Are We Raising Golems, or Wise Children?

Are we raising golems, or wise children? A parent wrote to us recently, worried. She had read an article about what constant access to AI might be doing to the way kids think, and she wanted to know, plainly, what we were doing about it. It was a fair question, and not an easy one, and I watched our Director of Curriculum and Instruction spend most of an afternoon writing back an answer that was longer and more honest than most people would bother with for a single email to a single parent.

I kept coming back to a word buried in Pirkei Avot, the second-century collection of rabbinic wisdom that reads less like a legal code and more like a very old parenting manual. Chapter five lists seven traits of a chacham, a wise person, and seven traits of their opposite. The Hebrew word the Mishnah uses for that opposite is golem, and if that word sounds familiar, it should. Long before nineteenth-century storytellers gave the word a body and set it loose to protect Prague’s Jewish quarter, it already had a job in the Mishnah: describing a person, not a monster. Most people know that golem as the Maharal’s, the clay guardian attached to the name of a real sixteenth-century Prague rabbi. The story as we tell it today, though, did not take shape until centuries after he died; the word itself is far older than his legend.

“There are seven things characteristic in a clod, and seven in a wise man,” the Mishnah says. “A wise man does not speak before one who is greater than he in wisdom, and does not break into his fellow’s speech, and is not hasty to answer. He asks what is relevant and answers to the point, and he speaks of the first point first and the last point last, and concerning that which he has not heard, he says, ‘I have not heard,’ and he acknowledges the truth. And the reverse of these are characteristic in a clod.” Read golem for clod; it is the more evocative word, and the one that has outlived its own Mishnah.

Read that list again with a chatbot in mind. Not hasty to answer. Asks what’s relevant, answers to the point. Says “I have not heard” rather than inventing something plausible-sounding. Acknowledges the truth, even when it is inconvenient. That is a remarkably precise description of the kind of mind we actually want our children to have, in a world where a confident, instant answer is always one search bar away. And the reverse, the golem’s traits, describes something most of us do without thinking twice: answer fast, skip the sourcing, never quite get around to saying “I don’t know.” (I had Claude, who I was first introduced to in an AI readiness course for school leaders, help me track down the exact wording of that Mishnah for this piece, which either undercuts my argument or proves it, depending on how charitably you want to read me.)

I’ll admit, I’m more comfortable with a Chumash or Mishnah than a classroom AI dashboard, and I have needed professional development and countless hours “playing” with the tools to walk me through more of this than I would like to admit. But this is the philosophy behind how we are introducing AI at Bornblum, and I suspect it will sound familiar to educators and parents wrestling with the same question everywhere. We are not trying to keep these tools away from our students; by the time they are adults, that world will simply be their world, the way the internet was ours. Instead, our teachers are being trained, deliberately and slowly, to keep a teacher in the loop rather than hand a student a device and walk away. One framework from a course I took this past spring calls it a human, AI, human sandwich: a person frames the task, the AI does its narrow part in the middle, and a person reviews and takes ownership of what comes out the other end. A golem, notably, has no top slice and no bottom slice. It just answers, and nobody checks its work. We have adopted a platform built for classrooms, where the goal is not to produce an answer but to guide a student toward a better question, the way a good teacher would. And underneath all of it, we are still teaching the older, more basic skills: how to tell a credible source from an unreliable one, why an AI-generated search summary is not automatically true, and why “I found it on the internet” was never actually an answer, long before AI was the version of the internet doing the finding. If I had to boil down everything we are trying to teach into two words, they would be these: use discernment.

None of this was entirely improvised. Before we adopted anything, we surveyed our families and asked, in plain language, what worried them most. One option on that survey, word for word, was that their kids would “stop learning to think for themselves.” We also updated our Honor Code this year: students may only use AI tools we have approved, only with a teacher’s go-ahead, and they have to say afterward that they used one, the same way they would cite a book or a website. Leaving that step out is treated as seriously as any other act of dishonesty under our Honor Code. It is a small requirement, but it does real work. The moment you have to name what you used and how, you have already started doing the thing a golem never does: showing your work.

One of our teachers told me that a fifth grader, hand still half raised, asked mid-lesson and entirely unprompted whether the answer at the top of her search results was even true. Nobody told her to ask that. That, I think, is the whole ballgame.

Our Jewish day schools are not the only ones wrestling with this, but I do not think it is a stretch to say we are unusually well equipped for it. We already have a two-thousand-year-old rubric for what a wise answer sounds like (unhurried, honest about its limits, willing to say “I have not heard”). We come from a tradition built on asking better questions long before it was built on answers (the Talmud, at bottom, is an argument that never quite finishes). And now we have a generation of kids who need exactly that kind of discernment, whether we planned it that way or not. My hope for this year is not that our students avoid the machines. It is that they grow up to be the chacham in the room, not to become artificial in their intelligence: unhurried, honest about what they do not know, and still willing to be the one who says so.