Rabbis Have a Moral Imperative to Use AI
Why Rabbis Should Use AI This High Holy Day Season
Every year around this time, rabbis begin disappearing.
We disappear into books, notes, drafts, abandoned openings, half-finished sermons, deleted paragraphs, and the peculiar anxiety that comes with knowing that more people will hear us speak on Rosh Hashanah and Yom Kippur than at any other point in the year.
The High Holy Day sermon carries an almost mythic weight in the rabbinate.
It has to be timely. Deep. Personal. Jewish. Relevant. Beautifully written. Spiritually meaningful. Somehow new.
And it has to appeal to everyone throughout our diverse congregations at the same time (without upsetting anyone.)
So we sit at our computers for hours trying to make all of that happen.
I want to suggest something that will make some of my colleagues uncomfortable:
Use AI.
Use it on your High Holy Day sermons.
I do.
And I should say what I mean by that, because I am using AI in a way that goes well beyond typing in a topic and asking for a sermon.
Over the past months I have spent hours arguing with AI about authorship, attribution, b’shem omro, environmental ethics, neurodivergence, the limits of AI as a thought partner, consciousness, and what responsible use by clergy might actually look like. Those conversations have become a kind of external intellectual memory. When I ask AI now to help me think about a new question, I am often asking it to synthesize a body of thinking that I have already developed over time.
That feels very different to me from asking a machine to have the idea for me.
A rabbi might begin with a page of disconnected thoughts and ask AI to help reveal the threads.
They might describe something they have been wrestling with and ask what questions they have failed to ask.
They might paste in a draft and ask where the argument loses coherence.
They might ask for ten possible structures and use none of them, while realizing that one of them has revealed the actual structure.
They might vaguely remember a teaching from the Talmud and ask AI to help locate it. Then they go to the text and check.
They might talk through an idea for twenty minutes until they finally hear themselves say the sentence they had been looking for.
The sermon still needs to emerge from a real encounter with Torah, community, theology, pastoral experience, doubt, failure, hope, and whatever this season is asking us to understand.
AI becomes part of the workshop.
For neurodivergent clergy, this can matter enormously.
Some of us can generate ideas all day and have tremendous difficulty organizing them. We can see connections everywhere and struggle to choose a sequence. We can know exactly what we want to say and stare at an empty page because turning thought into structure requires a different set of cognitive skills.
AI can provide executive-function scaffolding.
A tool that helps someone organize their own thoughts does not make those thoughts less their own.
And then there is something larger at stake.
A congregant called me this week to tell me that his wife had been placed on hospice.
That is rabbinic work too.
Actually, that is the work.
A rabbi should have time to call back. To sit with the family. To show up without watching the clock because three sermons still need to be finished.
September arrives, and rabbis are simultaneously preparing sermons, services, Torah, music, logistics, pastoral conversations, funerals, hospital visits, family crises, staff meetings, congregational conflicts, and everything else that does not politely stop because Rosh Hashanah is coming.
Suppose AI saves five hours on a sermon.
Those five hours are part of the ethical calculation.
Maybe they are spent with that family.
Maybe they are spent calling someone who is having a terrible year.
Maybe they are spent outside for an hour, actually experiencing late summer turning toward fall.
Maybe they are spent in prayer.
There is a teaching at the end of Jerusalem Talmud, Kiddushin 4:12:3 (66b, halakhah 9) that I love:
עתיד אדם ליתן דין וחשבון על כל מה שראת עינו ולא אכל
A person will ultimately have to give an accounting for what their eyes saw and did not eat.
The passage comes in a striking context. The Mishnah has just lifted Torah study above a whole range of other mitzvot, describing it as equal to them all. Then the Yerushalmi gives us this teaching about being held accountable for the good things in the world that we saw and failed to experience.
I hear a tension there that matters deeply to me: do not become so absorbed in the life of the mind, even in Torah, that you fail to live in the world God has given you.
Rabbi Elazar takes the teaching seriously enough to set aside money so that each year he can taste the different things available to him.
That feels especially relevant during the High Holy Day season. We can become so consumed with preparing to speak about life that we fail to live the life we are preparing to speak about.
We write about awe while staring at a laptop.
We preach presence after spending three weeks frantic.
We tell people to notice the world, repair their relationships, make room for joy, sit with mortality, and listen for the still, small voice.
And I wonder whether AI belongs inside that same theological question.
If human beings have developed a tool that can help clergy spend less time on structure, searching, or administrative cognitive labor, and more time with a family whose loved one has entered hospice, more time in prayer, more time actually encountering the season they are preparing to preach about, then perhaps there is also a din v’cheshbon in refusing to use it.
I do not mean that every new technology is therefore God’s will, or that usefulness settles its ethics. It doesn’t.
I mean that the religious question cannot only be, “What danger does this tool present?”
It also has to include: “What capacity has become available to us, and what good are we failing to do if we refuse to use it thoughtfully?”
At that point, the ethical question begins to look different.
We spend a great deal of time asking whether rabbis may use AI.
I think there are circumstances in which we should.
If this technology can make rabbis more present to congregants, more available for pastoral care, more able to engage personally with the season, or more capable of expressing what they genuinely mean, then responsible use can become part of responsible rabbinic practice.
That carries obligations.
Check your sources.
If AI tells you that Rashi said something, read Rashi.
If it produces a beautiful Hasidic teaching and cannot tell you where it comes from, don’t preach it as inherited Torah.
If a paragraph says something you do not believe, delete it.
If you cannot explain why a sentence is in your sermon, it probably should not be there.
And when AI has played a meaningful role in creating a piece, disclosure is easy enough. There is no need to pretend we work without tools.
We have always used tools.
Books. Concordances. Search engines. Editors. Teachers. Chavruta partners. Dictation. Spellcheck. Google Docs. Other people reading our drafts and saying, “I have no idea what you mean here.”
AI is a much more powerful tool, which means it requires more discernment.
The Jewish image I keep returning to is the Golem.
The Golem can do extraordinary things. It can help. It can protect. It can also grow beyond the wisdom of the person who set it in motion.
Its power demands responsibility from the human being who gives it instructions.
In the words of Spider-Man’s uncle, “with great power comes great responsibility.”
That feels closer to our actual situation than either utopian enthusiasm or categorical refusal.
The danger is real.
So is the possibility.
I want rabbis writing sermons whose words they mean.
I want us checking our Torah.
I want us wrestling with what this moment asks of us.
I also want us to remember that the purpose of the rabbinate is larger than producing excellent prose.
When rabbis stand before their congregations on the High Holy Days, people are not gathering to admire their ability to arrange sentences without assistance.
They are asking them to be there.
To have wrestled.
To have listened.
To have something worth saying.
If AI can help us get closer to that, I think that not only should we use it, but that we have a moral imperative to do so.
Full disclosure: I used AI in writing this piece. I used it as a thought partner, drawing on many hours of previous conversations in which I had worked through these questions, and to help me organize, structure, challenge, and edit the argument. The ideas, theological claims, judgments, and final choices are mine. I checked the sources myself and revised the finished piece before publication.
