What makes
an AI Jewish?
A corpus of texts, a way of interpreting them, a set of commitments, a relationship to a community: which of these shapes an AI’s behavior, and how do they work together?
Texts & traditionsAn AI research program
What should
intelligence
become?
Gilgul explores how Jewish texts, traditions of argument, and human relationships can shape the next generation of artificial intelligence.
Explore the inquiryJewish thought. Artificial intelligence. Human becoming.01 / An open question
The inquiry
Knowing more is only
part of the question.
As AI becomes more capable, the question of how it should relate to us becomes more urgent.
What should it learn from a tradition? How should it reason when values conflict? Can it challenge us in ways that help us grow?
Gilgul brings Jewish thought into this inquiry: a tradition in which learning is a relationship with texts, with other people, and with the person one might become.
The research agenda
Three questions.
A shared field of inquiry.
What makes
How should
an AI reason?
We investigate how interpretive methods drawn from Jewish learning influence the way models approach moral questions, weigh competing claims, and revise an answer.
Methods & evaluationWho do we
become with AI?
An AI can affirm what we already believe. Could it also help us examine our commitments and grow toward the people we aspire to be? This asks as much of the relationship as it does of the model.
Relationships & growthOur approach
Close reading.
Clear questions.
Experiments that test them.
Our program connects philosophical inquiry with empirical work on AI behavior. We study what changes when a model’s source texts, reasoning instructions, and role in a conversation change.
We distinguish the ideas we want to explore from the claims experiments can support. The research is developing; its questions remain open.
About Gilgul
An old tradition.
A new turn.
Gilgul is a research program taking shape through conversations between Zohar Atkins and Ben Wolfson about Jewish thought and artificial intelligence.
We are developing essays, research questions, and experiments on moral reasoning and human–AI relationships, with room for dialogue across intellectual and religious traditions.
Work in progress
Our first essays and research updates are in development. This is the beginning of the inquiry.