In this episode we engage with renowned physicist Sara Walker on topics of life's origin, artificial intelligence, and our cosmic future. We delve into Walker's career journey, which starts from her fascination with physics at a community college and leads her to question the origin of life as a theoretical physicist at Arizona State University.
Walker talks about how understanding life's origins is an open frontier that could potentially reshape our perspective of ourselves and our future, the role of AI in comprehending life's structure and origins, and its planetary-scale implications for humanity.
We also hear Walker's advice for young researchers stepping into this field. Tune in for a captivating exploration into life's deepest enigmas!
Sara Imari Walker is an American theoretical physicist and astrobiologist with research interests in the origins of life, astrobiology, physics of life, emergence, complex and dynamical systems, and artificial life.
Philipp Lenssen from Germany has been exploring between technology and art for all his life. He developed sandbox universes Manyland and wrote a technology blog for 7 years. He's currently working on new daily pictures at Instagram.com/PhilippLenssen.
Allison Duettmann: Hi, everyone. Welcome to Foresight Existential Hope Podcast. We have a special episode with Sara Walker that we have been looking forward to a really long time. I am not even quite sure where I first came across you, but once I did, you were everywhere. It is to the extent that I think you are one of the only people, that I know of at least, that have gone on Lex Fridman twice, once solo and once as a debate with Lee Cronin on anything from the origin of life to alien life and long-term futures in space. It was a really inspiring debate, and it definitely tackled a lot of different ground. I also did some research and found you with Danial Dennett and Nick Lane. You are very broadly focused, which may be an oxymoron, researcher and mentor for people trying to find their way around those questions related to the origins of life, the complexity of life, different life forms, there are other life forms out there, when will we meet them, is it possible that they are already around, and what our long-term potential in space, and so far. I think we will not run out of fun, controversial topics to talk about today. Firstly, if you had to describe your own life story in a few sentences, what would it be? How would you describe your own path, and where are you currently? I know you wear different hats as well.
Sara Walker: I am currently based at Arizona State University, but I have affiliations with a couple of other places, which is fun to diversify what you think about. I guess in my earlier days, I got really romanticized by physics, which is maybe a little odd. I went to community college when I started university, and I took a physics course. I just thought physics was the most amazing topic that we could understand the universe at such a deep, fundamental level. I was kind of dead set at that stage that I would become a theoretical physicist, although I did not really know what that meant because I thought it meant studying cosmology and particle physics and things that physics had been historically good at. However, the real reason I was attracted to physics in the first place was because of this idea of deep explanations that actually drive humanity forward and really allow us to explain more, but from those explanations emerge new technology and new ideas.
So when I really started to realize that that was sort of like my core interest was already after I was a PhD student and I had started working on the problem of the origin of life. The origin of life is a really interesting problem because it's one of those open scientific questions, where it's so hard that I'm not even sure that we've framed the question right yet, which is one of the reasons it's hard to get to an answer. So for me, that was incredibly exciting because it was like this open frontier, where it seems so obvious that we should know what life is, yet we realize we do not when we try to get to these deep explanations. So if we could really come to understand what we are, that would open a potentially entirely new vision for what the future looks like. I think those are the things that really excite me about focusing on the origin of life problem, and why I think it's one of the most important problems of our time.
Allison Duettmann: Yes, and I think many of the other problems seem to stem from that.
Sara Walker: Yes!
Allison Duettmann: The way you mentioned you are interested in physics for the explanation, we just had David Deutsch on. He is really about all of the proper explanations.
Sara Walker: I am amazed! Yeah, he really inspired me a lot. I love his approach to thinking about how explanations transform the future, so I think about that a lot when I think about what theories are.
Allison Duettmann: It is relatively meta, you know? I think it is kind of percolated throughout every part of your existence, whether it is a scientific domain or, frankly, your personal one. Okay, wonderful. If someone else was coming into the field, and I know that you have taught, mentored, and instructed a bunch of young folks, what would you tell them? What are your research questions that you address? What is an explanation of the space you are in if someone else came new to the table?
Sara Walker: Yeah, I mean, I think the easiest way to think about it is really that there is only one question I'm really interested in answering, and that is: What is life? Part of my motivation for that is really because I want to understand the origin of life and the transition from the universe without life to a universe with life, and that seems like a pretty radical transformation. However, because that question is so broad and so deep, it touches a lot of areas. My research group is actually fairly eclectic; I have a lot of Ph.D. students that come in with seem like totally different interests. For instance, one of my Ph.D. students is interested in whether there are laws that govern civilizations and their long-term future and how they get off planet. Another one is really interested in artificial general intelligence, and how do we build intelligent machines. Another one has a geochemist background and really interested in the geochemical underpinnings of ecosystems.
While you get this sort of like very eclectic mix of people from exoplanets to AI to everything else, the core unifying theme is understanding that life manifests in a lot of different things that we study. If we can actually develop a deeper understanding across those different domains, we might actually solve this really hard problem. On the surface, it looks like it is a lot of different stuff, but for me, it is actually really this one question. Also, if you look at the history of physics, it is very true that when we like sort of tunnel through and make these new explanations that are very expansive, they unify and compress a whole bunch of stuff that we thought was totally different before we really came up with that kind of explanation. So I guess we're trying to be as open-minded as possible as to what the explanation of life is, and trying to just study that problem in as many domains as possible to try to figure out what is actually common across all these different ways of asking the question.
