In his 1969 novel “The Golem” (translated into English in 1982), Isaac Bashevis Singer brings to life another Golem story, the first of which dates to the Jewish mystic Rabbi Judah Loew ben Bezalel (the Maharal) in the late 1500s. Singer’s take explores the responsibility of creation, power and restraint, and the nature of humanity. In the book, Rabbi Leib seeks to defend the Jewish community against the blood libel by using mystical incantations to animate Joseph, a giant made of clay. The rabbi’s wife, Genendel, persuades Leib to use Joseph to dig up a legendary hidden treasure to help the poor, and this misuse causes the Golem to grow out of control. Joseph begins to develop human-like emotions and desires. As the Golem becomes unmanageable and potentially dangerous, Rabbi Leib is forced to deactivate him by erasing the Holy Name from his forehead.

Every generation invents new machines and then reaches backward for old stories to explain why the machines feel unsettling. We did this with clocks and trains and televisions and the internet. When technology changes faster than our language, myth becomes a kind of emergency vocabulary. It gives us a way to name what we sense before we can prove it. In the AI era, one metaphor keeps returning in Jewish conversation with special force; the Golem, the human made servant that follows instructions, grows powerful, and threatens to exceed its maker’s control.

It’s tempting to dismiss the Golem as an antiquated warning, the kind of tale once told to children to scare them away from forbidden knowledge. But the Golem story isn’t primarily about forbidden knowledge, it’s about responsibility. It’s about what happens when we create power without building the moral and communal structures to govern it.

That’s why the metaphor fits. Not because AI is literally a clay creature lumbering through the streets of Prague, but because we are living through a moment in which humans are animating systems that can act at scale, that can amplify harm, distort reality, and reshape the conditions of social life. The Golem is not a prediction, it’s a mirror.

What the Golem Story is Really About

 Across versions of the Golem legend the Golem is created to protect the Jewish community. It is born from urgency and is an answer to danger. It is, in that sense, a sort of security technology.

The detail that matters most is that the Golem stories rarely end with uncomplicated triumph. Sometimes the creature grows too strong and sometimes it misunderstands instructions. Sometimes it continues working after it should stop and sometimes it becomes dangerous not because it turns evil, but because it has power without judgment. That dynamic is the key. The Golem is a being of execution without discernment. And if that sounds familiar, it should.

AI systems can do things at speed and scale that humans cannot. They can generate, recommend, rank, summarize, surveil, simulate, and predict. They can carry out a goal across millions of interactions. But what they cannot reliably do is the human work of moral discernment, understanding context, weighing competing obligations, sensing when not to act, or noticing when an instruction is wrong even if it is well-formed. This Golem obeys. It scales. It cannot and does not ask, Should I?

Modern Golems Aren’t Made of Clay, They’re Made of Incentives

There’s a second reason the metaphor fits, one we talk about less because it’s not as romantic. A classic Golem story has a single creator and a single creature, one rabbi, one servant. Modern AI has an entire infrastructure of companies optimizing growth (with platforms developed for engagement), institutions pursuing efficiency, governments improving security, users seeking convenience, and investors insisting on a return. When you combine AI with incentives, you get something that looks like a Golem ecosystem, enterprises that keep moving because the environment rewards movement even when it causes harm.

This is one of the deepest modern parallels. In many AI deployments, there is no clean off switch. There is only market pressure, competitive fear, organizational dependency, bureaucratic inertia, political risk, and reputational management. The Golem doesn’t run amok because it becomes evil, it runs amok because it becomes useful. And usefulness is the most persuasive force in modern life.

The Name on the Forehead: Language as Power and as Vulnerability

 In many stories the Golem is animated through sacred language. Sometimes the word is emet (truth), written on the Golem’s forehead. Remove a letter turning emet into met (dead), and the Golem collapses back into clay. That detail is almost too perfect for the AI era.

We are now surrounded by systems that are animated by language like prompts, policies, terms of service, and the silent language of metrics – engagement, retention, and growth. AI doesn’t just respond to language, it is built from language. It learns patterns in text, generates text, persuades through text. Our Golems run on words. And the same way the Golem can be turned off by altering a word, AI systems can be redirected, or weaponized, by altering language. A definition becomes a political weapon, a euphemism becomes a shield for hate, and neutral summary becomes propaganda.

