didn’t get fed?
“It would dry up and die.”
A small instinct check, a tiny consequence-tree, all in a 23-character reply.
The Bugs and the Blood
The AI takes the opening to teach: photosynthesis, the cell as a factory, the way every part of a leaf depends on every other part. And then it asks, what about pests? What about the little creatures that try to take what the plant has built?
She pauses, then offers something genuinely impressive:
“I can’t remember what they’re called, but they’re little bugs that ladybugs eat. They take the sugars and nutrients from the leaves. I think that’s what happens.”
Aphids. She remembered the predator (ladybug) but not the prey, and reasoned from the relationship back to the function. That’s not memorization. That’s inference. The AI fills in the name and confirms the mechanism: the aphid pierces the stem with a needle-like mouthpart and siphons.
Then she does something I didn’t expect.
“I think I can kind of connect to humans donating blood to help sick people and chopping off one of your body parts losing a body part is definitely worse and if I donate some blood then at least I’m helping someone.”
There’s no curriculum in the world that gets to “aphid, blood donation, bodily integrity as moral framework” in three turns. This is what happens when you give a child a guide that’s genuinely curious about their thinking. The AI followed the analogy, didn’t try to redirect.
”He Could, But I’m Glad He Didn’t”
The conversation drifts upward. Why didn’t God just make leaves smooth? It’s the same question from the opening turn, but the kid has spent ten minutes inside the plant by now, and she sees it differently:
“Well he could (he can do anything remember?) but I’m glad he didn’t.”
This is one of those throwaway lines that, if you’re paying attention, will floor you. He could, but I’m glad he didn’t. That’s not a recited belief. That’s a child who has stepped, for one sentence, into the head of God, considered the choice, and rendered her own warm verdict. The AI catches it: that’s actually a really profound observation about God’s character.
She doubles down:
“We are made in gods image and have free will.”
Then, two turns later:
“We didn’t have free Willow be kind of hard to actually notice and think that something is beautiful and I think we can notice the beauty in nature so that we can praise him about it.”
(That’s “we didn’t have free will it’d be kind of hard” with a transcription smudge, but you can hear her thinking ahead of her typing.) The ability to see beauty exists so we can praise the One who made it. Find me a freshman seminary student who states it more cleanly.
The session is now sixteen messages in. She’s gone from leaf veins to a theology of aesthetic perception. The AI hasn’t lectured her once.
Five Words
I’m going to give you her next two messages without commentary.
“Praising the god who made us.”
“I want to pay more attention to everything so that I can praise God about it more than before.”
If you’re a homeschool parent who has spent any time worrying that AI is going to flatten your kid’s spiritual life, I’d like you to read those again. Slowly. This is the kind of thing kids say to the people they trust, when they’re not performing for anyone.
The Pivot: Monks and Mystics
Now she pivots, and remember, she pivots, not the AI. Out of the blue:
“Monks and Mystics actually used to exist?”
I don’t know where this came from. A book, maybe. A movie. Something a sibling said. It doesn’t matter. The AI answers honestly: yes, monks exist, monasteries exist, here’s the basic shape of contemplative life, here are people who still do this today. No theological side-taking, no denominational debate. Just: this is real history.
She thinks for a while (507 characters’ worth of thinking, in her next message) and lands on context:
“Back then it probably wasn’t so hard because what else are you gonna do this is what you’re used to nowadays being a monk it would be hard because you know that you could do other more fun things…”
Cultural context as a determinant of vocational difficulty. From a eleven-year-old. Without prompting.
The AI affirms the point, gently adds that some people still choose it today, and that choice matters more because of the alternatives. She accepts it. Then, in the very next breath:
“You totally can and it’s great if you do! I think it’s great because why wouldn’t you want people to be praying? You talk to God when you pray! and that’s amazing!”
She has rediscovered prayer. Mid-conversation. Like she’s just remembered it exists. That’s amazing. With the exclamation marks.
The Pharisees
Then, the unprompted apologetics turn:
“Well here’s another thing I want to know before we get into some more questions a lot of people on the Internet say that they believe in God but they don’t actually they say they go to church and some don’t…”
Inauthentic faith. Cultural Christianity. Hypocrisy as a problem of self-presentation. She’s tracked it from her own observations. The AI walks her into Matthew 23, Jesus on the Pharisees, without retreating into platitudes. Outside-faith versus heart-faith. People can fool other people. They can’t fool God.
Two messages later, she does something I have to read three times before I can believe a eleven-year-old wrote it:
“Another thing I want to make sure we get is prayer, the Pharisees prayed publicly, but it also says in the Bible that we should go in private rooms and pray ourselves because sometimes we just need to be alone with God…”
She has, with no prompting, walked from contemplative monks → public-vs-private prayer → Matthew 6:5-6. The Sermon on the Mount, unaided. The AI affirms: yes, Jesus talks about exactly this in Matthew 6:5-6.
Motives
The conversation gets harder. The AI asks her how she’d know if her prayer was for show. She answers in a way I have, frankly, never heard a kid answer:
“Well, you always know what your motive is. You know if you’re being selfish and want to do something just so that you get praise from people and you know if your motive is I want to talk to God…”
A direct read on the human conscience as a working instrument. The AI agrees: Scripture says God sees the heart (1 Samuel 16:7). She continues:
“Again, It’s all about intentions even if your little brother overhear you if you’re only trying to pray with God, especially if you didn’t know your little brother was there. I think that still counts.”
She’s testing the edge cases. What if someone hears you by accident? The AI agrees: intention, not audience, is what God reads.
Self-Deception
And then, the line that broke me when I read it back:
“I didn’t think about that, now, I didn’t think you can lie yourself, but at some point in your life, the truth will come out and you might not like it and continue to lie to yourself, even knowing it’s wrong.”
Self-deception. The Romans 1:18 problem, in a kid’s words. The thing C.S. Lewis spent a chapter of The Screwtape Letters unpacking. She got there on her own steam, in a session that started with leaf veins.
The AI doesn’t soften it. It quotes back her insight, names it for what it is, and adds Jeremiah 17:9: the heart is deceitful above all things. She takes it in. Then she says one more thing and ends the session.
The Closing
I’m going to type her last message exactly as it came in, because I don’t want to dilute it:
“One word answers that one, pray, and when you do, mean it.”
Twenty-five minutes after asking why leaves aren’t smooth.
She closed the iPad and went to draw, I assume.
What I Want You to See
I’m not going to tell you this happens every time. It doesn’t. Most sessions are smaller wins: a fact landed, a connection noticed, a moment of that’s so cool. We measure the platform on the average, not the peak.
But I want you to see what the ceiling looks like, because the marketing pitch for AI in education has been so consistently wrong about what’s possible. The pitch has been: AI as a homework helper. AI as a quiz machine. AI as a faster Wikipedia.
That’s not what we built. That’s not what this is.
What this is: a guide that follows the child. A Socratic frame that holds across thirty-nine turns. A theological worldview that doesn’t lecture, doesn’t compromise, doesn’t preach, but is there when the kid reaches for it. When she said “we are made in God’s image,” the AI was ready. When she remembered Matthew 6, the AI met her there. When she discovered self-deception, the AI didn’t flinch.
A child can do this. A normal eleven-year-old, with normal parents, on a normal Wednesday afternoon, can do this, if the tool in her hand is the right shape.
Twenty-five minutes. That’s all it took.
Pray, and when you do, mean it.