Allison Duettmann: Wow, that is crazy. So it literally goes from civilizational, scaling laws to sort of AGI.
Sara Walker: And find the origin of life and chemistry. Everything pretty much subsequent to the origin of life transition on this planet is life. It is pretty interesting. In some ways, you might think about what is happening with our technology as being the origin of life recapitulating at a different scale. You know, think about planetary-scale technology. If you think about the origin of life as a planetary scale transformation in geochemistry that happened 3.8 billion years ago and has elaborated ever since, you start to see these sort of recurring structures across scales, and technology seems to be doing something interesting now. It's a fun time to be an origin of life scientist and look for patterns across these different systems.
Allison Duettmann: Could you speak a bit more about patterns? I mean, I think from Game Theory, like one that maybe has driven humans, is reciprocal altruism so far, more or less at least. There are some that are not even meta. I would not even say, for example, that David Deutsch
Sara Walker: Actually, the thing that I'm most interested in is what kinds of things get to exist. For instance, our imagination seems like it's unbounded, and there are a lot of things we can imagine could exist but don't exist. Yet, at the same time, our imagination, at some sense, defines the horizon of what gets to come to exist in modern civilization. So we are building all of these new technologies based on what we can imagine is possible. For me, the sort of the deepest structure, the physics of life, is actually that the universe doesn't just generate every possible thing that could exist; it only generates some things, and they are historically contingent, so there is some selection and information that has to be built up over time in order to make complex things.
So in some ways, I think about the physics of life as being the physics that generates novelty or creativity in the universe. Therefore, if you think about what we're doing as a sort of globally integrated bio-techno sphere right now, then we are one of the most creative structures we know of in the universe, which means we are probably one of the most alive. That process, to me is quite fascinating. If I look across life, and I say, “What's the most distinctive thing life is doing that it can't regularize in terms of current disruptions in physics?” It would be this process of generating novelty and selecting on things that are generated to actually build the future. I do not think the future is predetermined but instead open. Basically, I think everything that exists on the planet now is actively constructing that.
Allison Duettmann: Well, that is a relatively optimistic take, you know. I think it is a great one. At the same time, you also said that there are specific things, such as the conception space of our imaginations actually influencing what can be produced in the universe. Do you think of it as this universe or the universe as the multiverse of different universes?
Sara Walker: Yeah. So it is interesting because I think the multiverse is a real thing but not in the sense we think about it. I don't believe in anything outside of like our current physical universe because I am a physicist, and I have these biases. However, I do think that our ability to imagine these possibilities is actually causal to what happens in this universe. If we couldn't imagine things then we cannot do them. So even our ability to reason counterfactually about what other universes is possible allows us to understand the physics of our universe better, which allows us to do more things in this universe. So I think imagination is actually opening up the possibility space in this universe. And that is what I would attribute sort of the causal power of multiverse - imagination - but I would not say the multiverse is a real thing for me.
Allison Duettmann: Got it, that is very interesting. So you have been in that field now for how long would you say?
Sara Walker: I finished my Ph.D. in 2005, so I have been at it for a while.
Allison Duettmann: That is crazy.
Sara Walker: No, sorry, that is my undergrad. My Ph.D. was in 2010, sorry I cannot even date myself. Time is a weird thing. So yeah, and I have been at ASU for, I think, 11 years now or 12 years, first as a postdoc and then on the faculty.
Allison Duettmann: I love ASU. I was once on a panel. I think it is really wonderful how they ask really deep questions with actual scientific rigor and expertise. I think some of the quirkiest events are always at ASU.
Sara Walker: I love ASU for that reason. It is really great.
Allison Duettmann: Okay, so you have been there now for about 10-11 years or something? Has there been any exciting kind of cultural shifts? Or any new ideas on the block? Where you are like, “Oh okay, we can now talk about this, whereas we were unable to before that.” I know in previous podcasts, you have talked about UFO sightings and so forth. Have there been any shifts stretching the Overton Window, even at ASU, where they already have a pretty broad window? Feel free to share anything of interest that someone new to the field may be able to get a heads up from.
Sara Walker: Yeah. I can tell you sort of a major one in my career, at least in origins of life, which I have been deeply fascinated to watch this shift. When I started in the field, I was a Ph.D. student in theoretical physics. I remember originally going to conferences, and it would just be something like 90 organic chemists and like 10 biologists, and maybe one physicist out of 101 people or something. Maybe the ratios were a little different than that. But the point is that the field was predominantly chemists that were really interested in synthesizing the building blocks of life. There wasn't a lot of emphasis on the sort of broader systems-level understanding of what life is, or thinking about life as not substrate specific, so no chemistry-specific phenomena. In the course of my career, there has been a radical shift, where people are starting to recognize that maybe life is much more about information processing and these sort of more abstract properties. We do not need to necessarily talk about them as being chemical, like properties that only appear in chemistry, let alone specific chemistry that we see life on Earth using.
