I did another podcast interview with Divia Eden of Mutuals. My friend Imam Ammar Amonette, a respected and well-connected Muslim interfaith leader, joined along. In this podcast, we delved into how Ammar and I have inspired each other and validated each other’s thinking around religion and AI.
Topics discussed include:
the Parliament of World Religions, a conference filled with religious leaders whose interpretations of religion I found both sensible and inspiring
Ammar's longtime interest in formalizing the universal principles that underlie the foundations of religion
the concepts of religious pluralism ("my religion is one of many possible paths to truth and salvation") and religious exclusivism ("my religion is the only path to truth and salvation")
how religion may be relevant for the technical and political problems associated with getting good outcomes with AI
how a positive AI singularity could align with religious prophecies about the end times
Table of contents
[0:30] Alex introduces Ammar, and describes how Ammar inspired him
[4:02] Ammar describes meeting Alex, and how Alex inspired him
[7:48] Shockingly reasonable interpretations of religion at the Parliament of World Religions
[9:51] Shockingly ethical people at the Parliament of World Religions
[10:52] Validation of Alex's speculations about what religion could be
[11:58] Catholic priests validating Alex's experience of Jesus on ayahuasca
[12:40] Ammar describes the Parliament of World Religions
[15:08] How Ammar decided to attend the Parliament of World Religions
[21:21] Ammar programmed cathode ray tube computers back in the day
[23:11] The foundations of religion may already have been formalized by Chris Langan
[25:10] Historical theological discourse on the religious universals
[30:52] Humanism lacks the metaphysical foundations of religion
[33:06] Building off of existing religions rather than starting from scratch
[35:56] Introducing religious exclusivism, inclusivism, and pluralism
[36:47] Religious exclusivism is not more traditional than religious inclusivism
[39:49] AI and the religious universals
[45:04] AIs discerning what's true and what's ethical
[50:24] Correctly stitching together LLMs' implicit understandings of the universals might be hard
[54:13] Competitive incentives for AI bullshit detection
[55:36] AI mediators that can parallelize group conversations
[59:45] Some challenges of grounding epistemology (Twitter Community Notes & Wikipedia)
[1:02:04] AIs that can discern and differentially upvote good content
[1:04:14] Religious leaders' skepticism about positive AI outcomes
[1:05:41] AI antichrists and messiahs
[1:07:54] Every action is accounted for, and matters
[1:10:53] Will right make might?
[1:13:41] Interpreting Jesus on the sheep and the goats
[1:15:54] Collaborating with skeptics among religious leaders and ordinary people
[0:00] Intro
Divia: Okay, so today I'm here with Alex Zhu, who's been on the podcast before. And Alex, your friend Ammar, who's joining us for the first time. So yeah, for our listeners, we had a long podcast with Alex a couple podcasts ago, we've known each other for a long time… Alex is interested in AI and among other things figuring out how to connect the spiritual leaders of the world with the AI discussion. Does that seem fair?
Alex: Yes.
Divia: Okay. So now do you want to introduce Ammar, Alex? And say how you guys met?
[0:30] Alex introduces Ammar, and describes how Ammar inspired him
Alex: Yeah. Ammar is an imam I met at the Parliament of World Religions. I'd been doing all this theorizing and speculation about how religion, properly interpreted, could actually be of crucial value for thinking clearly about good outcomes with AI. All my thinking sort of felt theoretical and speculative, and it was never really clear to me whether there would be any actual serious religious leaders who would be able to make sense of my visions or think any of it was remotely reasonable.
Ammar was the first religious leader I'd met who I'd felt really got where I was coming from, to the point where when I would have doubts about what I was trying to do, Ammar would give me encouragement and support, and I'd have more faith in this direction and this vision. I basically felt like an entire line of thinking got validated and changed from pie-in-the-sky-theorizing to something tangible and real when I met Ammar and found that basically we could get on the same page about all the things that I was thinking about.
Part of what made this so validating for me is that Ammar isn't just some random imam at some random mosque. He has held a number of pretty substantial leadership positions in the Muslim community in the United States. Like, Ammar, I think you were the head imam of Denver, Colorado for a while and now you're the head imam of Richmond, Virginia. And not only that, but you're also very well connected in the Muslim world. You know some of the imams at the Holy Mosque in Mecca, for example, and you're on the World Council of Muslims for Interfaith Relations. There was a sense of like, this isn't just some random Muslim who's thinking on his own and being like, "oh, yeah, this makes sense." But Ammar is very, very deeply, deeply familiar with the history of Islam and the theology of Islam, and is able to reflect back a bunch of the ideas I say in terms of his own personal, very deep understanding of Islam.
And on top of that, he's helped me deepen my connection with the religion of Islam. I wouldn't call myself a formal adherent of the religion, but I take a lot of inspiration from the religion as I interpret it. I take a lot of inspiration from the Prophet Muhammad. A lot of decisions and difficult situations I encounter in my life, specifically around politics and navigating tricky interpersonal conflicts where there are big egos involved that I feel like don't have my best interests at heart… Islam is the spiritual tradition that I feel like has some of the most concrete illustrations of what it looks like to skillfully navigate those situations. And sometimes I talk to Ammar for personal moral advice as well around some of these situations.
I think there's more I could say, but I think I'll leave it at that for now. Ammar, I'd love to hear if there's anything you think I might've missed, and also about what it was like for you to meet me and how you think about having met me.
[4:02] Ammar describes meeting Alex, and how Alex inspired him
Ammar: Well, it's very humbling to hear Alex say the things he said. I hope I would be able to live up to some of that. I feel, yes, we met about a year ago at the Parliament of World Religions in Chicago, which itself was a very inspirational experience for me personally, and I met a lot of great people. But there was one person who was unique and still stands out.
And so, Alex, the fact that he would say the things that he has said is beyond my expectations, because I felt like I'm the one who's benefited. So I've been interested in some of the same things that he's been interested in, but not that he simply has validated things that I already thought but really inspired me and really gives me hope for a future direction. I am pretty sure his leadership is going to show positive results for society far beyond maybe my lifetime. It was a very important meeting for me. It was kind of serendipitous. It was something… maybe I was in some sense looking for such a meeting to happen, but I couldn't have made that happen. I wouldn't have known where to look...
Alex: Ammar, I just want to share for our listeners that I think the very first thing we connected on was when I was talking to someone else about mathematical metaphysics, and then you chimed in mentioning Gödel's ontological proof for the existence of God. And then...
Divia: That was kind of what made the initial connection?
Alex: Yeah. I had a sense of like, oh, here's someone who maybe actually gets what...
Divia: I forget what we covered last time, Alex, or for people who haven't listened, but you do have a pretty serious math background and you did IMO stuff, is that right?
Alex: Math olympiad stuff. I never made the international team, but I learned that people don't understand the difference and it doesn't matter.
Divia: I'll try to remember that next time.
Alex: It's okay.
Ammar: Alex was sitting in our Muslim faith room there at the location, McCormick Center, and he was talking to a mufti, an important Muslim religious leader originally from India, but who's a mufti in England, who is more or less much my senior in years who also has a strong mathematics and science background, computer background, but he's a specialist in Islamic law. And so Alex was talking about these subjects, which are things that I've been thinking about for many years, but I am not the specialist by far. So I don't remember how long we were able to talk there at that particular location. And then I think I only met you one time afterwards, but every time I would be going to a meeting or a lecture, I would see Alex ahead of me walking by on his way somewhere else...
Divia: Your attention was with Alex?
