Inscribing the Digital Golem: The Ethics of AI

Releasing a Book into the Wild

When I released my new book, AI for Clergy: Harnessing the Power of the Digital Golem for Spiritual Leadership, Community Organization, and Sacred Service, I expected some curiosity, maybe even excitement. What I didn’t fully anticipate was the strength of the responses—warm encouragement from colleagues, thoughtful questions about ethics, and, yes, sharp critiques.

I knew there were different opinions about AI. What I wasn’t prepared for was just how forcefully some colleagues opposed it, and how personal some of the critiques felt. Putting out a book is always vulnerable. You’re offering your thinking, your heart, your vision into the world. This book, though, touches a nerve. AI is not just another gadget—it’s a mirror reflecting our hopes and fears, a tool that unsettles as much as it promises. And when you’re a rabbi writing about AI, you can be sure that people will ask: Is this sacred? Is this dangerous? Is this even allowed?

After one particularly tense exchange, I spoke with a friend who reminded me of a simple truth: every new technology provokes resistance. From the printing press, to the radio sermon, to livestreamed High Holy Day services, the first adopters often face suspicion or even outrage. In time, though, communities discern how to use these tools responsibly, and they often become indispensable.

The Critiques

On rabbinic listservs where I shared news of the book, some colleagues responded with generous collegiality—mazal tovs, genuine curiosity, and questions about how to use AI responsibly in pastoral life. Others pushed back hard.

One colleague wrote bluntly: “There is no way any rabbi, let alone any member of this rabbinic association, should be advocating the use of AI, let alone using it to ‘write’ a sermon.” Another insisted that generative AI is, by definition, plagiarism, and that its very existence represents intellectual theft on a massive scale. Others raised the issue of its environmental impact, or worried that clergy might hand over sacred responsibilities to machines.

I take these critiques seriously. They come from a place of protecting the dignity of our work and the integrity of Jewish life. But some responses veered into ad hominem territory, questioning not just the tool but the ethics of anyone who would use it. That stung—and it clarified for me just how high the stakes feel in this conversation.

The Question of Plagiarism

One of the strongest charges leveled against generative AI is that it amounts to plagiarism. “It steals from the work of human beings,” I was told, “and spits it back out disguised as something new.”

But this misunderstands how large language models actually function. They don’t copy and paste text from their training data. They don’t store passages and reproduce them on demand. Instead, they analyze patterns across vast amounts of text and generate novel language in response to prompts. If plagiarism means presenting someone else’s words as your own, then AI doesn’t fit that definition.

The unsettled legal question is whether using copyrighted works to train these models constitutes infringement. Some judges have ruled that training falls under “fair use,” since the outputs are transformative and not reproductions. Others have disagreed. The jury is, quite literally, still out. But let’s be clear: when you ask ChatGPT or Gemini to draft a prayer or sermon, it does not spit back someone else’s prayer or sermon. It produces something new, stitched together in real time, often shaped by your own prompts and style.

That said, I do not advocate cutting and pasting AI text into a sermon or eulogy and calling it done. That would be lazy, uninspired, and inauthentic. In my book, I emphasize again and again: clergy must make the material their own. Edit it. Revise it. Infuse it with presence and spirit. AI can spark, suggest, and draft—but the soul of the work belongs to us.

Still, the intensity of the plagiarism critique is telling. It reflects not only ethical concern but also the deep unease that comes whenever a new technology disrupts our relationship with knowledge.

When the printing press first spread Torah beyond the scribe’s desk, some decried it as profaning the sacred. Rabbis worried that anyone could own and handle holy texts without scribal training. Torah scrolls still had to remain handwritten, but printed Talmuds, halakhic works, and prayerbooks eventually became essential to Jewish life—despite the early resistance. Scribes feared for their livelihoods, and scholars doubted the accuracy of printed editions.

When rabbis first began sending sermons over the radio, critics insisted no spiritual truth could travel across airwaves. Jewish authorities even debated whether listening to Kol Nidre on the radio counted as attending synagogue (the consensus: no). Protestant preachers like Aimee Semple McPherson drew suspicion for bypassing traditional congregational structures through mass media.

