A talk I gave at the Victoria University of Wellington / Te Herenga Waka seminar series on AI and Society on 9 October 2026.

In the first chapter of the book of Genesis, the first book in both the Jewish and Christian bibles, and the origin story for both religions, God creates Adam and Eve “b’tzelmenu kidmotanu” – in God’s image, after God’s likeness. Twelve verses later, in the second creation story, God breathes life into Adam’s nostrils, causing him to become a “nefesh chaya”, or living soul.

That humans are created in God’s image, and given a soul by the Divine, are core beliefs in the Jewish religion. The soul, the thing that makes us live and breathe as humans, is unique to us, given to us by our creator. Melila Hellner-Eshed at the Shalom Hartmann Institute says that each of us is a pixel in a photograph of the face of the Shekhina, the presence of God in the world, or what Christians would refer to as the Holy Spirit.

In the third chapter of Genesis, we humans start using tools – the first being an unnamed implement that Adam and Eve used to sew fig leaves together to hide their nakedness once they had eaten the forbidden fruit of the tree of knowledge of good and evil.

Judaism embraces the use of new tools and technologies that improve our lives. Under Jewish law, almost anything is permissible that will save a human soul. As an extreme example, even though pigs are not kosher, it is permissible to use a pig’s organ such as a kidney to extend a human life. In his 2023 paper on the subject, Dr John Loike writes: “Judaism teaches that human beings must always consider whether a new technology will better the world or elicit destruction.” Hold that thought.

I’m a software engineer by trade – Computer Science degree from Berkeley in 1981, honours thesis on Natural Language Understanding – which was pretty “out there” in 1981 – and at my age I’ve been cutting code for over 83% of my life. I use Large Language Models (LLMs) like Claude daily in my work. When stored in a computer’s memory, an LLM is just a very, very long number. Numbers do not have a soul. Nor does the computer hardware that runs the LLM have a soul.

There is far more to how humans live and think than can be captured in any mathematical model. We are much more complex than any currently available machine. Our consciousness and spirituality are beyond current capabilities of scientific description. I believe they live in the space opened up by Gödel’s first incompleteness theorem, which basically states that in any complex system that relies on axioms, there are things that are true that you cannot prove.

Back to the creation story, if we are created in God’s image, after God’s likeness, by the reflexive property of likeness, does that also mean that God is like us too?

I was recently struck by commentary in our prayerbooks that we read on Yom Kippur, our Day of Atonement, when we humans ask each other for forgiveness for our wrongdoings to each other, and ask God for forgiveness for our wrongdoings to God. In our prayerbook, Rabbi Brent Chaim Spodek and Ruth Messenger write:

Just as the shadow of a person does whatever that person does, so, too, does the divine do what we do. Divinity is the shadow or reflection of human action. If we save a human life, so too does God; if we decide to end a human life, God does also. At some level, the utterly transcendent divinity is right at hand, for the divinity we hope to worship is a shadow of ourselves, our best parts and our worst. If you want to see God save the innocent, you need to get off the couch and save the innocent. If you want to see God feed the hungry, you need to feed the hungry. If you want to see God stand by while the innocent suffer, all you need to do is stand by and do nothing yourself.

Which brings us to the moral responsibility of Big Tech, the providers of the LLMs we love to use. In the same way that we are created in God’s image, I would argue that the LLMs we use are created in the image of the humans that trained them and prompted them.

When a Grok-driven autonomous weapon kills a person, so do Elon and his shareholders. When ChatGPT advises your teenager to take her own life, so do Sam Altman, OpenAI, and their employees. And when you train or prompt a network of autonomous agents to hack a bank, you share that responsibility, accountability, and culpability with the humans that run your favourite AI platform. The AI itself is just a number running on a machine. It has no real knowledge of good and evil, other than its human-provided training.

Our sages were interested in the possibility of animated, anthropomorphic beings created from inanimate matter. The concept of a “golem”, goes back in Jewish thought to talmudic times. Tractate Sanhedrin, written around the year 400 CE, discusses how Adam started his life as a golem, until he received his soul through the Divine breath. In the 16th century, a legend arose of the “Golem of Prague”, who was created to defend the local Jewish community from antisemitic attacks, but fell in love and became a violent monster. And there are many other concurrent golem stories, including the Golem of Chelm, and the Golem of Vilno. Our sages knew that such a being without a soul and its consequent accountability to the Divine could be very dangerous.

For all of that, I’m an optimist. Claude Code has enabled me to do in an hour what used to take me a week, and made possible tasks that I would have considered unfeasible only a couple of years ago. And I can honestly say that I’ve learned a lot from AI, and AI has made me a better engineer.

I believe we’ll overcome the societal challenges presented to us by rapid technological advancement.

That is, if we can maintain control over our golems and the humans that build them. And if we stand by and do nothing, so will God.