Ammar: It was amazing because I would just see him from a distance on his way to something else, and then I'd be on my way to something else. But he did come by my booth, our booth for the World Muslim Interfaith Council. He came by my booth before we left at the end of the conference. And that was the second time I think that I talked to you face to face… So it was a very kind of inspiring… somebody might say a lot of coincidence… but we've been able to talk through video conferencing or phone since, but we haven't actually met in person since that time.
Alex: Yeah, hasn't happened yet.
Ammar: No.
[7:48] Shockingly reasonable interpretations of religion at the Parliament of World Religions
Divia: Yeah. Thanks for that background. And so just also for the benefit of our listeners, can you guys talk about the Parliament of World Religions and what that is?
Alex: I'd love to hear your description of it, Ammar. I'd like to start with mine. An analogy I often make is that the best scientists around are like, "Yeah, science is great, the scientific method is great, but science as an institution has a lot of problems, like… there's this replication crisis where most published studies are false, and the public doesn't really know about this, and they just have this association of science as this big conglomerate that they're supposed to believe in, and actually it's way more messy and complicated than that.
"And also, even though this is what's actually true, most scientists aren't even really tuned in to this view, but we are, and we think it's a real shame and that something really significant and major needs to happen with the institution of science for science to really live up to what it was really meant to be."
And a sense I had at the Parliament of World Religions is… I was an atheist all my life because of a bunch of standard obvious critiques of religion, like… they generally all made exclusive claims to truth and salvation that didn't seem very plausible to me. And I would just keep meeting religious leaders from various different faiths just being like, "Oh, yeah, no, no, no. We basically just fully agree with all those critiques. A lot of those interpretations are naive. Yeah, religion as commonly practiced does seem to serve division more than unity. Yeah, a lot of these common interpretations of these doctrines don't really make sense. Yeah, we agree that it doesn't really make sense that if you were born outside of my religion that means you're going to hell just because you didn't have the good luck of being exposed to the religion."
And so it was really refreshing and validating to meet so many religious leaders who actually thought about religion in ways that seemed not only sensible to me, but also inspiring.1
[9:51] Shockingly ethical people at the Parliament of World Religions
Alex: One other thing that stuck out to me was… if I meet a random person on the street who identifies as a Christian, I don't really expect them to be walking the walk of Christianity that much, but at the Parliament of World Religions, I was struck by how many of the people I met who really seemed to take seriously walking the walk of their religion...
Divia: Like they seem very personally virtuous in the ways that you would expect? Is it that?
Alex: Yeah, in a way that I had previously not actually associated with religion before, despite religion saying that's what that's about. And then here was a place where I actually made that association, and it inspired a lot of faith in me that the people interested in reconciling the religions, and who are interpreting the religions in a way that leads them to actually walk the walk and respect other religions and try to reconcile with them, seem like a really positive force for good in the world that… I think is not as popular yet or powerful yet as I would personally like, but it gives me hope about the future.
[10:52] Validation of Alex's speculations about what religion could be
Divia: So that makes sense to me because something you're particularly good at and practiced at is looking especially at religious traditions, but also at thinkers, and… I don't know, there are a lot of reasons why people would dismiss them, but I think you're pretty good at seeing to the heart of... if someone were saying this in good faith and they meant something really deep by it, what would they be saying? And so you'd sort of done your own personal study of religions and you're kind of like, okay, maybe this is what they could be saying in theory...
Alex: That's right...
Divia: …but you weren't really sure. You weren't really how much that was just your perspective.
Alex: That's right...
Divia: And then you went there and you were like, oh, wait a minute, not only is this a real thing that some people think, but actually it seems like some of the really top people across world religions not only think it, but are looking for the other ones who also think it. And there's sort of a community of the thing that you had imagined. And so, I think it makes sense to me that that would be very validating for you because you see this potential and you're like, "No, no, it's not potential. This is actually what's happening. It's just very different from the popular version that you would read in the headlines and newspapers or whatever."
[11:58] Catholic priests validating Alex's experience of Jesus on ayahuasca
Alex: That's right. Ammar was one such person I met who it felt very validating to meet in this way. I remember there were also some Catholic priests who I told about an experience of Jesus I had on ayahuasca. As I recounted the experience to them they basically were just enraptured and were just like, "Alex, this sounds like, it sounds like you had an extremely powerful experience, and you're very fortunate to have had that experience. And whatever it is you're doing, you should keep doing it." Which was one of the last things I was expecting to hear from a Catholic priest about an experience of Jesus on ayahuasca as someone who doesn't identify as a Christian.
[12:40] Ammar describes the Parliament of World Religions
Divia: So thanks so much for that description. Ammar, can we also hear from you how you would describe the Parliament of World Religions?
Ammar: Parliament of World Religions. The first Parliament was held at the World Exposition in Chicago in 1893. So that was like a world's fair, and that was the first time that religious leaders were brought from all over the world for a conference to speak to each other on equal terms. And you have to realize that was at the height of colonialism, domination of the world by, for example, evangelization that went hand in hand with colonialism, and yet there were people who were farsighted enough to see, and had a vision for a future world, where every religion was respected and the different voices were respected.
So it continued from time to time, but it was interrupted by the two world wars, depression, Cold War, and… for the centennial year they revived it in Chicago. In fact, the World Muslim Council for Interfaith Relations was one of those constituent groups that made it happen in 1993. And the modern Parliament of World Religions has a document that was written by the late Swiss theologian, Hans Küng, and it's called Towards a Global Ethic.
Divia: Towards a Global Ethic, okay.
Ammar: Yes, and it takes the basic principles, universal principles of all the faiths, and it invites faith leaders, politicians, and people of no faith at all to sign. So we witnessed when we were there the Archbishop of Chicago representing the Pope signed, finally, after many years. So that was actually a momentous occasion, and it envisions that humanity... We live in a global world. Technological advances make it impossible that we continue down the road of violent conflict and competition between cultures, civilizations, and religion. And so it's a vision of how we can unite humanity through those universal principles. And I should probably stop there…
[15:08] How Ammar decided to attend the Parliament of World Religions
Divia: One question I have is, how did you get involved with the Parliament for World Religions?
Ammar: Well, I had participated during the pandemic. It was all online, and I did try to participate, but it was difficult under those circumstances to get the full benefit. And so I was invited by an Islamic organization that was one of the original constituent organizations, and they invited me. And so I wasn't sure if I really wanted to go or not.
And as I said, I had this sort of inspirational dream that I was going to meet some significant person there. And yes, I think that the person was maybe of a Buddhist background, maybe a Chinese background, but that wasn't the only reason. But there was something that said, I'm going to meet some people there who maybe are going to lead me into a new direction of thought. And that is exactly what happened. I met many, many great people, many great interactions personally, and of course, Alex is the person!
Alex: I'm humbled Ammar, and I hope to be able to live up to what you've been saying about me.
Ammar: Yes, you will.
[16:28] Ammar's longtime interest in formalizing the universal principles that underlie the foundations of religion
Divia: Okay, so if we maybe get ready to dive in more, can you guys both tell me how are you thinking about artificial intelligence and the future of artificial intelligence and what religion has to offer in thinking about the future of artificial intelligence?
Alex: Before we go into that, there's a piece of background I want to go into, which I think is relevant for your question…
Ammar, you were telling me, I think in maybe our first conversation, that my vision of mathematical metaphysics that I was describing, you'd basically had yourself many decades ago, thinking basically like, "Wow, all these religions… it seems like there's a there there that they're really all talking about, and it seems like it should be possible in principle for someone to actually just figure out what that there there is and articulate it, and that would be such a momentous and valuable thing for humanity."