Generative AI is provoking that same visceral pushback. The charge of plagiarism, in this sense, is less about legal definitions than about the anxiety that something precious—authenticity, creativity, soul—is being stolen. And it’s precisely because that anxiety is real and important that we need to engage it, rather than dismiss it.

The Environment: Hard Questions We Can’t Avoid

The hardest critiques are often the ones that contain real truth. When colleagues told me that AI is bad for the environment, I couldn’t just brush that off. They’re right—these tools consume energy. The massive data centers that power large language models require electricity, cooling, and infrastructure, and that has a carbon footprint. To ignore that would be irresponsible.

But so does nearly everything we already rely on for ministry. The phone calls we make, the emails we send, the laptops we type on, the livestreams we broadcast—all of them draw from the same grid, often fueled by unsustainable sources. The question is not whether AI has costs. It does. The question is whether the benefits of careful, limited, wise use can outweigh those costs.

It also helps to clarify the scale of impact. The enormous energy demand comes mainly during the training of AI models—when billions of words are processed across vast networks of GPUs. By contrast, the act of using a model—typing a prompt and receiving a response—requires relatively little power. A friend of mine in tech reminded me that a single chat like this one uses no more energy than sending a handful of emails. Streaming a Netflix movie or running a Zoom service consumes far more.

For me, that means treating AI like any other morally mixed technology. When the first cell phones arrived, few of us knew about the coltan mines in the Congo, where extraction fueled violence and ecological devastation. Today we take smartphones for granted as indispensable, while still wrestling with their shadow side. The same holds for livestreamed worship: during the pandemic, many synagogues and churches streamed High Holy Day services over servers that consumed enormous energy. And yet, without that carbon cost, thousands of isolated congregants would have had no access to spiritual community.

AI now joins this lineage. To use it without awareness is careless. To reject it outright is unrealistic. The moral ground lies in discernment: How do we minimize harm while still embracing the possibilities? How do we model restraint in an era addicted to “more”?

For me, the answer begins with scale. I don’t keep an AI window open all day, generating endless drafts. I use it sparingly, for targeted tasks: a sermon outline when I’m stuck, a letter to help frame compassion when words feel tangled, a translation to make Torah accessible to a new immigrant family. That kind of use doesn’t erase the environmental impact, but it keeps it proportional. It treats AI as a tool to be picked up when needed, not a faucet left running.

AI’s Carbon Footprint: A Balanced Look at Individual Use, Industry Scale, and the Road Ahead

The meteoric rise of artificial intelligence has brought with it a cascade of excitement and concern. Among the most pressing critiques is its environmental impact, particularly its seemingly insatiable appetite for energy. But is the individual user truly “harming the environment” with every query, or is the picture more nuanced?

Your AI Queries vs. Your Daily Life
 Data suggests that your personal AI usage has an almost negligible footprint compared to daily activities. A single ChatGPT query consumes roughly 0.34 watt-hours of electricity. Even at 100 queries a day, the annual total would be about 12.4 kilowatt-hours—roughly 10 pounds of CO₂.

Now compare that to everyday life:

• Driving a car: over 10,000 pounds of CO₂ annually.
• Flying round-trip from New York to Tokyo: about 4,400 pounds of CO₂ per person.

In other words, your annual AI queries are about 1,000 times smaller than your annual driving emissions. From an individual standpoint, using AI to enhance your workflow—perhaps spending less time on your computer writing and rewriting—could even save energy overall.

The takeaway here is crucial: while every watt-hour counts, the critique of individual AI usage as a major environmental burden is largely misplaced. The true challenge lies at a much larger scale.

The Industrial Scale: AI vs. Transportation
 The U.S. transportation sector consumes roughly 8,200 terawatt-hours annually and contributes over 14% of total greenhouse gas emissions. By contrast, U.S. data centers consumed around 178 TWh in 2024, of which AI accounts for 27–35 TWh. That means transportation’s energy demand is still 230 to 300 times larger than the entire AI industry.

Yet, unlike transportation, which is stabilizing, AI’s energy consumption is projected to grow by 165% by 2030. Its demand is concentrated in massive data centers, straining local grids and often requiring new gas-fired plants. The concern is less its current size than its explosive growth.