So you can have a conceptual shift where you say the chemistry life on Earth uses is not the only possibility; it might be other possible chemistries. But then there is another conceptual shift where you could say when we are talking about life, it might be a broader class of phenomena, and of course, we can think about silicon-based life, but even broader than that, can you think about life as a phenomenon exists across scale. It is not just an individual cell, but populations of cells or hierarchies in biology, such as ecosystems or planetary scale processes equally being alive. Overall, those things have always been sort of themes, obviously. People have been talking about those kinds of processes and different ways of describing life for a long time, but I think it's becoming much more mainstream in the origin of life specifically to bring in some of these ideas from complex systems and in other ways of thinking. It has been really exciting. Yeah, that was the biggest one.
Allison Duettmann: What is your definition of life that is substrate, or at least more substrate independent?
Sara Walker: Yeah, I have a lot of different definitions of life. The thing I'm thinking about most is the relationship between life, information, and time. I can get to that in a minute. Sort of the working things I say when I talk about what life is, I say life is how information structures matter across space and time. So for me, when I talk about an example of life, it's not like an individual cell or an individual human being. It's the idea that there was some transition in the physics of our planet early on, and now we've entered this open ended cascade, where memory of the past is retained that actually allows building new possibilities in the future. It is that entire structure, across space and time, over 3.8 billion years so far, and whatever it generates in the future, that is an example of life.
I think that's a really different conceptualization of what life is and you might historically hear, particularly in my field. You hear things like life is a chemical system capable of Darwinian evolution or something which is very narrow and specific. However, one of the reasons I think you need to go broad and abstract is because, when looking at the history of physics, these deep extractions end up looking a lot different from the phenomena they were originally intending to describe. They are much more encompassing in explanatory breadth and reach, to use David Deutsch’s terms, in terms of what they can actually describe. I think that is going to be much more useful.
The relationship to time actually comes out of the theory that I'm developing with Lee Cronin called assembly theory, which aims to explain features that we associate to life and also be used as a tool for solving the origin of life. We are basically trying to build a theory that allows us to design the experiment to solve the origin of life. That theory really suggests that one of the most interesting things is the information necessary to build an object becomes an intrinsic feature of the object. This has an implication at least for me, philosophically, that's super interesting, that complex objects or living objects actually have a size and time.
So I have a friend Michael Lachman who talks about himself being 3.8 billion years old because some pieces of all of us are that old. We all share the inner structure of the ribosomes that are in our bodies helping make proteins. And that structure is a structure that is a pattern that has persisted on this planet for 3.8 billion years. I sometimes talk about that as one of the earliest technologies. If we're thinking about building future technologies, we haven't built anything we expect to last for 3.8 billion years, yet life has invented such technologies in the past, so it might be possible.
Allison Duettmann: I think there is a really good analogy that Mark Miller discussed with me, and he is a Senior Fellow. He basically said, “Hey look, once we have some cool longevity, if we ever get there, we need to radically update our understanding of what it is to be us.” Currently, it is this identity similarity over time that makes me think I am me. If we live very long time periods, however, and we have this propensity for radical change, then it is really a more broad definition of growing into something that is being proud of the version of yourself. He sort of analogized it with the definition of a civilization. It is not like I am truly a part of the Roman civilization, but to some extent, I am the product. I am humbly and gracefully looking back at a few of the things that they did. I think you have this larger definition of life that almost evolves and emerges overtime, which is really interesting.
Sara Walker: I like that a lot. Sometimes I think of myself as a temporary aggregate of information. I see it sometimes as though all of us are just like these temporary aggregates of this larger structure that is unfolding over time and building more things. Features of that persist in the aggregate into new things and new objects over time. But all of the particular things you see in the biosphere are temporary, yet all of that information is still propagating through material substrate. For me, that is actually what life is. The thing that is super interesting is that temporal structure, and you have to take time very seriously I think when you're thinking about what life is. Thinking about ourselves as being very old objects that have a large size in time brings a really interesting dimensionality to physics that I don't think has been present before. Things can only exist after the information exists to build them. So there is this whole structure you can add to the space of things that can be created in the universe, if you take time seriously.
Allison Duettmann: Interesting. Okay, so now, there are things that we can build, even just information wise, that we were not able to do at all before?
Sara Walker: Yes.
Allison Duettmann: This notion is generally a nice way of dealing with existential doom. If you are part of this larger informational evolution really, then your physical body is perhaps sad but not as sad as it would otherwise be.
Sara Walker: Yeah. I mean, I think there is always an issue of our self-identity as an individual. However, then there is the imprint we leave in the world, and I think those imprints are things much more likely to be immortal potentially than my identity as a self, which may just be a construction of my mind and not real to begin with.
Allison Duettmann: That is the other little problem, I guess. Okay, well, I am super curious. The recent explosion of interest in AI, which I think has always been strong in the Bay area, is being seen more in the mainstream. Frankly, most of my days are mostly spent with a radically refreshing, nonlinear library of AI and safety discussions. I think progress is accelerating in a baffling way. I wonder if you plug into this at all, or if you are, how may that influence your definition of life? I know on a different channel, you discussed this exponential population growth of AI life forms that we may be witnessing soon. I am super curious how the recent explosion in the intelligence field has influenced your definition of life so far.
