“The words of the mouth are more potent than those of the heart. For it is from the mouth that we utter blessings and curses, and from the mouth that we go out to war and peace. Therefore, one should be careful with one’s mouth and speak only words of truth and kindness.” (Talmud, Shabbat 104b)
What beautiful, sermonic words, perfect for Yom Kippur. I could proceed from here, exploring the importance of how we use our words for good, something that can so easily become a weapon of harm. Problem is, it isn’t from the Talmud. It is a text that RSY Netzer workers found last year when they tried to put together a source sheet with the help of Chat GPT. As Rabbi Mark explored on Rosh Hashanah, it isn’t always the most reliable source of information. But the above, shared with colleagues by Rabbi Deborah Blausten whom the movement workers consulted with, was totally made up, and also rather believable! As Rabbi Blausten herself said at the time;
it has given them the ‘Jewish wisdom’ they thought they wanted, it’s totally plausible and yet totally made up. It’s arguably a reflection of mistaking AI for search, but it certainly blurs the boundaries between fact and fiction. On one hand it’s fake, on the other hand, it’s made up from the sum total of assembled (albeit translated) information available to the AI so it’s in this ambiguous and confusing category of both having nothing to do with our textual tradition and also everything to do with it.
Chat GPT is an Alternative Intelligence that generates information upon request, anything from a sermon to a speech, fake news to impersonation. The jury seems to be out, however, on whether I might be replaced by an AI robot. I won’t ask for a show of hands as to who would find that preferable, but back in June, a Church in Bavaria gave it a try[1]. A theologian and philosopher, Jonas Simmerlein, designed the AI Priest for a service held at a Protestant conference.
“I told the artificial intelligence, “‘We are at the church congress, you are a preacher … what would a church service look like?’ In his ChatGPT prompt, he asked for the inclusion of psalms, prayers, and a blessing at the end. ‘You end up with a pretty solid church service’ Simmerlein said”.
As with every service, not least here, there were mixed reactions from the congregation. The main response, however, seems to have been that the service lacked heart. I bet you would all miss my mistakes if a monotone robot were in my place. This seems to have been affirmed by a study in Japan which had visitors to a Buddhist temple, Taoist and Christian services, listen to a sermon delivered by a robot priest – a million dollar android with a human like face that was programmed to give 25 minute talks. You’ll be pleased to know I am programmed to be briefer. Those congregants who heard the preacher were asked about making a donation to the temple or church as part of the study (something we are less averse to) and in all cases donations were significantly less likely to be received when the robot preached than when a real person had.
Perhaps it is our sense of the value of something that takes effort and work and has been written especially for the occasion, as opposed to something generated by a generic summation of internet knowledge. But perhaps it is our desire to connect to one another and to the humanity around us that produced these results. AI begs many questions of us, but for me the most pressing one is about our humanity. The next 25 hours we will be travelling together through a landscape that asks us to consider what it means to be human, and what it will look like in the coming year. But when it comes to AI, are we more like God the creator?
As creators, have we thought through the implications of what we have made (we might ask the same of God in creating us!) Max Tegmark, a Cosmologist and Professor of Physics at the Massachusetts Institute of Technology – MIT, ponders many of the questions raised by the advent of AI in his international bestseller Life 3.0. Many of us fear that AI will become evil, or act evilly in the world to our detriment. Tegmark argues it is not a question of evil, but of whether the goals of AI will align with our goals. Today we ask, what are our goals, and perhaps, do they align with those of our creator, or, I suspect more crucially, with those of creation. While we worry about the goals of AI, we also have to take account of the fact that too often our own goals do not benefit us, or the world around us, sometimes in the short term, and more often in the long term.
We will confess as individuals and as a community over the coming day. And our goals as both individuals and communities cannot be separated – as a Midrash[2] famously teaches (and this one really is from our texts, not from chat GPT), a passenger in a boat began drilling under his seat. His fellow passengers were appalled ‘ What on earth do you think you’re doing?’ they demanded – ‘Don’t worry!’ He retorted ‘I’m only drilling under my seat!’ We are all in the same boat, and I for one would love to keep sailing with you all. The reality is, that AI will be sailing with us too, and while some may be fearful of it, it also opens up incredible opportunity and advances for humanity, if we can align our goals.
Stuart Russell, an AI scientist, addresses the challenge and writes that “The difficult part, of course, is that our objectives are in us (all eight billion of us, in all our glorious variety) and not in the machines. It is, nonetheless, possible to build machines that are beneficial in exactly this sense. Inevitably, these machines will be uncertain about our objectives – after all, we are uncertain about them ourselves”[3]
This statement fascinates me in part because of my own personal theology. The Torah reading we will hear tomorrow morning describes a conversation between Moses and God during which Moses wants to see all that God is. God says No. No one can see all of God and live, our brains just can’t manage the magnitude. While they were atop mount Sinai having a chat, the Children of Israel were gathered around the mountain. In midrashim that deal with the giving of Torah, we are told that God appeared to each person that was there just as they were able to see God. God appeared in some 600,000 different ways – able to speak to each individual and what they were able to conceive of. This has always made me wonder if being made in God’s image isn’t about each of us as individuals, but about our immense human diversity. Each person has a glimpse of the whole, no one knows or sees it all (perhaps that is part of our fear with the amount of knowledge AI can potentially access or hold at any given time). We can learn from everyone else’s glimpses, and perhaps we can only ever come close to understanding what God is when we realise it is the breadth and depth of human diversity that is betzelem Elohim – in the image of God.
When it comes to AI, are we the God-like creators? If we are, AI potentially seems to hold the ability to understand us far better than we can understand our creator. Though I have wondered if God struggles to understand us just as much. When Eliana was about 8 weeks old. I had what my sleep deprived brain thought might be an insight into God the parent – Avinu Malkeinu. We had struggled and battled to conceive for 6 years, and we were finally holding this amazing piece of creation. And she wouldn’t stop crying. She had been fed. She had been changed. She was being rocked and cuddled… and she wouldn’t stop crying. We had finally, with some help, made our own little corner of creation. And we didn’t have a clue what she needed or why she was crying and how to stop it. Perhaps God is equally confused by our inability to figure things out, to prioritise the right thing, to decide on the right goals, and to look after one another. Perhaps we are as baffling to our creator as baby Eliana was to us. As she’s here I won’t say that she remains somewhat baffling at times!
But as we are the creators of AI, we have potentially created something that can know everything we know and much more. But can it understand us? It can imitate us, be in our image if we make it so, but there is something fundamental about what it means to be human. Sometimes I think it lies in the understanding. Sometimes I think it lies in the not understanding, in the wrestling we face the world with. We get it wrong, a lot. That is why we have this period in the Jewish calendar. We will wrestle with ourselves over the next day, and then we will go back to wrestling with the world, trying to understand, trying to do our best. Sometimes being our worst. We are a mysterious lot, and surprisingly, the future of AI is also still somewhat mysterious. We have the opportunity to influence and decide on the future as a society, or to ignore it and see what happens. This is also what Yom Kippur offers us – the opportunity to set our goals, to decide on what or who we want to be over the coming year – even though life may mean the path isn’t as planned. We have an opportunity over the coming hours to think ahead, to try to work out who we want to be in the year to come, what goals and objectives do we want to set for ourselves, and do they align with what is good for anyone beyond ourselves.
The consensus of the specialists on AI I spent time reading this summer is that there is still time – time to establish the ethics and the boundaries and the limits of AI. Time to ensure we align the goals of this mind blowing technology with our own. But we also need time to decide on our own goals.
Israeli academic and philosopher Yuval Noah Harari argues earlier this year that the real danger from AI comes in its ability to master language, and particularly story telling – creating narratives and a sense of intimacy with humans that can change behaviours and manipulate civilisations[4]. He writes that ‘Fear of ai has haunted humankind for only the past few decades. But for thousands of years humans have been haunted by a much deeper fear. We have always appreciated the power of stories and images to manipulate our minds and to create illusions”. Over the coming day let us face our fears, and decide what the story we want to write should say. Our fears should not replace our hopes, and as we decide on the story line we want to see unfold, let it be a story that considers the world beyond ourselves. There is a lot to do, but we have to be part of the conversation as we walk forward!
May we be blessed in the coming year with the ability to see a story of hope and life unfolding before us, and a community in which we can live that story with support and good grace. Cain yehi ratzon, may this be God’s will, venomar, Amen.
[1] AI-powered church service in Germany draws a large crowd | Ars Technica
[2] Vayikra Rabbah 4:6
[3] Human Compatible 2019
[4] The Economist, April 2023, ‘AI has hacked the operating system of human civilisation’.