Judaism's three doors into AI

One open, one opening, one closed.

Whenever a new technology shows up, religions can respond in two ways.

One kind of response is outward-facing: religions can influence shape the technology’s development or usage. This happens all the time: consider, for example, the availability of contraceptives and abortifacients in the United States, the regulation of the printing press in the early modern period, or the drive to develop mechanical clocks in medieval Europe.

The second thing is more inward facing: the technology can influence the religion itself. This, too, is very common: to pick a random example, I wrote my dissertation about how sundials, clocks, and other timekeeping devices completely changed how Jews discussed and regulated time over more than two thousand years.

These two responses aren’t mutually exclusive, but the dichotomy is helpful in evaluating how a religious has reacted to a given technology. I want to use them here to talk about Judaism’s response to AI so far.

On the outward-facing front, I’ve argued previously that the Jewish response is woefully inadequate and must be shored up while the public is still debating AI’s morality. The inward-facing response, on the other hand, is significantly more developed, but it is also in its infancy. In this post I want to map out what’s been done already, what’s happening now, and what ought to happen down the road.

Door #1: Halakhah / Jewish Law

At some point in the last century, a funny thing happened: people began using Jewish law—halakhah—as a kind of speculative language, as a way to talk about technologies that were emergent or even fictional. Exactly why this happened is a story for a different post, but some combination of a highly knowledgeable Orthodox laity, the erosion of rabbinic authority, and the adoption of the essay as a format for halakhic discussion have resulted in a situation where Jewish law—historically not known for being a first responder—is now deployed as a way to talk about lab-grown meat (not commercially available) or whether dinosaurs should be kosher (entirely fantastical).

Halakhah has been good for speculation about AI (here’s my favorite piece). Today, the vast majority of Jewish writings about artificial intelligence place halakhic language front and center, despite the fact that these responses are written by people who have no ability to enforce anything, for audiences that have almost no ability to follow through on their recommendations.

Now, a halakhic response to AI isn’t necessarily bad, but it has two major shortcomings.

The first is that AI is bigger than the law: sure, questions about liability for self-driving cars fit nicely in a halakhic framework, but questions about the overall morality of facial recognition do not.

The second, related concern is that it is relatively easy to formulate a legal response but quite hard to formulate a non-legal response, because legal responses have well-formulated questions and accepted rules for answering them whereas non-legal responses generally don’t. This means there’s a real danger that Jewish thought about AI will just stagnate in legal writings that cannot be enforced. It makes sense that AI started as a legal conversation—but it will be a failure if it ends there.

Door #2: Pedagogy

People respond to new technologies based on their existing competencies. Plenty of engaged Jews can speak halakhah, so a halakhic response was inevitable. But plenty of engaged Jews are also teachers, and in the past few months—really just since the debut of ChatGPT—it has become clear that the second major Jewish response to AI will engage with the technology’s implications for teaching and the classroom.

Some of these questions will be generic. AI-generated essays are already causing headaches and opportunities for teachers; the Jewish Education Project announced its next conference with the subject line “Will ChatGPT Write Your Next Lesson Plan?” A big chunk of the Jewish response to these thing won’t be particularly Jewish. The whole world is changing.

But a Jewish pedagogy for the AI age is still possible—in fact, it’s necessary. This is still new (for the world and for me) but there are at least three areas where a generic pedagogical approach won’t be enough.

1. How do you respond to a student who can instantly generate Torah—that is, text that bears all the hallmarks of sacred writings? What is the status of this text, and what does it mean that anything you can dream of can effortlessly be rendered in these styles? How do you maintain the idea of a “holy tongue”—meaning not just Hebrew, but special modes of writing—when the cost of translation is almost zero? (I wrote about this at length back in 2021.)

2. How do you incorporate AI into a curriculum or a school without falling into the trap that it is morally neutral? How do you use AI’s radical intrusion into the classroom as a way to talk about the moral significance of technology generally? How do you use a religious school’s power to instill moral ideas to create Jewish communities with a healthy relationship towards AI?

3. How does the idea of learning Torah l’shma—for its own sake—change or get reinforced when AI changes our conception of what school is supposed to teach?

I’ll confess that this is the area in which I have spent the least time, though the demand for answers I am seeing from Jewish educators is pushing me to spend more time here. This, at present, is the current frontier; it’s lively, the stakes are high, and unlike the legal response we reasonably expect it to yield concrete and enforceable policies. Watch this space; in a year from now I think it will be much more developed.

Door #3: Theology

Contemporary Jewish theology is a glorious and fascinating mess. There is no rhyme or reason to the way that it is taught, and there’s often a pretty big gap between the “official” line and what people will say if you press them. More interestingly, Jewish theology is (quietly) a place of enormous creative energy: left to their own devices, I’ve discovered that Jews not infrequently develop their own personal theologies. These are kept quiet out of concern that they will be received as heretical, but the truth is that it’s pretty hard to be a heretic and nobody is really minding the store. In short, Jews of all different denominations are surprisingly open to new ways of thinking about God, even if it’s a God they don’t believe in. (Again, there is a lot more to be written about this! Be patient, I will get there.)

Jewish thinkers have not written much about the implications of AI for Jewish theology, but AI clearly poses theological problems for Judaism. Here are a few ways of expressing those problems:

• If the premise of Jewish theology is a relationship between a unique God and a unique humanity, can it accommodate an AI that is, in some sense, a person?

• Jewish theology expresses a relationship between human and divine creative abilities. What does it mean that we can create machines of such sophistication?

• If our morality is premised on the idea that humans are created in God’s image, what happens when the line between humans and non-humans gets blurred? How do we ensure that humans do not get devalued? Do AIs deserve some modicum of the respect we give to people?

• Is the God/human relationship a good parallel for the human/AI relationship? What does it mean to go from being the created to being the creator?

• If tech companies begin implying that AI is God-like, what should the Jewish response look like?

Some of these questions may be more compelling to you than others. Some are attempts to solve technical problems, whereas others allow for wider thinking. Some, I’d even venture, could rejuvenate interest in Jewish theology itself.

There are Jewish sources that can be brought to play for all of these questions, and the conversation can spin out in fascinating ways. If Jewish thinkers are able to get over their fear of doing theology in public, if they’re empowered to take on this task, I think they will find that the theological angle both powerful and rhetorically influential. Law forces you into binaries, and pedagogy leads to policies—but theology rewards vivid dreams. It is also, I think, the necessary basis for any kind of moral response. This makes the production of AI theology not just a nice-to-have, but a moral imperative.

I hope this survey of the state of responses is useful to you, and as always I want to know what you think of it. In future installments I will delve into each of these topics more deeply.