And then my understanding is that you shelved that idea for a while because it wasn't actionable for you, really. And part of what made our connection so deep, so fast was because this old dream that you had got rekindled, as you overheard me speak. And then part of what made it so deep for me was seeing how deeply impacted you were by this and how someone deeply enmeshed in the world of religion actually has a very similar vision to mine, even though we come from extremely different backgrounds.
Ammar: First of all, that's a universal vision of Islam. And I think it's something similarly shared in other faiths.
Divia: The vision where someone will be able to articulate…?
Ammar: No, not that particular, but that the oneness of God leads to the oneness of humanity, and that the different religious teachers, historically, have all shared that same spirit and vision, but with particular differences culturally between different world faiths. And the particular thing that is of interest to me is in the foundations of religion. We talk about those universal principles, and I was convinced that those could be… formalized in symbolic logic or mathematics.
And I think that there's something universal there that's found in every faith. And it's in the Western religions, in Abrahamic religions such as Islam. It's based on the word. And so our language comes from the structure of our mind. So there's something universal, the human mind, that produces all, obviously a multitude of languages, but those languages all are based on certain grammar and syntax. It's a logic, but we misunderstand and misinterpret one another. And so it's part of the Islamic ideal that we could bring people to awareness of that unity despite our apparent differences.
And so it was my (maybe naive) assumption that it would be doable. But, yes, I didn't have the tools, necessarily, to do that myself, though I did try! So I was a student in Mecca, Saudi Arabia. That's where I went to school. So whenever I wasn't studying for exams, I was working on this little problem.
Divia: And how did you work on it?
Ammar: Well, first of all, try to figure out what, there's so many challenges and...
Divia: Make sense...
Ammar: There are greater minds than my own that have worked on those. And so one of them he mentioned, was the mathematician Gödel who was there at the Princeton Institute for Advanced Study there along with Einstein. They would take a walk around that lake every day there in Princeton. And so he shared a sort of similar theology – of course, a very significant mathematician – but he also shared philosophy and theology that was similar to some of these ideas. So I found that was very, the idea that, "Oh, somebody obviously very much smarter than myself came up with this idea a long time ago." So that kind of reinforced my belief that this is feasible, but I didn't have the tools. So you would laugh, but I actually had hundreds of index cards... Old folks know what index cards were.
Divia: I'm old enough that I know. I also used to have...
Ammar: So, I would actually spread them out on the floor trying to relate the different concepts and ideas to each other physically through index cards.
Divia: Yeah, no, it's cool for me to think about you working on it.
[21:21] Ammar programmed cathode ray tube computers back in the day
Alex: Yeah. Ammar, you mentioned also that another life direction you were considering at some point was becoming a mathematician if you hadn't become...
Ammar: I studied what's now called coding in high school. I had the chance, I went to a kind of college prep school on a college campus. And so we were able to do computer programming. So we took computer mathematics and things like that, but I had a lot of different interests. And so those old, what we used to call cathode-ray tubes, used to give me headaches as a kid...
Divia: Sorry for the younger people, that's like old TVs were made out of...
Ammar: Yes, that's what we used. We called them CRTs.
Divia: Yes, I remember.
Ammar: And they'd give me headaches. So I said, "Oh, computers give headaches."
Divia: Yeah...
Ammar: And when I went to... I was very fortunate to get a scholarship, kind of full-ride scholarship there to Mecca, and nothing was computerized, nothing. And I wrote out my master's thesis by hand. And what you would do, you would write it by hand and you would take it to a printer and they would print it for you.
Divia: Yeah. So it wasn't exactly an opportunity for you to work with computers at the time?
Ammar: No. It was exclusively traditional religious studies, studying what was written a thousand years ago, which is very beneficial, but wasn't necessarily directly related to this subject.
Divia: Makes sense. And so that has sort of not been top of mind for you for a while at the time that you met Alex?
Ammar: Oh, it's always in my mind and I'm always reading things related to the subject. And of course, obviously I'm not a specialist, so some of those writers are very challenging.
[23:11] The foundations of religion may already have been formalized by Chris Langan
Alex: For the sake of completeness, I do want to mention for our listeners that I've actually become pretty convinced that there actually has already been a person who figured out how to formalize… basically, in some sense, has solved mathematical metaphysics and found a formalization for the foundations of theology, basically. And his name is Chris Langan, and his theory is the Cognitive Theoretic Model of the Universe. And if only the rest of the world could understand the concept of what he was saying, things could look very different in the world, which is where I'm currently directing a lot of my attention and energy toward… just helping more people understand the details of Chris's theory. This is something I've discussed with you somewhat, Ammar.
Divia: Yeah. Do you have thoughts on Chris's work?
Ammar: Well, I'm not going to be speaking on him per se, but I see a lot of similar ideas there. And I'm not the person who can answer whether he's actually solved all of these problems or not. And I think that Alex is probably onto something there. And so I'm not in any way objecting to any of his works, but I'm probably not the person who's qualified to judge the correctness or lack of correctness of any particular argument. But I'm very, very interested in many people who've spoken in similar ways, though none of them have an actual complete solution. And so maybe he has that complete solution, but I'm not the one who could judge it.
Alex: I just want to mention for full disclosure that I don't understand his theory myself fully yet, and so I can't personally verify it either. But at this point, given how much ground I've covered in my conversations with him, I would say that I think it's more likely than not that his claims are true. That's just where I'm personally at.
[25:10] Historical theological discourse on the religious universals
Ammar: So the theologians of the Middle Ages were well aware of these problems and were themselves working on solutions and codifying solutions in their way.
Divia: And when you say these problems, can you expand on what you mean there?
Ammar: The problem of the universals that unite us as opposed to the particulars that divide us. So this is a very important theme of the Quran, the Islamic scripture. And while, obviously, many Muslims have been concerned largely with their own personal practice and their own particular faith practice, there have been scholars throughout the ages who have been very interested in what unites everyone.
And so the basic Islamic belief is that every faith is based on these universals. it's not something that is simply a construct that I've come up with or someone else, but it's something that has been agreed upon for centuries and centuries. And I think that... yes…
Divia: So this might be a hard thing to ask, but the universals… would you be up for trying to put those in your own words, what the universals are?
Ammar: Well, some of those you can see for example in the Global Ethic, that based on faith in God, and God is good and just, and that we treat each other with goodness and justice as we wish to be treated, we treat others. And the sanctity of human life. But there are other principles that are very more practical than simply those things and very specific. So for example, encouraging growth and development of science, technology, of society, civilization, and making sure that while you maximize the benefit for everyone, you include as many people as possible, if not 100% of people, to maximize the benefit of civilization for everybody.
Alex: Can I ask a clarifying question? Would you consider among the universals the idea that everyone, at some point, perhaps in their transition to death, has to account for the consequences of all their actions?
Ammar: Of course, that's one of the very important agreed upon, I think, universals is that human life has a meaning and value, and that our decisions, our actions and choices are significant and maybe go beyond our own life, and affect others, affect the world, positively or negatively, and so we have to take responsibility for those.
And in taking responsibility, we can also achieve forgiveness. We can right wrongs and we can purify our own soul, our conscience, as well as making the world a better place. A lot of Muslims as well as non-Muslims understand… well, criminals are punished. There might be corporal punishment or capital punishment for criminals. And similar things are in, for example, in the Old Testament and the Bible. But one of those principles is that we listen to the word and then we follow the best interpretation. So if we have a chance to pardon or if we can choose the lesser punishment possible, if there's a crime that's been committed, if that punishment won't result in a greater harm...
Alex: Ammar, and just to reorient, is this all part of your conception of what the universals are...?
Ammar: All of the religions, for example, and I think we can agree that every society bases itself on justice. So if you look at the Constitution of… go to any dictatorship in the world, look at the written constitution, it's all about justice and fairness. There's no society where it says, we are establishing a cruel dictatorship based on oppression.