Sara Walker: Yeah. I have been trying to understand exactly that question for a little while, and how we might be able to think about the transitions happening now and a longer view, if we contextualize them in the history of life on Earth and the future that we think life is going. I think about it in terms of major evolutionary transitions. So if we look at the history of life on Earth, people will talk about major transitions in evolution as being transitions in information processing and storage. You had, for example, with the advent of cells, the technology, the ribosomes, DNA, and digital storage of information, and then new information processing, modalities had emerged for multicellular organisms to exist, then societies, and now what we are seeing are basically multi societal structures emerging with the global integration of humans.
Basically, what I think is happening that is super interesting to me is that we are taking a lot of technologies that were invented very early on in the history of life on earth, if you want to take a technological lens on biology in the same sense that you take a bio biological lens on technology. We are now recapitulating them at larger and larger scale. If you want to think about the entire planet as a living structure, it is emerging new layers of complexity, and that is really what the technological transition is. To take a non AI example, you can look at things like eyes, right? Early in the history of life, no organism could see, and then photon receptors had to evolve. Then when you get to multicellular organisms, you had individual cells that could detect light, but then you get 70 different kinds of cell types that have different light sensitivities that form the human eye. From there, you then get to societies, and we still built things like microscopes and telescopes, so that we can see sort of different modalities of the structure of reality. Now, what we are seeing with the kind of technology we are inventing is taking patterns in the brain and trying to actually put them into technology. So I think if you take this sort of more global view, and you think about the globally integrated system, it seems clear that we are just emerging new concepts at a higher scale.
Overall, I think about like the Technosphere itself as a living system, and just saying that these AI are just different ways of like integrating that system at this global scale, so that life can persist at that scale. That might be necessary to actually getting off planet because if you think about a planet as a sort of a living structure, in order for it to reproduce itself, and to actually have another planet come alive, it might have to go through this technological phase of development. So those are sort of the ways I think about thinking about this sort of informational perspective on life, and what structures we are building now and how they lead to sort of more growth and open ended complexity and what that may look like.
Allison Duettmann: Very cool. It also reminds me of another bit that Mark Miller mentioned, which is this notion of civilization as a super intelligence. There has been computer intelligence for a while, which has been a part. Maybe the fraction of intelligence over time that gets contributed by machines is going to be larger than human’s contributions. Nevertheless, the larger structure is still sort of a nation at large. I though that was always super interesting. It does bring light to this more symbiotic notion of different races being more cooperative. It sounds like you have a pretty optimistic take of AI.
Sara Walker: I do. I do not see it as competition, in the same way I am not in competition with the individual cells in my body. I am a collective feature of them.There are some discussions that about AI or that the AI are competing with us as individual humans, rather than thinking about the AI as being an emergent structures of the interaction of societies of humans. So when I hear questions about existential risk in AI, I think more about, “Oh, we're trying to build an immune system.” So it is a question to me of what would be the planetary scale immune system for technology? Because that's based on how you would protect yourself for longevity.
Overall, I don't really see a lot of discussions in existential risk because it is very focused on this notion of the individual rather than thinking about these things as these kind of larger scale patterns that we are embedded in. We are not the pinnacle of evolution or the most alive structure as individual humans on the planet. Actually, we cannot even survive out of societies at this point. The same way that individual cells in your body cannot live outside your body, individual humans also cannot live out of human societies. I think the average lifespan of a human outside of society nowadays, because we don't have wilderness training and stuff, is about three weeks or something. So I do not think that we think about ourselves as collectives. But if we do, and we talk about AI at that scale, I think it changes the discussion of what it is.
However, I should also say to your point about this is very hopeful, I have a very hard time with negative future scenarios. My brain just does not work that way. I feel as though everything's hopeful. So I am one of those people that turns even the most pessimistic things into positive things. It is probably because I have this deep, intrinsic view that the narratives we build become the future, and because I think theories matter so much to what gets constructed that I have a hard time envisioning negative futures because I feel that is the easiest way to make them happen. So you cannot build something you cannot imagine.
Allison Duettmann: Yeah, I certainly agree with that. Last night, I listened to “We’re All Gonna Die” by Eliezer Yudkowsky and Bankless, which I think is the recent pinnacle of AI doom. It was pretty sobering, and he has good arguments. I think one thing to mention about your point about the more hopeful interpretation of AI is that there are some texts with Will MacAskill within EA where he does not classify a conscious AI outliving humans as an existential risk. At that point, that would still be something that is conscious. It is as though living potentially a life somewhat human-aligned. I think one worry that people have is that the AI we are creating may not be conscious of being sentient. So possibly, we are creating this “Disneyland” with extremely good optimizers, which is like building the Dyson sphere. There would be no one there to live and experience them. I think that is one of the risks. Do you have any of that at all, as in some sort of takeover that is a structure of a conscious or sentient life form at least?
Sara Walker: I guess I just do not see any concrete evidence that suggests to me that that is a real possibility. I think it goes back to like imagination and what's possible. I think that there are a lot of things that we can imagine based on our kind of primitive understanding of things. No matter where we are in history, we do not really fully understand ourselves or our place in the universe, or what the universe is. We have approximations, and we have adequate explanations for the time we live in. I do not see that being something that I feel is a physically real possibility. I just do not see, in the history of life, evidence of such things or evidence that we are actually building technologies that display those properties, even if they might be currently based on some optimization function. I just do not know how that manifests into a physical thing, not just an algorithm. Also, algorithms are physical for me, but they are not physical in the way that people talk about them in software. I cannot build a physical intuition around those scenarios. I do not see any consistency with them.