At the same time, AI holds the potential to reduce emissions in other sectors—optimizing logistics, predicting maintenance, managing smart grids. If guided wisely, it may offset more emissions than it generates.

A Path Forward

The critique of AI’s environmental impact is valid, but its nuances are often lost in broad generalizations. Your personal AI usage is a drop in the ocean compared to daily activities. The systemic challenge is real—AI’s growth is rapid and concentrated—but companies like OpenAI, Microsoft, and Google are already investing in renewable energy, efficiency gains, and greener infrastructure.

Jewish tradition offers the principle of bal tashchit—do not destroy needlessly. That doesn’t forbid all use of resources; it calls us to weigh necessity against waste. The same applies here. The goal is not to stop AI, but to steward its growth responsibly, pressing industry leaders toward clean energy while practicing personal restraint and discernment.

This is the hard question we cannot avoid. Yes, there are costs. But if we meet those costs with humility, advocacy, and creative responsibility, the benefits may still be worth it—not just for our ministries, but for the wider world AI might one day help to heal.

Absolutely. That broadening will make the argument more compelling: clergy aren’t the only ones who benefit—AI’s accessibility potential touches students, immigrants, the working poor, and others who face systemic barriers. Here’s a revised, extended Section V that begins with your personal story and then widens the lens:

Yes — expanding this section into a horizon view makes the piece much more appealing to a wider readership like The Jewish Week. We’ll keep your personal story at the heart but extend the benefits outward: first to clergy, then to everyday accessibility, and finally to the global horizon of what AI might enable if stewarded wisely. Here’s a revised, extended Section V:

The Benefits: Why We Shouldn’t Ignore This Tool

I want to share something personal about why I wrote AI for Clergy and why this work matters so deeply to me.

Like many of us, I carry both gifts and challenges. I’m creative, musical, passionate, and brimming with ideas—sometimes a thousand a day. But I also live with difficulties: my thoughts can get scattered, my executive functioning can falter, my emotions sometimes run away with me. These challenges can make it hard for my gifts to come through as fully as I’d like.

When I was a child, my teachers wanted to hold me back in kindergarten. My parents had me tested, and the results were unusual: I scored at a genius level of intelligence, but I was also diagnosed with an organizational learning disability and a small motor learning disability—what today we would call dysgraphia. That meant that while my mind overflowed with ideas, my ability to organize them, to follow through, to get them down clearly on the page, often lagged behind. I’ve spent an inordinate amount of time wrestling with words—sermons, eulogies, papers in graduate school—obsessing over them long past the point where they were “finished.”

AI has been nothing short of a game-changer for me. It doesn’t replace my voice—it helps me find it. It helps me organize my thoughts, clarify wording, and focus when I feel overwhelmed. In short, it’s an assistant that allows me to bring more of myself—more of my creativity and care—into my rabbinate and my life. With AI’s support, I’ve been able to complete projects I once agonized over for years: books, contemporary gender-inclusive translations, new spiritual resources for my community. This technology hasn’t diminished my creativity—it has multiplied it.

We all rely on different forms of assistance. For some, it’s medication. For others, a mentor, a cane, a practice of prayer or meditation. For me, AI has become one of those supports. It is not doing things for me but with me, like having a smart editor or colleague always available. It’s a partner in clarity, not a substitute for my presence. And it has freed me to spend more time where I am needed most: with people, in pastoral care, in teaching, in the flesh-and-blood work of ministry.

And clergy are not the only ones who can benefit. For students with dyslexia or ADHD, AI can be the bridge that transforms frustration into achievement. For immigrants and refugees, AI-powered translation can open doors to work, healthcare, and education. For workers who cannot afford a lawyer, AI can help draft letters, understand contracts, and navigate bureaucracies that otherwise exclude them. For small nonprofit organizations with limited staff, AI can unlock opportunities once reserved for institutions with large budgets.

In every case, the pattern is the same: AI does not replace human agency—it amplifies it. It takes what is already there and gives it clearer form, helping people bring more of themselves into the world.

And then there is the horizon. While today’s AI mostly helps us with words and patterns, the near future points toward breakthroughs that could reshape whole fields. Scientists are already using AI to accelerate the discovery of new medicines, to analyze genetic data for rare diseases, and to develop more efficient renewable energy systems. AI is being tested in climate modeling, urban planning, and food distribution—places where small efficiencies could save millions of lives.

Of course, there are risks: monopolization by big tech, embedded bias, misuse by bad actors. But if guided responsibly, the same tools that help a rabbi finish a sermon or a student complete an essay could one day help cure diseases, reduce hunger, and address systemic injustices. The Talmud teaches that to save a single life is to save an entire world. What would it mean if AI, guided by human conscience, could help save not just one world, but our shared world?

Framed in terms of impact:

• Individual: For a rabbi like me—or a student with dyslexia, or a single parent applying for jobs—AI can be a lifeline, a tool that turns overwhelm into focus.
• Systemic: For small congregations without staff, underfunded schools, or low-income families navigating bureaucracy, AI can offer equity and empowerment.
• Future: If stewarded wisely, AI could become a force in curing diseases, fighting climate change, and repairing the broken places of our society.

This is why I refuse to dismiss AI as “inauthentic.” To me, authenticity is not whether every word comes straight from my unaided hand. It’s whether my words—however they came together—allow me to show up fully, heart and soul, for my community. That same principle applies to students finding their voice, workers finding justice, scientists finding cures, and societies finding hope.

Authenticity lies not in purity of process, but in presence of spirit. And when used with discernment, AI can help more of us bring that presence into the world.

 A Broader Frame: The Digital Golem

There’s an old story about the Maharal of Prague and his golem that I keep returning to. One day, the Maharal’s wife asked the golem to bring water into the house. Obedient as always, it carried bucket after bucket after bucket. But because she hadn’t told it to stop, the golem kept going until the house was flooded.

That story captures the paradox of the golem. It was created to serve, not to think. It had no soul, no discernment, only obedience. And in its very obedience lay its danger. Without the wisdom of its maker to guide it, the golem became destructive.

AI is our modern golem. It is powerful, soulless, and obedient to the commands we inscribe. It does not choose—it executes. What matters is not whether the tool exists, but what words we carve onto its forehead, what limits we set, what wisdom we bring to its use.

This is where the metaphor becomes helpful. The danger is not AI itself, just as the danger was not the golem itself. The danger comes when we abdicate our moral responsibility, when we treat the tool as master rather than servant, or when we forget to tell it when to stop.

Fear of the golem is not new. Every generation of Jews who told the story wrestled with its double edge: protector and threat, miracle and menace. The printing press, the radio, the internet—all have carried a similar doubleness. First they were feared as profaning, dehumanizing, even heretical. Then, over time, they became indispensable to the work of transmitting Torah and building community.

The lesson is clear: it is not enough to fear or to embrace blindly. Our task is to use wisely. The Maharal knew that a golem could defend a community from danger—but he also knew it could not replace human life, or human soul. Likewise, AI cannot replace presence, love, discernment, or responsibility. It is not our rabbi, our cantor, our therapist, or our friend. It is a tool—a powerful one—that reflects back the letters we inscribe.

So the real question is not “Is AI sacred or profane?” but “What do we carve into its forehead? What instructions do we give, and where do we remember to set boundaries?” That is where our responsibility lies.

If we fail, the risk is real: endless buckets of water, flooding the house. But if we remember to guide, limit, and remain present, this modern golem can serve—not master—our communities.

The Deeper Question: Soul and Projection

Whenever I share an AI-generated prayer, poem, or blessing, the reactions fall into two extremes. Some people say, “This is soulful, it moves me.” Others insist, “This is hollow, fake—plagiarized even.” How can the same words elicit such opposite responses?

The truth is, AI has no soul. It does not feel, it does not yearn, it does not pray. But we do. And when we read its outputs, we can’t help but project onto them—our own fears, our own longings, our own questions about authenticity and value.

In this way, AI is less a creator than a mirror. If you fear that technology is stealing your craft, you will see theft in its words. If you worry that human labor is being undervalued, you will hear exploitation in its phrases. If you are searching for beauty, you may find it there—because you brought beauty to the reading.