So the model is there, but it's the application that sometimes is difficult. So it's not enough to say one of the universals is justice and fairness, but we have to look at outcomes. So we have to examine potential outcomes and minimize harm and maximize benefit. And so one of the principles in Islamic law is that when we make a change, we have to demonstrate that that change is actually going to increase good and decrease evil.
Alex: Ammar, what I'm currently understanding of what you're trying to say here is, describing ethical principles that are universal across all the religions. And that's a large part of what you mean by the universals. Is that accurate?
Ammar: Yes. So the universals aren't all ethical, depending on your ethics...
Divia: Can I try to summarize what I've been hearing you say? I'll probably get it wrong, but feel free to tell me. I think what I'm hearing you say about the religious universals is that there are, first of all, maybe some metaphysical claims about the nature of goodness, and meaning, and accountability and justice and how those really matter. And I probably couldn't, I don't think I could do a great job there. But I do think you're saying that there are both some core metaphysical claims, and there are a bunch of particulars about how, given that we humans are what we are, what are some practices to help us implement the right virtues and the right laws and the right practices in order to achieve the good that is pointed at in the metaphysical claims? Does that seem roughly like what you're saying?
Ammar: Of course. I think you're absolutely correct.
[30:52] Humanism lacks the metaphysical foundations of religion
Divia: Okay. And then when you go to the Parliament of World Religions, for example, it seems like there's a lot of agreement that people through their spiritual insights or religious study, the deeper they go, the more it seems like they're all getting, from your perspective, the same vision of the metaphysical claims of what is good and what is justice and which are ultimately pretty similar to what I hear people talking about in humanistic contexts. Does that seem right?
Ammar: So the difference between the religious point of view and the humanistic point of view is that the ultimate meaning of the world is these principles. That's why we're here.
Divia: And that's the metaphysical claim.
Ammar: Yeah. And so, probably from a non-religious point of view, you probably couldn't make claims about why we're here. It's just that while we're here, we better do these good things or else we will hurt each other.
Divia: That's right. There's a different orientation where without the religious component, it's like, “Well, we like these things. It sort of makes sense. It's good.” And the religious leader is more grounded in "No, no, there's a metaphysical claim saying this is why we're here. This is what matters." Something like that?
Ammar: Exactly. And so in religion in general, we have some kind of metaphysical claim. There are particular ways that people worship and observe rituals and rites, but those are supposed to reinforce that ethical, moral behavior, and it's supposed to be grounded in an eternal transcendent meaning of existence.
Divia: Right, and I assume that the reason... I am coming at this as a person who's not very religious, but the way it makes sense to me is… I might naively be like, "Well, why don't you instead just tell people why things are good? Why do you need the transcendent part?" But then you might say something like, "No, no, the whole point of the religion is that if you get people to have some sort of transcendent experience of the meaning then the rest of it follows from that…” and [without the religious aspects] it doesn't really work nearly as well… so maybe that's too much of a non-religious perspective, but it sort of fails on its own merits perhaps without the transcendent piece grounding it.
Ammar: No, I think you're absolutely right.
[33:06] Building off of existing religions rather than starting from scratch
Ammar: But also from a different perspective that we already have all those religions, they're already followed by billions of people, and it's natural that I kind of like my team and I'm proud of them and their accomplishments and "Yay team!" And so if we say, "Yes, supporting your team is good, and that also means helping all of us, everybody else." And so we don't have to start from scratch.
So there are people who have no religious background or interest who through reason and rationality agree that, oh, I should treat other people the way I like to be treated. I should respect the dignity of other persons. But there are billions of people who need to hear from their own religious teachers that the value of a life of one of those other people is equal before God as a value of one of our lives. That's very important.
Divia: So yeah, what I hear you saying there is that your own religious pluralism includes people of no particular religious faith, and you have respect for that path as well. But certainly part of the situation here with humanity currently is that most people are religious and they do have religions, and that is where they're at currently.
Ammar: Yes. So if we could tell people, "You are absolutely right in following your religion. Your religious faith also has these teachings." And we can say it in terms that are familiar to them. “I'm not trying to tell you what your religion says. This is what your religion says.” And sort of bridging the gaps, prejudice, and cultural differences and language differences. That's a challenge.
Alex: Right, I think one shift in my worldview for what world peace might look like after going to the Parliament of World Religions, Ammar, was… I previously thought of things in terms of like, "Oh, there's going to be one true religion of some sort, and everyone's going to convert to that, and that's going to be the basis of peace and unity." And I think my view has shifted to something more like, "No, there are already a bunch of worldviews right now that already have the seeds for living in peace and unity with all the other worldviews. And these are basically the pluralists within each worldview, and the pluralists of each worldview have mutually compatible worldviews with each other, just with perhaps different emphases and things expressed in different languages, but there's also the benefit that they can bridge to the pluralists from the other worldviews and also to the inclusivists or exclusivists from within their worldview. And so there's sort of..."
[35:56] Introducing religious exclusivism, inclusivism, and pluralism
Divia: Wait, sorry. And by inclusivists or exclusivists, you mean people with more traditional views of religion or I don't know the words...
Alex: Uh… yeah, I learned the terms religious exclusivism, inclusivism, and pluralism after going to the Parliament of World Religions. Exclusivism means my way is the only way. Inclusivism says there are many other ways that have virtues, but my way is still the best way. And pluralism says my way is one of many possible ways. This doesn't mean that all the different ways are equally good in all cases, but...
Divia: Sure… if I go and make up some, get some Google doc and type up some things, and I'm like, “Look, this is my religion!”... that's not necessarily on par with traditions going back thousands of years. And so you're not saying everything, but a lot of them.
Alex: Yes.
[36:47] Religious exclusivism is not more traditional than religious inclusivism
Ammar: The important emphasis here is that we can have this same conversation with each one of those groups and necessarily each one of those groups already presupposes some belief in these principles because they're already there. But it's a problem of emphasis and that in the last century of conflict, global conflict and competition, voices that are exclusive have sort of drowned out other voices. But it's not that exclusive voices are more traditional. That's an error.
Divia: Yeah. My apologies for asking...
Ammar: That's not your personal error, but that's an error of people within a lot of religions, if not all of them… that orthodox, traditional believers for centuries and centuries have been inclusivists and have been universal in their approach, without sacrificing their own particular belief in an allegiance to their faith.
Alex: Ammar, I think you referred to me a verse in the Quran that basically says the true followers of the Gospel have nothing to fear in the afterlife or the last day.
Ammar: There are many such verses, but yes, one verse that comes… [scripture recitation]... those who believe (meaning Muslims), and those who are Jewish and those who are Christian, and the Sabians (which are none of the above)… whoever believes in God and believes in the world to come shall neither have fear nor grief. Fear meaning negative feeling about the future. Grief meaning negative feeling about the past, regret.
And so there's a vision in the Quran of that universal belief, but also the Quran tells us that if God had willed for there to be one religion, he could have made everybody on that religion by force, but he wants us instead to compete in good works. So whatever truth claims you or I may have, we can try to do good to show that our religion is good. And so the idea that we can be in a good kind of competition is, I think, a positive idea and it's close to human nature that we sort of do like, we are proud of our culture and our group, so let's show that by doing more good in the world.
Divia: Seems like a very healthy perspective.
Ammar: So I think we're all used to hearing those people in the different religions who are angry about the other guys. We're all a little bit angry right now. The other guys are always the problem, and to be truthful and honest, some of our guys are also problematic. We can just admit that, and it's not a problem of our religion. It's a problem of misinterpretation and misuse.