Allison Duettmann: I love it. I am not going to push on this further. I am going to hand it over to Beatrice now, who is really digging into the more Xhope part of this discussion. I like the way your mind is working on default for positive scenarios. I wonder if that is just a natural thing. Perhaps to close out: Is it a more natural thing where you end up thinking about the positive things more? Or is it a more specifically targeted because you think it is instrumental to not think about them very much?
Sara Walker: I think it is both. Naturally, my personality is disposed that way, but I have to take this meta-perspective on the things that I do. So I was doing what I was calling these thought experiments and the theoretical physics of theoretical physicists, which is sort of along the lines, again, of how David Deutsch thinks about explanations. However, a very specific example is the invention of Newtonian gravity, which is a particular kind of abstraction, a theory, that emerged on our planet after many billions of years of evolution, and then many millions of years of human evolution, and 1000s of years of modern thought. You can actually trace the whole lineage of things leading to Newton's idea of gravity. This is a sort of regular pattern that we see persisting in our environment, and then what you see downstream of that is that we build things like satellites, and new technologies emerge out of that understanding.
Because I think of theories as being these properties already that emerge out of evolutionary structures, and then they impart are actually what steers and builds the future, I have to think about constantly correcting my structure, which is life. I am doing the same thing by trying to build theories to understand life. That sort of transition kind of constantly is forcing me to have this positive outlook. I want to make sure that whatever theory we have that describes us also helps perpetuate that we have some longevity associated to us. I actually have a sort of theory building of not Occam’s razor, but instead a theory building of positivity. What are the theories? What are the explanations we can give ourselves to build the largest possible future?
Allison Duettmann: The Walker razor. I like it. That is a great point. We are ready to hand it over to Beatrice. Thank you so much; this was very fun!
Sara Walker: Thank you.
Beatrice Erkers: Thank you so much for joining Sara. It seems like you have answered a bunch of my questions already. You seem very positive about the future, which is usually what I ask. Also, your work seems quite existential in a lot of what you have talked about so far. It is very interesting to hear about your background and origin, so I am curious to hear what you say about the future. So you seem very connected to these questions, but as we want more people to engage with these questions, I was wondering if you could share if you have ever had a particular experience that made you feel excited, connected, and hopeful about the future.
Sara Walker: Maybe not a particular experience, but I think something that has been characteristic of my personality, and also throughout my career, is when people say things cannot be done. I just want to demonstrate they can be done. When I was a postdoc, or actually when I was a Ph.D. student coming out of a cosmology Ph.D. group, I was trying to figure out how can I get a job as a postdoc setting origins of life. In particular, as a theoretical physicist, I had no idea how to do this. A lot of people advised me not to work on the origin life because they thought it wasn't a solvable problem. Instead of backing away from that, I thought, “Well, isn't this great?” No one knows how to solve it.
Therefore, no one's actually asking the questions about how to solve it. If I start doing that, it might not mean I am going to make progress, but it means there is a space there to actually work in and try to do something. If no one's doing it, it is obviously not going to happen. That would be a self-fulfilling prophecy, so I just wanted to try to do it. I think that that is sort of the way I like to work. I like to take things people think are not possible or look at things in ways that people have not before. So there might be a narrative people are using to describe something, and it seems to me as if that is a closed scenario. It is not really helping us see new ways of thinking about that problem that opened up new spaces. Overall, I like to play with ideas and just taking negativity and turning it into positivity. I do not know if that answers your question.
Beatrice Erkers: I think it does. It provides a lot of context on what makes you feel excited about the long-term future.
Sara Walker: Yeah. I guess for young people who want to go into fields and people tell them it is not possible: I say if you are super passionate about that and you think there is something there that other people are not seeing, then it is sometimes worth going for.
Beatrice Erkers: Yeah! Are there any areas or technologies that you think are undervalued right now that more people should pursue?
Sara Walker: I actually think the origin of life problem for technology is deeply undervalued, and of course, that is coming from some bias. I have been working for a while with Lee Cronin developing this origin of life theory of life, but also on this idea of whether or not we can build an origin of life experiment. The idea is to see it as a search problem through chemical space though, not computer algorithms or a search engine. Instead, can you actually think of the space of molecules and actually build an evolutionary engine that operates in chemistry and searches for new life forms? The reason I am so excited about that idea is that I come from a tradition in particle physics of theory motivating these large-scale experiments to prove fundamental aspects of our universe, and we do not think about the origin of life that way at all. It is the same kind of scale problem in terms of physics.
However, if we solve that problem and we really know how the universe generates information, and then it generates novelty based on information, I think that is a much more embodied way of thinking about what software is and what algorithms is and how they emerge in the universe from this sort of disembodied way we currently talk about it in technology. Overall, I see this really interesting future in technology that is much more embodied if we can solve the origin of life problem. Whereas right now, I think the reason we do not understand technology is because it is so abstracted from the physicality that it is really hard for us to even know what we are creating.