If emet matters, it’s because in a language driven world, truth is the difference between animation and collapse, between moral life and moral chaos.

The Misunderstood Instruction Problem: Alignment is a Golem Problem

One of the most persistent features of the Golem myth is literal mindedness. The creature does what it is told, not what is meant. AI systems regularly produce Golem outcomes when the goal is specified imperfectly. When a platform seeks engagement and gets outrage it is not because the system is malicious, but because it is obedient to the wrong objective. The Golem warns us that an animated, powerful executor with imperfect instruction deliver’s the instruction’s consequences, not the original intention.

The Deeper Fear is That the Machines Will Unmake the Human

Public anxiety about AI often defaults to science fiction with killer robots, apocalypse, and machines replacing humanity. But Jewish anxiety frequently has a different tone shaped by historical experience, the fear of dehumanization. This is because Jews have lived through what happens when a society trains itself to see certain people as less than human. This is where the Golem metaphor becomes psychologically precise.

The Golem is not a human, it is a human-like force. If it becomes the model for how power operates, it can reshape the moral environment. It can normalize a world where decisions happen without accountability, speech becomes automated, people become data profiles, truth becomes optional, and suffering becomes a statistic. In that world, the danger isn’t only what AI does to us, it’s what it trains us to accept. A society that relies on Golems begins to think like Golems.

The Golem as Protector: Why the Metaphor Isn’t Anti-Technology

 The Golem is created to protect Jews. It is a defensive response to real danger. This certainty prevents the metaphor from collapsing into simplistic techno-panic. The Golem story contains a moral paradox: Protection is necessary, but power is dangerous. Refusing power can leave the vulnerable exposed. Using power can corrupt the user and harm the innocent.

That isn’t a reason to reject AI, but it is a reason to treat AI as moral power, not neutral machinery. The Golem metaphor calls on us to develop AI in a way that doesn’t betray what we were trying to protect.

The Jewish Corrective: Tools Must Remain Subordinate to Human Responsibility

 Jewish tradition doesn’t reject tools, but it does reject the act of treating a human-made thing as an ultimate authority. The danger is not only that AI will be wrong, but that we will treat it as an authority because it speaks with confidence and scale.

This is Golem idolatry, when the creature’s output becomes more real than reality. We see it, for example, when generated image replaces evidence. The antidote is not hostility to AI, but insistence on a hierarchy wherein humans remain responsible, institutions remain accountable, and truth remains non-negotiable.

The Off Switch Today is Governance

 In the legend, the Golem is shut down by removing a letter. Unfortunately, modern systems don’t work that way. This is an unglamorous truth, but the letter that deactivates a modern Golem is often an institutional decision to prioritize social responsibility over profits. That feels less mythic than a sacred name, but it is the real work. And since AI companies cannot be trusted to voluntarily self-regulate against practices that spread misinformation and enable hatred, government enforced regulations are necessary.

Why This Metaphor Matters Specifically for Jewish Life

 The Golem story, at its core, is a communal story. It forces the community to build what is needed to protect the community, and ponder what happens if our protection becomes its own threat. That is exactly the dilemna Jewish institutions now face as they adopt AI tools for communication, education, fundraising, security, and operations.

The Golem legend exists to teach us that power without wisdom becomes dangerous, even when created for good reasons. AI is an amplifier that magnifies competence and confusion, safety and surveillance, truth and propaganda. The question is whether we will treat it like a servant, and not a sovereign, never a substitute for moral responsibility.

In the age of digital Golems, our task is not to fear the old metaphors. Our task is to learn from them, to build systems that protect without dehumanizing, that inform without hypnotizing, and that preserve emet in a world increasingly optimized for viral illusoriness. The Golem story ends badly when people forget that they made a force of execution without judgment. But the story ends well, if it ends well at all, when the community remembers that what we create does not release us from responsibility, it intensifies it.

© The Times of Israel (Blogs)