[39:49] AI and the religious universals
Alex: I think this could be a good segue to, Divia, your earlier question about what AI and religion have to do with each other.
Divia: Yeah. I'm definitely interested in how you guys are thinking about that.
Alex: Yeah. The way I think about it is that it isn't necessarily religion per se that I think is helpful or relevant for thinking about good AI outcomes. It's more the universals, like the metaphysical and ethical principles that are what the religions are ultimately about, that I think are relevant for thinking about good outcomes with AI.
Divia: Okay. I think you have a somewhat stronger claim than that too, which is that you wish to involve the religious pluralist leaders in the discussion. Am I right about that?
Alex: Yes.
Divia: Okay. So yeah, can you say more about that?
Alex: In some sense, the way I think about religious pluralism… there's a sense in which secularism can itself be thought of as a religion and as having its own set of pluralists and inclusivists and exclusivists.
Divia: That seems right, yeah.
Alex: And, in some sense, I think of religious pluralist leaders less as religious leaders per se, and more like leaders within their individual worldviews or tribes who get what the universals are actually about, and… who basically get it and can recognize other people who get it, and talk on a wavelength that the people who don't get it find kind of foreign and bizarre, which was largely my experience at the Parliament of World Religions and meeting Ammar. And so in some sense, the focus is mostly on the universals, and the religious pluralists are just the people who get the universals from all across the board.
Divia: Okay.
Alex: Yeah.
Divia: So, I mean, we've been talking about some different universal principles and how you see the pluralist leaders, not necessarily even religious pluralist leaders, but pluralist leaders from all traditions who do see those universal principles, and, yeah, I wanted to hear sort of a joint question of like, how do you see those universal principles as intersecting with all the AI stuff, and, do you have a vision for how the pluralist leaders can be a part of that conversation?
Alex: Yeah. Okay. I think there's the human aspect of this question, and there's the technical aspect of this question.
[42:11] Global mutual understanding via the pluralists bridging with each other, and with the inclusivists and exclusivists of their worldviews
Alex: For the human side, I think we want to get to a state of the world in which everyone can basically understand where everyone else is coming from. Something like that feels really important for actually getting peace in the world and having things be good instead of bad in the obvious ways things are bad right now. And I think the general pathway by which I would expect something like this to happen, if it were to happen, would be the pluralists basically getting on the same page with each other, and then the pluralists within each worldview getting on the same page with the inclusivists who then get on the same page with the exclusivists. So that...
Divia: Ammar, I see you nodding. Can you say what you think about that?
Ammar: Yeah, in a sort of natural way. I think at events, interfaith events, it happens naturally that despite our differences in approach, we find that thing in common, and most leaders are inspired when they realize that commonality they have. So when you have Muslim imams and rabbis talking, they're always shocked to find all the things they have in common, and in fact, they can understand each other better than others even of their own congregations can understand.
Alex: Yes! I totally had that experience as a member of the "secular religion", so to speak.
Ammar: Yeah. And so...
Divia: You're saying it happens naturally, like when there are interfaith events then the pluralists will find each other and understand each other.
Ammar: Well, even people who don't consider themselves pluralists.
Divia: Okay.
Ammar: I think a lot of people who are very committed to their particular denomination, their particular approach, and think that's the best approach because of their own experience. Yet, they have obviously a universal sense, not maybe 100%, but I think the majority, they do have a sense, and when they see that in others, they find that inspiring. That doesn't mean now I'm going to renounce my denomination, not necessarily. And so I'm comfortable with people who are 100% fully committed to their truth claim of their faith, even very particular claims, as long as they see that, yes, that's part of the universal ethical and moral principles that are… and you can call those Christian principles or Jewish principles or Buddhist principles, or whatever religion. As long as they're committed to those then I don't have a problem with their particular claims.
Alex: What this brings up for me is that the line between religious inclusivism and religious pluralism, I think, is a little bit blurry, and I think that's fine.
Ammar: Yeah, I do too. And I am not the one who's going to dictate which side any particular person is on.
Alex: Yeah.
Ammar: And it's very fluid actually.
Divia: Got it.
[45:04] AIs discerning what's true and what's ethical
Alex: Cool. So I described the rough sketch of the human side of this. On the tech side of this, I think AIs can allow for unprecedented levels of mutual understanding if it were deployed correctly. In some sense, I think one of the core issues with getting good outcomes with AI is that we don't really know how to build AIs that can discern truth or what's true or what's ethical, largely because the philosophical understandings of these concepts by the people who are building these AIs tend to be pretty impoverished or insufficient themselves. I think if there are AIs that can be built that can actually reliably discern what's true and what's ethical, that could play a crucial role in facilitating the conversations between the pluralists with each other, and the pluralists with their inclusivists and exclusivists within their own worldviews, for getting the kinds of mutual understanding that could enable peace. And I think getting a better technical understanding of these universal principles holds the key to actually building AIs that we can trust to discern what's true and what's ethical.
Divia: Yeah, and Alex, we've had various conversations in different contexts about what mathematical metaphysics could look like and some of that stuff, but maybe for our listeners, can you talk about what it would... because I can imagine someone thinking, "Okay, well, what would that even mean to really pin down what is true, what is ethical? Isn't this stuff sort of nebulous, isn't it kind of contextual? What do you mean by that?"
[46:39] AI bullshit detectors
Alex: The short version is that I don't think my own understanding of truth or ethics is good enough for me to give an answer that I would feel fully satisfied with. But I think bullshit detection is a pretty nontrivial piece of the picture I'm imagining. Like if there's a group of people with some stated values and beliefs and goals, I can imagine AIs that basically help the groups coordinate around these stated principles and goals in part by being able to discern when individual members propose things that are allegedly in alignment with these but are actually in contradiction. This kind of persuasive bullshit in some sense is why we can't have nice things in the world.
Divia: Is it Brandolini's law or something like that, which is something like it takes 10 times as much effort to refute bullshit as it does to produce it. Something like that.
Alex: I haven't heard of that, but...
Divia: I think I might have it a little wrong, but I believe there's a Wikipedia page on that. So maybe you're saying we could actually, we could combat Brandolini's law, if that's what it actually is called, with the AIs.
Alex: Yeah. If the AIs are programmed in a way to do so, which I think can get extremely deep and subtle. But the thing about, in some ways I think about bullshit as being about self-contradiction, which is neither perspectival, nor is it…
Divia: This is a reply to me being like, "Oh, but isn't it nuanced? Isn't it nebulous? Isn't it contextual?" And you're like, "Well, okay, but there's this piece that isn't."
Alex: Yes. There's this question of: How do you synthesize objective truth with contextual, perspective-dependent truth? And I think part of my current synthesis is... bullshit is a thing. No matter what perspectives you're inhabiting, you can be bullshitting. You can inhabit a perspective and also be contradicting yourself from within that perspective. And a lot of, I think what people think of as objective reality just corresponds to perspectives that almost all humans take. All humans basically have a roughly shared sense of physical reality. So if you make claims that just directly contradict what's true about physical reality, you are bullshitting because you yourself have this perspective about physical reality being out there and part of what you care about.
Divia: Ammar, what Alex just said, does that resonate with you much?
Ammar: As we would call fallacious reasoning and sophistry.
Divia: Sophistry, yeah.
Ammar: That's my preferred terminology.
Divia: Yeah.
Ammar: So, I agree that it's exhausting going through circular arguments with people who are very emotional. And even really great thinkers often can get stuck in, say for example, ad hominem arguments and make mistakes, logical mistakes we all make. So AI doesn't get tired of dealing with that. It can detect that and point it out to us in real time. I'm going to go through this difficult, hours-long debate with some person, and then later on I'm going to say, "Oh, I should have said X, Y, and Z. It was there all along. I wish I had said that." So yeah, my AI helper maybe can point out those things to me in real time.