Beatrice Erkers: I am interested in your use of the word “embodied.” For example, that is one of the things about the whole concept of how we think of existential hope at Foresight. We think about how we can make people feel a more embodied vision of the future. How do you think about using that? Also, do you have a specific vision of existential hope for the future?
Sara Walker: Actually, I think I have kind of a different idea of what's physical to most people. So most people think matter is physical and atoms are physical, yet algorithms and mathematics are not physical, and they only exist in some platonic world. Personally, what I think is physical is actually what most people will call informational. For instance, if you look at complex matter, the actual physics that governs things like us in complex matter is how those objects are constructed across time and what the necessary operations are for building those objects. For me, that's actually the underlying structure, the physics of life. So if you take something like a molecule, most people might look at the molecule as a physical molecule in a three-dimensional structure and weight’s mass. Contrarily, I would say the actual physical structure of that molecule, that's relevant for thinking about it as an evolutionary object and an object constructed by the universe, is all the ways that the universe could build that molecule based on the physics of our universe. That then becomes the object in the theory.
So I have this sort of view of embodiment, such as what is physical in this kind of temporally extended sense, of understanding where objects come from, what are the ways of making them, and how that actually makes new things come into existence. I don't think that is really translated into something that we understand in general physics yet, although we are trying to work on that. I think if we understand that, then there is some capacity to steer our future in some sense. We have this idea that biological systems have agency and goals, but we do not know how to impose those things as a planetary-scale process. So I wrote this paper with some collaborators on this idea of planetary scale intelligence, which is focused on how the planet as a whole does not act like an intelligent system. However, you can see some rudimentary processes that might suggest it could be an intelligent system. For example, people think climate change is a really bad thing, and of course, it might be an existential threat. Yet, what is super interesting about what's happening with climate change is that we are modeling possible futures for the planet in a way that we are trying to respond and now steer those possible futures.
So if you think about the earliest emergence of life and emergence of agency in systems evolving goals are understanding what the future space of possibilities is and trying to steer among those, then you are starting to see some of these features of life, again, occurring at a planetary scale. So I would say my existential hope is that life is really an open-ended system and that we will really see these features reemerging at different scales. I hope this will continue to persist into the future and generate more things and more possibilities, and hopefully, also improve the quality of life. I feel like solving the origin of life is a critical transition in the evolution of any biosphere because the biosphere learns about its environment, learns to see, learns to hear, learns to think, learns to invent tools and technology, learns something about the laws of the universe that work it lives in, and then it learns something about itself, which then allows it to do things. It is just super fascinating to get closer and closer to understanding what you are. The universe is doing that through us. It is just all super interesting. I don't know what it looks like on the other side of that transition, but I want to help make it happen!
Beatrice Erkers: Thank you. I am a bit curious, then, because you and Allison just discussed this thing about sentient AI not being considered an existential risk or threat. Rick, in the chat, had a question: Is consciousness all that matters? Or would you be okay with humanity being replaced by AI life that does not have maybe the same values we do?
Sara Walker: I am pretty sure I am conscious, so I share the biases of most conscious things; at least, I think I'm conscious of what I am saying to you guys right now. With that, I guess I kind of hope there's an inner world. However, I also think this framing of consciousness is always focused on this inner subjectivity and not so much like what consciousness does. Allison mentioned Dan Dennett earlier, but I really liked this paper he had on what the hard question of consciousness is, what does it do? I had also been thinking for a while about that question. For me, the thing that's super interesting about human minds is our ability to imagine again, counterfactual possibilities and then make them actual possibilities. I cannot help but think like, if you really want to talk about consciousness as a phenomenon that exists in our universe, it's not really about what's happening inside an individual mind. But if you have a lot of lights on and a lot of physical systems, then that actually has more future generative potential than not.
So I think if you looked at the space of selected futures in the broadest possible spaces, I think consciousness would actually be a key feature of that. Again, this is the optimistic narrative, but for me, that's one of the reasons that I am really interested in understanding consciousness more as a property of collectives and interactions of collective agents than thinking about it. The intersubjectivity matters, but it matters to particular dynamical outcomes that couldn't happen without that. I do not see a lot of framing of discussion of consciousness in humans or machines from that perspective, but I think this is really what people are starting to get at with like human-AI interactions, in the sense of the conscious experience, maybe between the two entities and not in one individual. I think those kinds of things are really interesting to think about.
Beatrice Erkers: Thank you. I have not heard that perspective, but that is interesting. We are jumping between questions. You touched quite a lot on how thinking of positive outcomes for the future is instrumental. It seems in the sort of general narrative of the world, it often ends up in dystopian visions. What do you think we could do to change that?
Sara Walker: I go back and forth about this. I'm actually not sure we need to change that. I do think we should inject more positivity and more creativity about what the future could look like, for the reasons I stated that we need to build it. But it's also in some sense that, when people are talking about existential threats and working on those problems, I am almost no longer worried about those threats because someone actually recognizes that they are a possibility, so they are working on trying to steer us away from it. Instead, it is the things that we are not thinking about that I'm worried about in some way.