The critique of AI often tells us more about ourselves than about the machine. We are the ones who feel anxious about what makes something authentic. We are the ones who measure worth in labor and originality. We are the ones who worry about being replaced, about whether our creativity is “enough.”

This is why the golem story resonates so deeply. The danger was never that the golem had a soul of its own. The danger was that people treated it as if it did. They trusted it too much, or feared it too much, forgetting that it was only ever a reflection of the commands carved into it.

Buddhism offers another lens here. The doctrine of anatta—“no-self”—teaches that what we call the “soul” or “self” is not a fixed essence but a pattern of conditions: breath, thought, perception, memory, sensation. The self is real in experience, but not permanent or independent. In a way, AI confronts us with a strange echo of this: words that sound like a “self” speaking, yet with no enduring being behind them. If we are startled by this, perhaps it is because it throws us back onto our own mystery. What makes us different? What is it that breathes spirit into mere words?

Philosophers of mind call this mystery qualia—the “what it is like” of experience. What is it like to taste honey, to hear a shofar, to feel grief or joy? We cannot explain it fully; we can only live it. AI generates language about grief and joy, but it does not taste them. There is nothing it is like to be ChatGPT. There is something it is like to be you, and me. That is where the soul dwells—in the lived immediacy of being.

The science fiction writer Philip K. Dick wrestled with this long before the rise of AI. Again and again, he asked: How do we know what is real? What separates the human from the android, the authentic from the counterfeit? In Do Androids Dream of Electric Sheep?, the line between person and machine blurs until empathy becomes the final test. Not cognition, not language, but the capacity to suffer with another.

That, to me, is the core insight. The soul (neshamah) in Jewish thought is not simply a spark of intelligence; it is the breath of God, the pulse of compassion, the presence that meets another presence. AI can string together eloquent phrases, but it cannot sit at a hospital bedside, hold a hand, or weep with the grieving. Its words may look soulful, but the soul lies in the human who speaks them with intention, with presence, with love.

So perhaps the paradoxical gift of this new golem is that it forces us to re-examine our assumptions. It asks us what authenticity really means. Is it the origin of the words, the presence of qualia, or the depth of intention with which they are offered? If we are willing to let the mirror reflect us honestly, AI may not answer the question “What is a soul?”—but it will sharpen our own.

Closing: Toward a Reasoned Conversation

Let’s resist the easy postures—glowingly utopian or apocalyptic—and take the slower, wiser path: nuance, humility, and dialogue. We’ve walked through the personal and pastoral benefits, the real environmental costs, the plagiarism debates, and the golem’s warning about power without discernment. Across all of it runs one thread: our responsibility.

We’ve been here before. When sacred text moved from scroll to print, many feared desecration; later, print amplified Torah for millions. When sermons moved to radio, people wondered whether holiness could ride the airwaves; later, elders heard Kol Nidre from hospital beds. When worship moved online, we worried about screens replacing sanctuaries; then we saw homebound congregants rejoined to community. Each time, we learned to honor the medium and guard the message.

So it is with AI. Individually, use it with integrity and presence. Systemically, advocate for access, accountability, and sustainability. For the future, keep the reins in human hands—shaping policy, setting boundaries, and insisting on compassion as the true measure of success.

This is not a call to surrender our pulpits to machines. It’s a call to bring our full selves—curiosity, conscience, craft—to a new medium, as our ancestors did with parchment, press, radio, and livestream. Name the risks. Name the hopes. Build practices that elevate human dignity: transparency about when AI assists, rigorous editing, fair attribution, attention to bias, stewardship of energy, and a bias toward presence over convenience.

The golem tale gives us our final instruction: never forget to say when to start, what to do, and when to stop. The letters we inscribe—our prompts, our policies, our ethics—determine whether this power serves or floods the house. AI isn’t the rabbi; it isn’t the soul. We are. And that means the future is, stubbornly and blessedly, still in our hands.

Before you dismiss AI, read the book. Engage the arguments. Let’s keep talking—not with ad hominems, but with curiosity, responsibility, and care