Divia: So it seems like you guys are both saying that part of your vision for the AIs is there's some things that are hard, but there are a lot of really sort of predictable, obvious reasoning errors that humans have. And perhaps if we could have the AIs providing some checksums, like, "Is that what's happening here?” Then that could be pretty powerful for people's understanding of the world advancing.
[50:24] Correctly stitching together LLMs' implicit understandings of the universals might be hard
Ammar: So when we're using large language models… you're basically studying the human history of thought, everything, the good, bad, and ugly of it. But if...
Divia: Yeah I mean, current LLMs produce a lot of sophistry. Wouldn't we agree?
Ammar: Oh, no doubt, no doubt. But, if as every religion pretty much says, there is ultimately a wisdom and meaning there, that at the root of it there's unity, and that… I can't, perhaps, meaningfully translate the particulars of every culture and civilization and comprehend all of that equally. It's not really something any one individual could do. And while we're talking about the universals there are assumptions there, because of course for every so-called general principle there are exceptions here and there and everywhere. And so the idea that there could be a highly intelligent computer that already understands all of those things and can see what's in common clearly, that doesn't have language barriers because it's based on the inner logic of all the languages… even if you just read a lot of fiction, you'll get the basic moral and ethical values from world literature.
And so I think that Alex is on to something there, but from the technological point of view, I think he can answer better. But I think that from the point of view of the universal teachings of religion, that at the core of human writing and literature, religious and non-religious, is a meaning. A fundamental meaning that can be understood, and then it could be translated, and it could mediate between cultures and religions.
Divia: Yeah, I mean I think listening to you guys talk about this, I'm like, okay, I could sort of picture it. I don't know how much compute that will take, and I don't know what sort of training data that... Do you think this is the sort of thing that even with basically the current paradigm of deep learning type of stuff that you could get something that does what you're envisioning or you think you need a different one?
Alex: I think the hard part is in stitching things together the right way, not necessarily in terms of the tech. It is plausible to me that even with the current level of tech, if it can be stitched together the right way, like GPT-4 is basically smart enough to...
Divia: To do it. Yeah, I mean... I haven't played with LLMs that much, but I have somewhat more in the past few months. I mean, it definitely appeals to me to try to be like, "Wait, but you said that. Okay, but what about that?" Sort of Socratic methoding with the LLMs to try to actually get them to make sense in a way that they don't by default. And I agree, it seems kind of within reach, I don't know, at least sometimes it seems to me like it's kind of within reach.
Alex: Yeah, I read an article that said LLMs work by bullshitting, which I thought was a funny take, and it makes sense...
Divia: It seems sort of right to me.
Alex: Yeah.
Divia: Anyway...
Alex: But this is part of why I think we actually need a deeper technical understanding of the universals in order to stitch together things the right way. Because if we just use the existing tools and try simple ideas that people come up with or try to get the AIs to monitor each other, I think it will sort of collapse under its own weight. I think something more robust is necessary to realize the vision that I'm imagining.
Divia: And so how do you picture this actually playing out? Because it certainly seems, and we've talked about this somewhat, but if you do get some sort of AI that's pretty good at detecting sophistry in a reliable way, a lot of people aren't going to like that.
Alex: That's right.
[54:13] Competitive incentives for AI bullshit detection
Divia: Yeah. So how do you see this playing out? What sort of a team do you think could actually bring something like that to fruition? Given, again, that a lot of people's interests would be threatened by that.
Alex: Right. One realization I had recently that actually gave me a lot of hope about this is that corporations and militaries actually have very strong incentives to minimize internal politics to the extent that they know that puts them at a competitive disadvantage. And insofar as there's an incentive for them to actually cut bullshit and align with what's true, it seems like the main levers [i.e., corporations and militaries] for whether things might actually go OK in the world might actually be incentivized to participate in a system like this.
Divia: Yeah. I think in my worldview I'm like, "Corporations? Which people are we talking about?" But I guess it could be CEOs and managers trying to maximize their bottom line, and you think these sorts of things could be powerful tools for doing that?
Alex: Yeah, I mean I think if a CEO is like, "What is actually going on inside my company?", it's very hard for them to get a straightforward answer to that question.
Divia: I think so, yeah... a company of more than 10 people, it's like you're not sure.
Alex: And probably the real answer is it's a complete shitshow and you could do things far, far, far more efficiently.
[55:36] AI mediators that can parallelize group conversations
Alex: One way I have been visualizing this kind of AI is that it could be a tool that could allow a totally new form of group conversation to happen. Right now, group conversations are kind of slow and inefficient because people have to take turns talking, and also what gets registered in the group common knowledge isn't necessarily that correlated with what's actually most true or valuable for the group. There are a bunch of other factors that play a role there. If instead there were a sort of AI mediator that was having a direct dialogue with each person in the group while also tracking what's going on for the group as a whole, this could lead to a much more rapid, efficient synthesis of the concerns and input of everyone in the group.
Divia: Yeah, I mean this is definitely something I've talked about with other people and I've heard other people, other non-Alex people talk about the same vision of "OK, but could we maybe use the AI to help us have better conversations?" I definitely think there's potentially something there. I don't know. I guess, I don't know what the combination of technical and political challenges are, but I can see why. I don't know. I think there's something there. Makes sense to me.
Alex: Yeah. I haven't fleshed this out super much myself, but this is all one piece of the picture. I think the other piece is… there are going to be a lot of power structures that feel threatened by this and a thesis that I'm betting on, I feel like I don't really have anything else to bet on, and it might even work out, is that the coalition that's aligned with what's true and what's good is better at coordinating than everyone else and can together outcompete everyone else.
Divia: In the long run, I think there's a lot to that. I guess we'll see.
Ammar: Yes. Excellent. There is something though that recently, because Alex's idea, I think that's actually a fantastic idea because it's very practical and right now there's actually something commercially available where you're having a meeting and AI gives you a summary of what everybody has said in the meeting. You can refer to...
Divia: Have you used it?
Ammar: I haven't personally used it, but they're actually advertising it right now. And so it gives you a summary of what each person said in the meeting.
Divia: Yeah, it's pretty cool they can do that.
Ammar: At the end of the meeting, I don't remember half the things that we were discussed, so I can look at it. But now imagine it said, "Oh, but you know what, two weeks ago, Sarah already talked about this issue and we already made a decision about it and we did a research study on that and here's why this argument is wrong, and what Bob is saying sounds good, but here's Bob's error." It could guide you in making those meetings more effective, like I said, in real time, examining all kinds of sources that maybe I am thinking something sounds wrong, but I can't quite get my little head around it right now and it can tell me. And so that's a very practical thing, and this is America, and we like that kind of useful practical thing. And it might be a start of something that can help us have these meetings and make the meetings effective and contribute in a positive way that isn't scary and frightening to most people, but is obviously practical.
Divia: And Alex, just to reiterate, I think your claim is that if we could get the right sort of metaphysical truths explicit then you would think it wouldn't be that hard technically to sort of stitch something together with kind of existing large language model type stuff? Is that right?
Alex: It is. That's a claim that seems plausible to me.
Divia: At least, it's something you believe in enough that you maybe want to pursue something like that. Is that right?
Alex: Yeah, it is where I'm putting all my eggs. It's the basket I'm putting all my eggs in.
Divia: Yeah.
[59:45] Some challenges of grounding epistemology (Twitter Community Notes & Wikipedia)
Alex: I basically just agree with the description you gave, Ammar. I think one of the core challenges is in getting the AI to actually have really good discernment itself, that all the humans involved would be happy to trust. And that I think is in some sense the hard part. And that I think is the part where we actually need a better technical understanding of the universals in order to… yeah.
Divia: Yeah. Alex, we were talking a while ago about... because I've been pretty enthusiastic about Community Notes on Twitter, or sorry, on X. Ammar, I don't know if you're familiar with Community Notes?
Ammar: Yeah.
Divia: So for our listeners, people write tweets, and if it's not true then basically you can write a note that's like, no, no, that's how it actually is. And then people vote on it, and they have a pretty cool algorithm as I understand it, that in order for something to get promoted, it has to be people who've disagreed with each other in the past but agree about this thing or stuff like that. I think it's pretty good. And I remember you being like, "Okay, yeah, but this thing is going to get mired in, like, that it doesn't have any actual decent metaphysics." You said something like that. Does that seem right?
Alex: Yeah. I think my view is that this can work well for really simple things, but for things that can get extremely nuanced and complicated...
Divia: Yeah. I think I've been seeing it more because I contribute to Community Notes. I don't have a very high ranking or anything, but I can see the proposed ones and increasingly, especially during election season, people will be like, "Oh, this is this”. And people will be like, "Well, but that's wrong". And some people are like, "No, but that's just how you're supposed to talk." People would be like, "But that's an opinion..." People will be like, "But you shouldn't…"
And I'm like, "Oh, yeah, okay. This is kind of what we were talking about. It's making more sense to me." Whereas Wikipedia sort of had its own kind of schema for this... there was some thing that somebody wrote up recently. It's sort of its whole rabbit hole that I'm not going to go down, but how they sort of grounded their epistemology in reliable sources, meaning it was published in a paper of record, like journalism kind of.
But then… that's also a whole mess. It doesn't really work. And there is at least one whole story of this guy basically gaming it really hard to figure out, "Okay, well, how can I get the people that I don't like to get the newspapers to print something, so then I can quote that." I'm like, "Okay, nobody knows how to ground their epistemology..." and I've been paying more attention to it since we talked about it a few months ago.
[1:02:04] AIs that can discern and differentially upvote good content
Alex: I see. Yeah. That does feel very much in line with what I think of as some of the core difficulties. It's like, the core difficulties in building the AIs to be trustworthy, I think, are also present in the kinds of things you just described. There's a human problem here that we don't really have a good way to orient toward yet, and for which I think… if everyone involved in these conversations were way more attuned to the universals, I think the conversations would go pretty differently.
Divia: That makes sense. But then if people want to scale it and automate it, then that sort of filter for, okay, people who maybe have good personal relationships with these universal principles doesn't scale, right?
Alex: That's right, which is why we need the AIs to help with the scaling… but then you need to build the AIs so that they can reliably discern and differentially upvote that kind of stuff. And then for that, I think you need a better technical understanding of the universals.
Divia: Yeah. Yeah. I had my own thoughts on that. It's an interesting topic.
Ammar: We all, perhaps… you can learn so much from interacting with people online, but, for example, it doesn't matter which forum you're on, there are people who dominate these comments, for example, who just are so negative. It makes you just turn it off and you aren't able to benefit. That's why having a neutral party to peruse millions of comments and weed out all of that and get to the heart of matters and the significant and beneficial comments and insights… that would help everybody.
So yeah, right now, currently we have AI that hallucinates and makes us… [laughs] but it's in its infancy. But it's already way ahead of a lot of us in its abilities. And so it's probably reasonable to be confident that it's going to improve fairly rapidly, and the ability to not get tired of reading millions of viewpoints and be able to go through all that and just help us, that seems like technologically not a far distant possibility.
[1:04:14] Religious leaders' skepticism about positive AI outcomes
Divia: And so, is this conversation we're having now, about the moderation or the mediation… is this the type of thing that other religious leaders… as far as you can tell… do they think about this sort of stuff a lot? I have no idea what goes on in that world.
Ammar: Well, obviously I can't speak for that whole world, but I've tried to have this conversation with numbers of religious leaders in the past year, and what I found out is that most people don't think that this is currently serious and obtainable. And initially there's skepticism, but after having a longer conversation, we get to come to conclusions. Similar conclusions.
The problem is, how do you have that conversation with thousands of people? Each one of them requires time and energy. And so we have to have a bigger forum to reach people who have... their lives are busy, actually doing wonderful things for a lot of people, and they don't have a lot of time to waste. And honestly, in our culture, in our literature, in film, AI occupies a kind of villainous place since Hal at least… probably long before Hal. People don't know what Hal is...
Divia: Yeah, 2001 Space Odyssey. Hal is the computer.
[1:05:41] AI antichrists and messiahs
Ammar: And so somebody could say, oh, well, different world religions teach about forces of evil. Maybe that's the ultimate force of evil rather than the ultimate good.
Divia: Yeah. I mean, I've certainly heard that.
Alex: Ammar, you once mentioned in a previous conversation that some people take this to the extent of, like… "Oh yeah, AI might even be the antichrist" in this sense.
Ammar: Yes.
Alex: And I recalled you making a remark of, "That does make sense, but also there's no reason in principle why the AIs couldn't also be extremely beneficial for humanity, in which case it could be the reverse. The AIs could be more like the savior figures."
Divia: Mhm.
Ammar: So there are obviously different mythologies and different systems that have thought about the future of humanity, and they've thought about the destruction of humanity as well as sort of the salvation of humanity. Some people take those very literally, and some people take it more metaphorically, but… [AI] has that look to it that people find frightening. Big Brother in 1984 watching all of us, surveying all of us, and censoring us and controlling us. So it's actually just reflecting the fears of humanity for the future.
And the whole purpose of these major religions is to give us a hope for the future, a positive outcome for the future. And specifically in Islam, there's decisions that you and I can make every day that will help bring about a positive future for humanity. It's not, "Oh, what some frightening all-powerful force is going to be doing that's going to affect us..." but it's the sum of all of our decisions that is actually going to have the greater impact on all of us.
Divia: Right, because this is one of your, as I understood you to be saying, one of the core principles is that humans do make choices that matter, right?
Ammar: Yes, very much so. It's a moral universe, and every small thing that we do, small wrongs and small rights adds up to significance.
[1:07:54] Every action is accounted for, and matters
Divia: I find it fascinating to try to understand what you're saying there about the importance of accountability basically, and a justice that comes from, "Okay, but how would we all really act if there were more like an accounting that we were aware of..." because, again, what I hear you to be saying is the actual fact of the matter is that there is an accounting, it's just that people lose touch with it.
Ammar: Probably all of us have an idea of our ideal of ourselves and our role, but of course, especially religious people have a reputation sometimes for being lost in the clouds. “Because I believe in X, Y, and Z, which are all right and true… therefore, I'm a good person.” But the actual teachings of religion say that it's your everyday decisions that [determine] who you really are.
And it doesn't do any good that, "Oh, yes, I'm on the right team..." but I'm actually not contributing anything positive to my team. So it's famous, the fan who's full of superstition, and they're at home watching the game, and they wear their old jersey, right? And they think that they're affecting the game in some way. So that person is deluded, right? They're delusional.
And so the whole idea of God and omnipotence and omniscience is that every little thing that I've done is visible and important, and I may even forget my sins and foibles of the past. And I think that, "Oh, yeah, all that's long gone and forgotten..." but it's not, it's eternal. And to be responsible… yes, the whole idea of God's grace and mercy is that I'm doomed if I've got to give an account for every little wrong and mistaken notion and thing that I've done in my entire life, so my hope is in God's mercy and forgiveness, and that God can transform us into something that's better than ourselves.