I think the fact that we're having those discussions, in some sense, is a protective mechanism of trying to understand. We have to talk through what the possibility spaces are, which includes talking about the positive and negative and what we value as positive or negative. We could not do that unless we were trying to build dystopian scenarios as well as utopian scenarios. Probably what we really want is something in between, because neither of those sounds like the real optimal. I would hope for just more discussions about the fact that we are actively steering it and more awareness of the agency that we have and what agency we want to give away to technology.
Beatrice Erkers: Yeah, okay, so more awareness of the fact that we are steering it and able to steer it. It sounds like then we are talking about people working on existential risk. But would you say that even in the general public, it doesn’t have to change in that way?
Sara Walker: No, I mean, I think change is always good and asking questions about change. Instead, I would ask, what is the value of having these discussions besides what we are saying? For example, I thought it was really funny when Elon Musk took over Twitter, and there were so many people talking about Twitter on Twitter. Twitter became very meta about itself. All the conversations were about how horrible this is. However, at the same time, if you look at Twitter as a collective platform, suddenly, it's having this sort of self-referential discussion about itself that it never had before as a collective.
So I think just recognizing the different levels that we're having these discussions and what the discussion is at the level of how we're talking about it between us and what we perceive of it, versus just stepping outside ourselves and looking at what is this collective discussion saying when we look at all the sides, is just critically important for understanding what we are doing when in large groups. Trying to take that meta-perspective of looking at the outside, and not just looking at our personal opinions and how we're arguing with each other, is critical for understanding how these systems behave because our opinions are very biased.
Beatrice Erkers: Yeah, well, thank you so much. When we talk about existential hope, we always bring up this thing of eucatastrophe, which is the opposite of a catastrophe. For instance, when there is an event, the following outcome would be of positive value. We always mention that it is a terrible name because of its similarity to catastrophe. Do you have any suggestions for a better name? It is okay if you don’t.
Sara Walker: That's a tough one. It would almost be better if you described what it was but didn't give me the word because now I'm trying to permute that word. You should try that experiment on someone else in the future. That's my suggestion. I don't have one. If I think of it, I will let you know, though.
Beatrice Erkers: I will try to rephrase it for another interview then. So, you already shared your existential hope vision. In this podcast, we always try to create a visual art piece on an existential hope scenario. I was wondering what you think would make a cool visualization. Do you have a vision, or a prompt of a vision, for the future that you think is quite inspiring?
Sara Walker: Yeah. I am very romanticized by this idea of life across all the scales of complexity on our planet, all the way up to the planetary. I enjoy thinking about this integrated structure on a planet where some features of it are 3.8 billion years in depth. We think we just exist in the current now on this planet, but it is inspiring to think of that entire structure. I also think it is interesting to think about what are the transitions that are building the future. This sort of open-ended possibility space is that planet, the biosphere emerging on that planet. The Technosphere understands more about itself and more about the universe. It lives in and can expand beyond that planet.
So as I said before, I think the kinds of transitions we are undergoing now are critical, although I am biased by my time and place in history. We are emerging all of these new layers of complexity and technology that we don't understand, but it is recapitulating at these different scales. At the same time, I think historically, we are at the moment where we should be ready to solve the origin of life and chemistry. If you look at those two things as our deep history, we are finally understanding that first moment of life, and we're recapitulating life now, reemerging at this different scale. It is just this really interesting loop across our history of what we're doing, and where we are going. I think it sets up this amazing, hopeful view of the future of what the possibility space is when we cross that threshold.
Beatrice Erkers: I love thinking about our origin more and using history to understand our future a bit more. The next question would be, based on everything you have told us here today, do you have any reading or listening recommendations? Where can people start diving in?
Sara Walker: Yeah, well, there was a podcast that Allison mentioned if you want to follow up on some of the ideas I talked about. David Deutsch came up a few times; I love his books. Paul Davies was my postdoc mentor, so I learned a lot about how I think about reality from him. He's a pretty open-minded person. He's written like 30 books and always tells me I need to catch up. He has a newer book on information and life that I think is nice. I also really like Ted Chiang's writing from a science fiction perspective. “Story of Your Life” is one of my favorite stories about alien life, in part because I'm obsessed with time being related to what life is. That has such an interesting, deep relationship between the different kinds of intelligences and minds and how they perceive time. There are tons of great stuff out there.
Beatrice Erkers: Thank you! I think those are all great. I love Ted Chiang as well. We have time for one more question in the chat. Creon was asking if you have studied the works on active inference, Markov blankets, and the free energy principle.
Creon Levit: Yeah, and not just have you studied it, but also, does it inspire you or change your research program in any substantive way?
Sara Walker: I am aware of it. I've read a few of the papers, and I have spoken with some people that work on those theories. I don't find it particularly useful myself because I think it seems ad hoc to me in some ways and not so natural. I think when you get to brains, and you are trying to understand and talk about brains outside of their evolutionary context, then some principles of that make sense. I think that was the direction that Karl Friston was coming from originally when he was proposing some of these ideas. However, I think when you're trying to build a physics from the bottom up and understand the origin of life, I don't think it's so helpful because it has a lot of, what I think are, unnatural assumptions that might actually be products of evolution. There are a lot of observers already in that theory, and this is a problem with a lot of theories.
Creon Levit: Okay, fair enough. I think actually that it is a little more applicable to biological and nonbrain systems than then you might think, but fine. I mean, you get to choose, and you're an expert. I would wonder maybe the same question about information theory in general. I see deep connections between information theory, origin of life, and evolution of life.
