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 it on your High Holy Day sermons.
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........