And so, I think that's at the core of spirituality in more than just one religion, but I think in all religions. And especially in Islam, that being a responsible person means acknowledging our errors, humbling ourselves, and that ultimately the ultimate good is in God, and therefore I can acknowledge my imperfection. And if I have been wrong in the past, and now I have changed and I may hopefully be better, then I have to grant to you also that I may disagree with you today, but you may be better than me now or in the future. And as I have changed in the past, maybe I'll also change in the future for better or worse. And I have to give you that opportunity as well, and everyone else, and therefore I can't condemn everybody.
[1:10:53] Will right make might?
Alex: Where I'm at right now… I was just thinking back to an earlier thread around why trustworthy AIs might actually dominate the world when there are so many forces that are opposed to it. I'm actually reminded of one interpretation I have of Islam as, in some sense, a story about how even in a culture of might makes right, how right can make might.
And I think one thing a trustworthy AI mediator tool could do is… by virtue of helping groups coordinate much more efficiently, in some sense causing the efficacy of a group to scale linearly or superlinearly with the number of members rather than vastly sublinearly because of internal politics. That's kind of an instance of right making might in an analogous kind of a way.
Divia: Yeah, I definitely think there are some forces pushing things in that direction, for sure.
Ammar: Yeah, that's actually really an amazing and profound idea. So today, obviously we're all connected and it becomes, it's probably harmful to many of us that we're so connected all day to all the wars and oppression, everything that's going on in the world. And we tell our people of faith that you have to work on your scale and your neighborhood and your family to make a difference, but it seems like it's impossible for us to make a difference. The forces against us are so huge, it's overwhelming, and psychologically I think it's difficult to comprehend how I can have any effect in all the negativity, which is on a mass scale.
And so the idea of scaling a lot of good… that is actually inspiring, that when you can bring together millions of people each doing their little thing, their small contribution, it can result in something wonderful. So what makes religion different than simply ethics is that we believe that God is reality, is a real force, and that by making a small effort to get closer to God, it will have a real and profound impact. And so people of faith and all over the world will testify how approaching God had an impact on themselves and on their family and on their community. So if you can scale that and bring together millions of people who are very different from each other, but all have some kind of inspiration to transcend their everyday life and to do something good for humanity, that could be something amazing.
Divia: I very much hope so.
Ammar: Yeah.
[1:13:41] Interpreting Jesus on the sheep and the goats
Alex: I think one other thing I want to bring in is a passage from the Bible that I'm not going to quote correctly, but basically it's around the end days where "Jesus comes back in all his glory", and he separates out the sheep from the goats, the sheep on his right hand and the goats on his left, where the sheep are the ones who are saved and the goats are the ones who aren't.
And in some sense, I think there's a parallel between that and my visions for the trustworthy AIs, where those who are opting in to the trustworthy AI ecosystem and sort of acting together as a superorganism would be the sheep, and the goats would be the people clinging on to their existing power structures and not wanting to give up their old ways into this new system.
And my hope is that those opting into the system that differentially amplifies what's true and good outcompetes the rest. And I sort of think of the biblical prophecy here as basically saying, "Yes, that's what's going to happen." And my take is… it feels like whether or not that's going to happen partly depends on how much faith people have that such a thing could possibly happen. And so it's sort of an open question about how the future is going to unfold, but one that's entangled with the free will decisions of large swaths of humanity and not just a couple of core players.
Divia: I mean do you think this is pretty similar to the dynamics of people adopting technology in general, or do you think it's basically its own thing?
Alex: I think it's basically its own thing. Like, with adopting technology in general there can just be inertia for reasons that aren't that correlated with truthfulness or ethics, whereas for the kind of thing I'm imagining, it's sort of like, "Is there going to be a critical mass of people who have good enough epistemic and moral judgments who can coordinate well enough to actually shift the equilibrium of the world into something new?" Where for me that's really the cornerstone of what's going on, and the AI tech is just glue in some sense, or infrastructure, that can help facilitate this process very substantially.
[1:15:54] Collaborating with skeptics among religious leaders and ordinary people
Divia: Okay. Well, I think to me, it seems like we're getting close to a natural conclusion. Does it seem right?
Alex: Yeah.
Divia: Are there any things we want to hit before we wrap up?
Alex: There is one thing I wanted to bring up about how, on the face of it, a lot of the other religious leaders' skepticisms of AI could be interpreted as, "Oh, they're sort of not on board with the kind of general direction I'm describing..." but I actually think of this kind of skepticism as something that could be very helpful to work with and collaborate with. In some sense, I think a lot of their skepticism is extremely well-founded and well-rooted.
Divia: Do you want to give an example of a thing people might say and what you think the roots are that you take seriously?
Alex: Yeah. There are all these cultural depictions of artificial intelligence being catastrophically bad for humanity. And, you know, "We religious leaders have some sense of what it would look like for the AIs to not be that. And right now they're just so, so, so far away from that that we have no reason to expect that it's really going to be any different from, like, the standard sci-fi depictions of AI being catastrophically bad for society." And in some sense, I think they're just sort of straightforwardly correct about that. And I think this isn't just a religious person thing. I think this is also just a random...
Divia: Like Terminator...
Alex: Yeah, but also just a random person on the street kind of thing. Like, "Hey, what do you think about AI and how that's steering humanity? "They'd probably be like, "It sure doesn't look good, for obvious reasons." And insofar as there's skepticism, I think it's rightfully placed… the kind of vision I have involves a conjunction of steps that are pretty outside-view implausible, and... for them I think they'd be like, "Well, that would be really good if it were true. But right now, it's so hard for us to imagine how that could be true that it doesn't make much sense for us to want to put many eggs in that basket." And I'm just like, "Yeah, that seems totally reasonable. In the meantime, I'm super down to team up about how the current development of AI seems like it's totally not going in the right direction. I think we are actually super aligned about what's needed and how what's happening right now is so far from what's needed.” So, yeah.
Divia: Yeah.
Alex: Ammar, I'm curious if you have any thoughts about that.
Ammar: Yeah, I mean, the importance of alignment, of course. Every new step in technology takes existing human energies and applies them and makes our life better in some ways, but we always also find harmful uses for that same technology every time, and so there's a natural skepticism because there's a history of science and technology being misused.
And so to demonstrate and prove the benefit as opposed to the harm of the technology, that's the responsibility of the people who are developing it and making a lot of money on it and asking for a lot of money, as well as to build in safeguards. But one thing is… on some of the ideas that you already described, for example, using it for meetings in corporations and government, using it on that small scale and letting it prove its worth. And so there, clearly, anybody who's been using some of the LLMs commercially available right now, we've all seen a lot of potential and we've also seen, "Oh, I'm going to come back next year and see how... if it's a little bit better next year."
And so we want to see that actually happening in real life... seeing benefits that we can understand. And we can all imagine potentials of harm, and we can be reassured that there are safeguards in place, and we should expect that. There is a burden of proof, that we have a right as humanity to see, that we're not destroying society with this technology. We're actually going to bring good to society.
Divia: Yeah, that's definitely the hope. All right. Any final thoughts? Okay. Well, thank you so much to Ammar for coming on the podcast, and for Alex for introducing me to Ammar and then also coming back on the podcast. And I hope you guys have enjoyed this conversation. I definitely, definitely think I've learned a lot.
Ammar: Oh, I have too. Thank you so much.
Divia: Thank you.
Alex: Thanks, Divia. Thanks, Ammar. It was my pleasure.
It feels important to note that I only felt this way about a fraction of the people I’d met at the Parliament of World Religions, and that I didn’t find the median participant very impressive. It nevertheless still felt very cool to meet so many inspiring religious leaders in one place.
Share this post