Sara Walker: Yeah, all these things are connected. Also, I should say there are some papers applying it, and those papers are interesting. But again, I think there's the issue of how do you build the experiment. How do you actually connect this to the properties of chemistry versus assuming you are labeling things and you can describe them in some probability distribution and information-theoretic way? So I have, in my own career, been a long-term advocate that somehow, something in the space of information or properties is related to the foundational physics of life. I think all of these things are touching in the right direction, but I do not think that they're formulated in the right way to describe the physics accurately.
Part of the reason information theory, in particular, is hard is because to do any kind of information theory, you actually have to build probability distributions. As such, you already need to be talking about events that are frequent enough to actually construct a probability distribution from. A lot of the physics of life that I've described that I think is most important is actually the mechanisms of generating novelty, so you basically want to talk about the first creation of something in the universe and then how it gets to be a highly reliable and probable event. Once it's a highly reliable, probable event, you already have this physical structure selected, and then you can describe it with information-theoretic principles. You can build probability distributions over how likely it is to happen. But still, it doesn't tell you how you generate those things in the first place.
What we're trying to do with the kinds of theories that we are developing with assembly theories is actually unify the novelty generation mechanism with the idea that you have objects in abundance. How do you actually bridge those scales from the universe's first creation of an object to when that object becomes a reliable feature of the universe? So from the first incarnation of the earliest idea of a cell phone all the way to everybody on the planet having a cell phone. Information theory might tell you properties of the networks of things that are generating information for cell phones on the planet today, but it would not be able to predict the emergence of a cell phone from past history, going all the way deep in time. So there are some features there that I think are missing from those kinds of descriptions.
Creon Levit: Is it another way of saying that, while the whole concept of novelty is arguably an information-theoretic concept, or at least can be used to analyze the idea of novelty in a formal way, just because something is novel, doesn't mean it's going to take off?
Sara Walker: Yes. That's right. I think about it in the sense that there is a possibility space of things that can exist, and things are being pulled into existence by the biosphere over time, and then there are reliable structures. This is actually what the evolution of life is, so you need to be able to unify that mechanism of basically generating the first instance ever in the entire universe of an object. Then that object can be produced reliably because there are now systems that know how to build it that were selected. The physics of life is so radically different than any physics we built before. That is one thing that is underappreciated, like you guys asked me. That is one. Everybody thinks life exists in known physics. No, but these are great questions Creon, as usual!
Creon Levit: I'm going to be quiet after this last one. Novelty itself, and the generation of novelty, is an interesting question. Perhaps in your view, is it that it is kind of necessary but not sufficient to explain what's going on because novelty doesn't mean that the novelty will be selected?
Sara Walker: Exactly. So novelty is insufficient, and abundant objects are insufficient. You need something that unifies those scales. That is actually sort of one of the main things that we have gone after with the stuff we do in assembly theory. We have ways of talking about those two things and unified dynamics. In order to do that, you get these new concepts of information in time that haven't existed. Also, I should be careful when I say information. I am always using that word as an approximation to a set of things that we associate with information in a variety of different linguistic meanings and also causation. So for me, those two concepts are the same, and some features of those concepts now overlap with what we usually talk about as matter and time.
Overall, language is hard because our language is like these rough patches of things that are out there, and we're moving around the meanings of words. This class of things that we described as informational is a really interesting class of things. So people sometimes mean information in a formal information theoretic sense, and sometimes they mean meanings, and sometimes they mean information technology. So there's just all these different sorts of ways of meaning it. What I really think is important is that life seems to be generating structures that are so low probability that you would never expect them a priori from standard physics. It seems to be able to do that in such a way that it not only generates those structures but then learns to produce those reliably. That feature, I think, is about causal structure and causal dynamics in the universe, and it is also associated with information. You would need a theory that describes those features.
Beatrice Erkers: Thank you so much for diving into that, and thank you, Creon, for continuing the interview. I think if you have 30 seconds or so, it would be a nice round-off if you could share the best advice you have ever got.
Sara Walker: Oh. It is very, very funny advice, but I got this advice when I was very young. My mom was very much into natural healers and Reiki and stuff. I think I had to take Reiki one or something when I was a teenager. The guy that was doing that with me said it's really important for you to learn those things but also unlearn what you have learned, so you can think about them in new ways. I think that was actually one of the best pieces of advice I ever got.
I have gone through entire formal training in physics, and what I learned is a very rigorous way of thinking about how reality works. I have also done it in such a way that I'm willing to forget, historically, any feature of what we learned in case it might be wrong, but I am also willing to try to rebuild it again in as rigorous a way as possible. I think that just stuck with me in some weird way. I love getting advice from completely unexpected things. You never know what you are going to learn from people, even if, you know, it is just really unexpected.
Beatrice Erkers: That seems like a great approach to life in general. Thank you so much for coming and sharing all of your perspectives with us today. It is really kind of you to share and very interesting. Thank you so much for this very positive outlook on the future. I feel revived.
Sara Walker: I love the idea of existential hope so much, so I was very pleased to participate. Thank you!
Recommended Reads Mentioned:
Talks and Videos: