This Man Thinks “AGI” Could Allow Humans To Live Forever, With Dr. Ben Goertzel

Ben Greenfield

My name is Ben Greenfield, and on this episode of the Boundless Life Podcast... But how accurate do you think the current biological testing landscape is?

Ben Goertzel

Oh, not very. When I really looked at the data underlying any of these biological age estimates, I mean, it's a little dodgy. I mean, we've all met people in the longevity field who are like, well, I'm 75, but I have a biological age of 43. But you're like, well, why do you look so And then they get a heart attack the next year, right? The long-lived flies have several SNPs in a certain gene that the short-lived flies don't have, and you can map that into humans.

Ben Greenfield

Welcome to the Boundless Life with me, your host, Ben Greenfield. I'm a personal trainer, exercise physiologist, and nutritionist, and I'm passionate about helping you discover unparalleled levels of health, fitness, longevity, and beyond. My guest on today's show is a cross-disciplinary scientist, entrepreneur, author, and a musician. Stay tuned for the end of the podcast for a very interesting anecdote about his band. His name is Dr. Ben Goertzel. He's the founder and CEO of SingularityNET. It's a decentralized AI network. He's the chairman of the Artificial General Intelligence Society. He came up with the term AGI, invented that. He's the chief scientist of robotics from Hanson Robotics and has a PhD in mathematics. So we have a wide-ranging discussion, especially as it applies to AGI, AI, and longevity. So all the show notes are going to be at bengreenfieldlife.com/AGIpodcast. Let's go talk to Ben. So Ben, I, until like a month and a half ago, hadn't heard that much about you. And I was at dinner with one of my buddies and he's like, you got to check out Ben Goertzel. And he's got all sorts of super fascinating work, particularly related to AGI, Artificial General Intelligence, and health, longevity, a lot of the things that we talk about on this show. So I left that dinner and I looked you up, including your interviews with Lex Fridman, with Joe Rogan, a few other shows you've done, and realized that I must be hiding under a rock because it's a very prolific guy out there with the same initials as me, but way smarter, doing some pretty cool stuff. So welcome to the show, man.

Ben Goertzel

Yeah, yeah. Good to be here.

Ben Greenfield

In my stalking of you, I saw that your dad actually apparently spent years writing a biography of Linus Pauling. He's come up on the show before related to like vitamin C. Was that kind of your dad's interest in that? What got you interested in nutrition in the first place?

Ben Goertzel

No, although that was an issue that was sort of around our family when I was a kid. In fact, that book on Pauling was my grandmother's passion, my father's mother. And she worked on that book for a long time because she and Pauling were involved together in the nuclear disarmament movement in the 1960s. Then she got Alzheimer's and started to lose her mind and couldn't finish the book. So my dad stepped in to finish the book for her. I mean, he's a political sociologist. Pauling wasn't his main thing. Then my dad pulled me in to sort of rewrite the parts of the book about Pauling's chemistry because neither my grandmother nor my dad knew much about chemistry. The thing with Pauling's interest in orthomolecular medicine and the potential health impact of vitamin C is my grandmother didn't believe it. She thought it was a bunch of quack nonsense. So she was disappointed like this great scientist she had admired had like gone down a fucked up path of promoting vitamins to cure everything. So she was like, Pauling says he never gets sick because vitamin C. But when I when I called his assistant yesterday, she said, No, he's a little bit under the weather today.

Ben Greenfield

Vitamin C never gets sick. He just gets under the weather. You forgot his vitamin C that day. It is interesting, though, because I think I think it's kind of nuanced, like it's such a potent antioxidant. There's been some studies that have shown that it can blunt, for example, the hormetic effects of exercise, like if you take high dose vitamin C.

Ben Goertzel

Yeah. So I guess in that context, which was a long time ago, I mean, there would have been like the early 90s, I get in that context, I did look at the data about vitamin C. And what I saw was, it was more positive than my grandma was thinking, like it actually did have significant value. On the other hand, the dose is a is a vexed question, like some of the dosing that Pauling and others were suggesting, may not always actually be beneficial to take, take that much. So it was it was interesting to look and I think one of the one of the main habits or lessons I got from my dad over the decades, and we're still we're working now on a book about the social and economic implications of AGI, actually, but you and your dad are? Yeah, yeah. But one of my dad's characteristics is, he's very willing to look at the data and revise his beliefs based on whatever, whatever the data say, like, it's not, it's not like he doesn't have emotions or biases, but he's, he's just willing to look at the data tables, look at what experiment was done. And, you know, then he was then he was wrong.

Ben Greenfield

So I It's the mark of a good scientist on a constant quest to prove yourself wrong. That's kind of what science is.

Ben Goertzel

Yeah, it's very rare in the political sociology field, which is where he spent his career, actually, because that's just very, very ideological, because the data is scant and ambiguous. And in the vitamin C thing, my dad and I sort of took that perspective. And then we ended up when we wrote the biography, just not being super decisive about that, because it sort of seemed like the jury was out like, yeah, taking a bit of vitamin C is good. Taking a huge dose is probably not always good, but could it be good for some people in some circumstances? Like, maybe, maybe? Yeah, maybe it could be right. So but I, I think I got more into the alternative medicine side, many years later, many years later than that, actually. And it was partly due to personal life stuff for like, my first wife had Lyme disease, and none of the standard things got rid of all the like, long tail of Lyme disease. And some, some, you know, out of the mainstream therapies are helpful. But then, from looking at the long lived fruit flies that a company called Genescient evolved, these are fruit flies that live like, at that time, two or three times as long as normal flies. Now we have ones like five or six times as long. But you can, right, you can see, like feeding different combinations of traditional Chinese herbs to these flies could make them live a long time, you can improve their hearts, improve their brains can, can totally overhaul them, right. So I mean, you can see that a lot of these sort of holistic medications, in an animal model where you can test everything really rigorously, you can see they're doing just right, you can test everything really rigorously.

Ben Greenfield

And in the case of fruit flies over multi generations pretty rapidly.

Ben Goertzel

Yeah, yeah. I mean, 10s of 1000s of generations, right? So yeah, yeah.

Ben Greenfield

Interesting. So when it comes to AI, which I know is something you've done quite a bit of work in AI and AGI, like, how do you how do you go from, from from generating ideas about a certain nutraceutical or compounds to applying that in a laboratory setting? Or do you can you replicate a fruit fly or a human model?

Ben Goertzel

We don't have that simulation model yet. And that's, that's really the sore point in terms of applying AI to longevity, overall, right. So I mean, what we can do, we can crunch all sorts of biological data about flies or mice or humans. And I mean, so you can, you can look at genomics data, proteomics data, metabolomics data. I mean, you can look at tissue specific data, flies at different points in their life cycle, and so on for people, people or whatever, right. So I mean, you can, you can do super interesting data analytics, and then you can feed in chemical information about different compounds. But then, at some point, once you have the idea, like, okay, the way to, you know, improve the fly's lifespan, or the way to get rid of Alzheimer model in the fly or something, right? Once you, once you think the way to do that, is to carry out a certain gene therapy, or a certain combination of nutraceuticals, or a certain small molecule, like, then pretty much you still have to try it out on the real organism.

Ben Greenfield

So when people talk about like a digital twin, is that just who we are? Is it just in its really like embryonic stages of development? We just don't work well enough yet for this sort of thing.

Ben Goertzel

Because I mean, you can, I mean, we don't have a digital twin, even of a nematode worm fully yet, let alone of a fly, a mouse or a human. I mean, there are simulation models of specific, like, biochemical pathways and specific subsystems. But I mean, in the end, you still have to try on the real organism. And that's, that's what I like about the fly model for longevity, of course, is, I mean, if a control fly lives a few days, and one of our long live flies lives a few weeks, I mean, then you can actually test longevity, like did it double it? Did these herbs double its lifespan? Did these herbs make it, you know, run mazes better when it got old or something, right?

Ben Greenfield

But talk to me, talk to me about applicability of that to say, you know, then a rodent or a primate or a human model, like how closely does genetics or biological systems or something like the fly model replicate that?

Ben Goertzel

This one of my favorite AI applications in the domain, right? Because there's something in the field of AI called transfer learning, which is how do you take a model of one thing and transfer it tells you something about the model of something else. So like, if you're, you know, if you're a champion at Othello, does it help you at all like Go, right? Can you transfer a little bit of your strategic knowledge and as a human, you can transfer a little bit, right? Like if you're, if you're a human who's a master in Othello, I mean, that doesn't make you a great go player, but some of what you learn transfers over as opposed to if you were starting from absolutely nothing, like there's some high level lessons, like, you know, increase the mobility you have, you have in the mid game or something. Right. So similarly, I mean, like take another example, if an AI can drive a car, does that give it a head start in learning to drive a truck? Right. I mean, for a human, right. It obviously gives a head start. Like there's more learning you have to do to drive the truck, but your car driving ability can transfer over to some degree. So in the same way, taking therapeutics that extend the lifespan of a fly and trying to transfer them over to a mouse or a human, I mean, it's not just straightforward, but it is a definable transfer learning problem. And this something that modern AI algorithms, you know, can be reasonably good at. And you can see some of the specific biological dimensions of that very easily because flies, for example, very rarely die of cancer. I mean, they can, but it's just not a major thing. So a lot of things you would do to prolong life in a fly, they might give cancer to a mouse or a human. And this ties in with the multifunctionality of many genes like p53 in humans, which are their tumor suppressor cells, but they also serve other functions.

Ben Greenfield

Right.

Ben Goertzel

So like if you, if you mess with how that gene functions in a fly, it may just be fine. It may prolong the life in a human. It may muck with the other stuff that gene was doing, suppressing tumors. Right. And then, yeah.

Ben Greenfield

So is, is the absence or the, the at least lower incidence of cancer in flies due to their shorter lifespan or is it something else?

Ben Goertzel

It doesn't seem to be just a shorter lifespan because our long-lived flies don't die of cancer either. And we've increased the lifespan a lot.

Ben Greenfield

Right.

Ben Goertzel

So we, I mean, that fly's body is just quite different and I'm not a, I'm not a Drosophila biologist. So I I'm not the best person to answer that in detail. I actually, you should interview Kennedy Schaal, who is the CEO of the company Rejuve Bio that we spun off from my AI project Singularity. And they're, they're doing all the therapeutics development and bioinformatics on the long-lived flies using our AI tools right now. Because I've learned just enough to be dangerous about Drosophila biology, but I'm not a real fly guy.

Ben Greenfield

Yeah. Yeah. You know you brought up the disease Lyme as well as a treatment like vitamin C when it comes to AI or AGI and feel free to differentiate between the two because we really haven't contrasted them before on this show. Where, where do those fit in beyond just say testing something in flies when it comes to, to assessing how we might treat a certain disease or how we might like, like test a certain nutrient? Like what's the actual process?

Ben Goertzel

So, yeah, I think there's the current state of things and then there's how I think things will be in a few years when we've made more progress toward actual AGI. So yeah, let me start by saying a little bit about AGI versus narrow application specific AI, and then we can see how that manifests itself in the longevity therapeutics domain, right? So when the, when the AI field started formally in the 1950s, I mean, informally a couple of decades before that, they were trying to make machines that would think, you know, with a grandeur in generality of human beings, and then even beyond that, like you see in science fiction. And this includes the ability to generalize wildly beyond one's training and programming, like to deal with the great unknown, right? Like in, in Star Trek, every week they go to a different planet, they deal with a different kind of alien civilization. You've got to be imaginative and go beyond everything you've learned. As the AI field evolved during the 70s, 80s, 90s, people realized that was harder than they thought. So AI became more a collection of specialized applications, like one AI for driving a vehicle, one AI for playing chess, one AI for predicting the market, one AI for diagnosing diseases. And even in that area, in that era, rather, you had specialized AIs that could do, you know, disease diagnosis better than human doctors and so forth, right? So, I mean, there already was significant successes, but I spent a lot of time in the early aughts and onward trying to reorient the AI field back toward the big picture problem, which I called AGI, artificial general intelligence, which is a term I launched in a book in 2005, which now has become very popular, right? Like open AI, AGI is, is, is upon us. But in 2005, when I wrote the book titled AGI, it was a very sort of obscure topic because most of the AI field didn't believe it was, it was real, right? And the manifestation for longevity medicine or medicine in general is like, once you're into the full-on AGI era, you should have AIs that just act as autonomous creative research scientists, right? Where the AI can, you know, come up with a theory, flesh out the theory by looking at the literature, design an experiment, run an experiment, analyze the results, you know, lather, rinse, repeat, right?

Ben Greenfield

Right. And continuously self-improve and self-evolve along the way, right?

Ben Goertzel

Yeah. I think we're not far from that, but we're not there yet, right? So, I mean, in the exact current practice today, it's more like as a longevity researcher, you're using an assemblage of different AI tools along with non-AI tools for bioinformatics, data visualization, whatever. And the AI tools are valuable parts of your toolkit for designing therapeutics. And then there's a number of different AI tools you can use, right? And so, I mean, there's neural nets to do machine learning where you can crunch a dataset that came out of an experiment and try to find what are the patterns in that dataset. What's that telling you about what causal pathways were revealed in the experiment? But one other thing we've been experimenting with a lot in our work at Rejuve Bio and then SingularityNET is using machine reasoning for discovery. So, it's target discovery and it's discovering new sort of combinations of therapies. So, this not so much running experiment and crunching the output using machine learning, although you can do that too. This like, okay, we have some vague idea as to maybe these pathways play a role in longevity. Can you dig through all the literature and think about it and come up with some new ideas about what might be some causes of longevity related to these datasets we loaded in, related to these papers we loaded in? And the AI is like doing the creative thinking, trying to come up with some new hypotheses.

Ben Greenfield

A lot of people would say, well, I could do that right now with any of the latest Claude or GPT models.

Ben Goertzel

You can do that. They're just not as good at it as our...

Ben Greenfield

When you say as good, do you mean as fast, as thorough or something else?

Ben Goertzel

No. What I mean is that these models like ChatGPT or Claude, and I mean, they're amazing. The ASTRA model is even more amazing. These are super brilliant things, but they do stick relatively close to their training data. I mean, all of these large transformer neural net models, which is the kind of AI underlying LLMs. I mean, these are all basically huge models of everything on the web. And I mean, that does very, very well in certain domains. Like if you're doing Python programming, there's so many useful Python programs on the web, like modeling all of those is great. What you find in something like mathematics, let's say, and that's my original background and my PhD was in math. I mean, they're very good at proving math theorems within known branches of math. And if you're trying to go out on a limb into a branch of math, that's new or not much is known, they're much less good, right? They're not retarded, but they're much less good because there's just less training data on the web about these weird, obscure little branches of math. And they can transfer knowledge from well-known branches of math to new ones with scant presence on the web, but they make a lot more mistakes. And with something like longevity biology, what you find is they just come back to what's in the literature every time.

Ben Greenfield

And are you contrasting that with AGI, saying that AGI is information generative?

Ben Goertzel

I'm contrasting that with a system that we have developed in SingularityNET and our sister project, BGI Labs and Rejuve Bio. I'm contrasting that with this probabilistic logic reasoning system called PLN, probabilistic logic networks, which we use together with LLMs because we have a different sort of knowledge graph. We fill it up using an LLM, having the LLM read a bunch of research papers and interpret a bunch of datasets and then spit knowledge into this knowledge graph that we have. But then we can do explicit probabilistic logic reasoning, like different kinds of AI thinking on this knowledge graph. And what I find is these different kinds of AI logical thinking are better at taking leaps beyond what's in the literature. So if you're looking for what does the literature say about this pathway causing brain aging, LLMs are super brilliant at that. But the nature of trying to do frontier science is exactly you're trying to go beyond what is in the literature. And sometimes that will contradict what's in the literature, right? And then you really have a hard time flogging the LLMs to question what's baked into the weights of the model and go in some other direction. So, I mean, to take a physics example, like if you trained an LLM on physics before quantum mechanics, I mean, you're not going to get that LLM to agree with you on the invention of quantum mechanics, let alone help, right? Because it's got baked into its weights, how the universe works, like it's a deterministic classical system, and it can't revise those weights during inference. So I think the thing is, Sam Altman could be right, that models like Astra could do 90% of human jobs or something after appropriate system integration. On the other hand, breaking radical new research frontiers, that would be in the other 10%, right?

Ben Greenfield

Yeah, it sounds to me like when you're describing probabilistic knowledge, it would be the difference between a human reading in a consumptive fashion a book versus theorizing in an imaginative fashion after reading the book.

Ben Goertzel

Yeah. On the other hand, I mean, every one or two-year-old child is theorizing in an imaginative fashion all the time, just to figure out stuff that's radically new to them, even though it's obvious to us, right? So we're concerned about the high end here, because that's where there's economic and scientific and humanistic value. But on the other hand, the process of leaping beyond your knowledge into the great unknown, that's how we grow up in the first place, right?

Ben Greenfield

Yeah. So boots on the streets here from a practical application standpoint, when I was having dinner with my friend, he's actually a former podcast guest, it was Matt Morgan, who's been on the show before. He said, well, Ben is solving many of the problems that we currently face in human longevity with AGI, potentially even radically extending human lifespan with application of what can be learned via AGI. What exactly are you doing that might result in a meaningful increase in human longevity?

Ben Goertzel

I think the most concrete and specific longevity project we're working on now in my team has to do with these fruit flies that I've already mentioned, which were originally experimentally evolved at UC Irvine. Then they were taken on by a company called Genescient, and now we're working on them within the Rejuve Bio, a spinoff of SingularityNET, because we already know, by using AI to model the genomes of the flies and connecting that with chemical databases, we found novel combinations of traditional Chinese herbs. You give them to a middle-aged fruit fly, they can more than double its life, right? And that's just herbs. It's very clear to us, if you can do a gene therapy and just twonk the same genes that the herbs are twonking, you could have an even bigger impact on the fly's life.

Ben Greenfield

When you say Chinese herbs, how many different herbs are we talking about?

Ben Goertzel

Five, I guess.

Ben Greenfield

Are we talking about also herbs that people would be generally familiar with?

Ben Goertzel

I mean, some, like Astragalus was one, and some were obscure names that I can't remember.

Ben Greenfield

Actually, I've come across Astragalus before when it comes to potentially decreasing the rate at which telomeres shorten. Forgive me if this a dumb question, but when you're saying five Chinese herbs, I'm assuming that these were selected or chosen or suggested by an AI or AGI model before application?

Ben Goertzel

Yeah, yeah, yeah. We fed in a big database of traditional Chinese medicine, and for each herb, that database tells you what are the known active chemical ingredients. Then you can look in PubChem chemical database. It will tell you of those chemical ingredients, like what proteins did they dock to, right? And then you can see what genes code for those proteins, and the AI can connect the dots. And what was fascinating there is, in traditional Chinese medicine, they have a certain freaky science telling how to combine these herbs, and the AI combines them in a totally different way. So they're mixing up the different plants in a way that TCM just would never do because of the way TCM tries to guide things, and that's-

Ben Greenfield

Interesting. Okay. So are there RCTs or human clinical trials using traditional Chinese medicine in the way that a Chinese-trained doctor or biologist might use it that have proven to be less effective than what AI suggests?

Ben Goertzel

We didn't try that on the fly, to be honest. No, we just tried what the AI suggested.

Ben Greenfield

Okay.

Ben Goertzel

We did, however, I mean, we tried a few of these substances on humans with positive impact.

Ben Greenfield

Oh,

Ben Goertzel

really? We showed that you can, in many cases, roll back moderate Alzheimer's to mild Alzheimer's in people using these things.

Ben Greenfield

Really? Now, I'm assuming, by the way, are we talking like case studies, anecdotes, or

Ben Goertzel

- It was a small clinical trial with dozens of people.

Ben Greenfield

Okay. Interesting. Wow.

Ben Goertzel

With nutraceuticals, you can do an informal trial yourself because there's not a lot of regulations on herbs. I mean, with another concoction of Chinese herbs, which has anti-inflammatory properties, I mean, we tried in it, and you can show it greatly decreases C-reactive protein levels in human bloodstream. But again, these are small trials on dozens of people done at low cost. And these trials were done by Genescient, not by my company. They're the company that bought the flies out of UC Irvine. Okay. Since we've been working with the flies, we haven't done our own trials yet, but we've done a lot more AI work. So, by looking at DNA and then the RNA in the flies and how they differ between the long-lived flies and normal flies, I mean, we've found sort of four transcription factors in the genomics that seem highly causal for the flies' longevity. So, I feel like we've found... I mean, there's like... So, a fruit fly has around 14,000 genes, and about 2,000 of them are quite different between the long-lived flies and the normal flies. So, it's a quite broad-based difference. It's not like there's one gene that you chrome. On the other hand, there's a few dozen of those genes that seem to have more causal impact, and many of those have to do with four different transcription factors. So, we sort of seem to have used AI to pinpoint some of the real biological network differences between the long-lived flies and the short-lived flies. Then there's two questions, right? Like, how do you target those either with drugs, herbs, or gene therapy? And then what happens when you map from fly into human? And it seems that most of this stuff maps very nicely from fly into human, because it's very deep at the level of genomic regulation rather than being about very specific fly processes. It's not like you get a stronger thorax and you live a long time or something, right? It's more about better meta-regulation of the genomic network, better regulation of DNA repair, and better function of energetics, like in mitochondria and so forth.

Ben Greenfield

Yeah. The combined application of molecular chemistry and nutraceuticals using an AI or AGI model seems relatively straightforward. But then, do you see any potential future in expanding beyond nutraceuticals? Yeah. What I'm saying when I say expanding on nutraceuticals, I mean, if you look at hyperbaric oxygen and red light and electrical fields and even simple modalities like exercise and sleep, is there a way to model those in as well?

Ben Goertzel

Yeah. I mean, in the opposite direction of that, first of all, before I come back to that, the feeling you get when looking at these genomic differences is like, fuck, if we could just do gene therapy on 20 genes at once without causing bad side effects, then maybe you're done, right? Because you can see which genes you just need to adjust their expression in different tissues, and then you can see the impact that it has on the biological networks. And then trying to adjust the expression of these genes using herbs or even small molecules is such a blunt instrument. So you sort of feel like molecular nanotech is going to be the real solution. I mean, it doesn't have to be metal nanomachines. It could be engineered enzymes or biomolecules, right? But it seems like engineering molecules, so you can insert them into the tissues in the body, and then they just like mechanically alter it.

Ben Greenfield

Like what you're describing would be something like, I don't know, let's use something simple, like an amino acid sequence peptide, like MOTS-c as an exercise mimetic, right? People might inject it, and it would cause an increase in AMPK, similar to what you might get from exercise or fasting. And it sounds to me like what you're saying is like, repeat that molecular mimicry for heat stress, for oxygen carry capacity, for mitochondrial membrane, whatever. And you could just go on down the line and replicate as many lifestyle factors as possible with molecules.

Ben Goertzel

There's also engineered enzymes, right? So like if you see the long-lived flies have several SNPs in a certain gene that the short-lived flies don't have, and you can map that into humans, then you can engineer an enzyme that just goes in and actually edits the DNA to put those variations in there, right? And I mean, you could do that in the test tube now, and there's no fundamental reason you can't do that with engineered enzymes. It's just the delivery mechanism, right? Like we don't have a good way to not have that enzyme dissipate on the way into the part of the body where it's supposed to do its editing, right? So in a very targeted way for this sort of a nano-engineered biomolecules like that, that feels like the holy grail solution. And if an AGI can design that like five years from now or something like that, then that's that. I mean, because that would let you just re-sculpt the genome of every tissue in your body, however the AI has designed.

Ben Greenfield

Yeah, it is interesting. I mean, I've undergone follistatin gene therapy and without changing protein intake or weight training protocol, I gained 11 pounds of lean mass in three months, which is arguably very, very difficult without steroids and massive increases in calorie intake from that simple intervention. I'm not saying that's 100% bulletproof and safe for everyone, but I mean, I imagine if you could rinse, wash and repeat that safely across a wide variety of organ pathways, it'd be interesting.

Ben Goertzel

So my friend, Liz Parrish, who lives near me in Seattle, she has a company called BioViva. And I mean, she developed therapy for telomere lengthening. So she did this therapy with like injections into all different tissues around her body and she re-grew her telomeres, right? But then as part of that same protocol, she did a bunch of follistatin injections also. And now, I mean, Liz is super ripped after that, right?

Ben Greenfield

Yeah, it is interesting. I think she's also, from what I understand, I believe done some klotho treatments as well.

Ben Goertzel

Yeah, yeah, that's right. And I mean, again, it's an end of one experiment. Just knowing her as a friend, it's interesting to see the impact. It's hard to, I don't know how the difference, since she tried a number of things, I don't know how to tell which of the effects on her is due to one thing versus the other. She's looking great. She's looking young and strong and healthy, right?

Ben Greenfield

Right. Yeah, yeah. Knock on wood. When it comes to the actual measurement of how well these things are working, particularly in humans, we went from like telomere measurements, which I think were flawed because they were just a small sampling of telomere length on white blood cells to biological aging clocks based on pace of aging, now to singular aging clocks for each organ. But how accurate do you think the current biological testing landscape is?

Ben Goertzel

Not very. I mean, I'm not a huge expert on that, but we sort of looked at that a bit in building an app. We were doing a Rejuve.ai, which the app sort of tried to take a bunch of medical information from the user and give them guidance on what to do for health and life extension. And we found users of that app, they wanted a biological age estimate. They wanted to know, well, since last time I used the app, I've gone from 42 down to 38 or something. But I mean, when I really looked at the data underlying any of these biological age estimates, I mean, it's a little dodgy. And I mean, we've all met people in the longevity field who are like, well, I'm 75, but I have a biological age of 43. But you're like, well, why do you look so fucking old then? And then they get a heart attack the next year, right?

Ben Greenfield

Yeah, yeah. And then, of course, there's also the, well, what happens if I drop you in the middle of the wilderness without food?

Ben Goertzel

Yeah, I mean, I think tissue-specific and like organ-specific is certainly much better. And I think you can get an accurate read, a more accurate read of the age of the effective biological age of your heart or lungs or muscle or specific body system. On the other hand, we don't really have a handle on the holistic systemic function of the body either, right? Like there's a whole problem that happens as you age with the sort of message passing through the extracellular matrix, right? And it's a combination of chemical, mechanical, and even optical with ultra-weak photon emissions going through the body between the cells. And that messes up when you get old also for a variety of complex reasons. So you could have a bunch of different tissues being very healthy. But if the extracellular matrix signaling fabric is fucked, then your whole body still isn't necessarily doing well.

Ben Greenfield

I'm sure a lot of people just raised an eyebrow. Were you referring to the signaling mechanism between mitochondria using photonic light patterns?

Ben Goertzel

That's part of it. But I mean, there's ultra-weak photon emissions in the brain also. And I mean, in the whole body, I mean, there's just a bunch of sort of molecular goop in the body, like in between the different organs and the cells. And there's something like a 0.6 or 0.65 correlation between the sort of creases in this extracellular matrix and the acupuncture meridians, actually. So, I mean, there's some like energetic flow happening in the extracellular matrix through the body. And this energy flow works worse as you get old for reasons that are not fully understood, right? And I mean, that's just as an indication that there's a holistic aspect to body function.

Ben Greenfield

Oh, yeah, absolutely. I mean, it's interesting to talk to someone, you, who might be described as a sort of technologist in the longevity realm, referring to things like meridian pathways, Chinese traditional medicine, and things that people might consider to be ancestral wisdom, you know?

Ben Goertzel

Yeah. I mean, I lived in China for 10 years, and my wife is mainland Chinese, so I had some contact with that world. I mean, I did a deep dive into TCM theory at one point. And I mean, it's just really weird and crazy, right? I mean, the whole vocabulary they use, the way they're thinking is often...

Ben Greenfield

I have a poster tucked away somewhere in my office, a TCM traditional Chinese medicine poster, and it's fascinating to look at all the different meridian energy flow pathways and how they refer to specific organs.

Ben Goertzel

It's also sort of closed in the way that modern science isn't. Like there's an ancient book, which is like the Bible of TCM. And now every new discovery has to be met back to what's in that book. So it's sort of like how the Supreme Court has to map everything back to the Constitution, which is so hard to change or something, right? For religious, people have to trace everything back to the Bible or the Koran. And I think that's a constraint, because even as TCM practitioners discover new stuff, they can't revise the foundations just for cultural and historical reasons.

Ben Greenfield

That's a real nuanced discussion, because it's highly highly subject to interpretation based on faith beliefs, right? So if an ancient text was believed, and perhaps even accurately is, influenced by some form of divination from the spirit world, or by God, and set in stone as that, then modifying or altering that outside of the scope of divinity might actually be incorrect. So that obviously depends on whether you're a faith-based person or not.

Ben Goertzel

The Chinese are always flexible. Everything is subject to reinterpretation. But I do think it's a very tradition-bound field, right? Which is different than... I mean, science can be that way also. But I think if we're going to cure aging, it's not going to be by being tradition-bound in any sense. But still, there's a lot of knowledge in TCM beyond the specific herbs, right? And I think they did, while their theories are weird and not always correct, they're trying to grapple with the holistic function of the body in a way that Western medicine rarely is. And that comes back to a point you raised at the very beginning of this conversation, which I think is key. For AGI to really solve aging beyond merely coming up with some therapies that will double our lifespan, which is already very useful, right? But for AGI to really comprehensively solve aging, I mean, I think we are going to need an accurate systems biology simulation of the whole organism, right? And then once you have that, the holistic nature of the body will just become very, very clear. Because you see that with weather systems, right? We now have better and better simulation models of weather systems. And you can see it's chaotic. It's not localized. A wind over here can cause a hurricane over there. And once you have a computer simulation model, then you can understand all that. We have the same with climate modeling or ecosystem modeling. Now, we just don't have effective whole biological system models yet. Now, ironically, I think we might be able to pull together decent ones right now, just using the totality of biological data sets gathered on the planet, plus LLMs, plus other AI tools, like the reasoning systems my team has developed. I think if you took every biological model and data set known to men and tried to have an AI system put together the best comprehensive body simulation you could based on this, I mean, that wouldn't be perfect, but it would be quite interesting. It's very complex.

Ben Greenfield

Do you envision that biological simulation being digital or robotic?

Ben Goertzel

I mean, I think you want it to be digital initially. Making it robotic is many orders of magnitude harder because robots are not.

Ben Greenfield

Yeah, obviously. Because then you're literally generating artificial tissue.

Ben Goertzel

I mean, robots are made of the wrong kind of stuff, right? I mean, robots don't even typically have skin that can feel anything.

Ben Greenfield

That's what I'm saying, is you would literally have to create a robot with simulated tissue.

Ben Goertzel

Yeah. I mean, I think coming up with a reasonably accurate digital simulation.

Ben Greenfield

Yeah. I guess the only advantage of the robot would be what I raised earlier, like the ability to test out things like hyperbaric or certain weight training protocols, etc.

Ben Goertzel

Building a robot out of engineered biological cells is probably harder than curing aging.

Ben Greenfield

Yeah, you could. But you did. Weren't you involved in like a humanoid robot, like an eldercare robot?

Ben Goertzel

Yes. I'm still working with humanoid robots now. I mean, we have a new robotics company called Mind Children. We have a little child robot named Codey. And then I actually have a band where I play the keyboards with a lead singer is a humanoid robot named Desdemona.

Ben Greenfield

Wait, what's the name of your band?

Ben Goertzel

Desdemona's Dream.

Ben Greenfield

Okay, cool. I'm going to put that in the show for those people listening.

Ben Goertzel

Cool. Yeah, yeah, yeah. But I mean, the thing is, I mean, the robot has a humanoid face that can smile and so forth. And I mean, she can move her hands around. But I mean, there's motors and metal inside her arms. I mean, her face has 35 motors pulling the strings to different expressions. So I mean, it's not intended as a biological body simulation, right? And I think doing that, that probably needs a super intelligence, not a mere human level, human level AGI. And then, of course, you have some weird ethical issues. Because if you've made this simulated human, then it probably can experience things like a human and like, randomly fucking with it in the lab is kind of a problem.

Ben Greenfield

Yeah, I think you'd have to truly traverse the impossible in order to pull that off, because you'd have to introduce a soul slash consciousness.

Ben Goertzel

Well, that's a different topic. I mean, yeah, yeah, that's a whole different podcast. Yeah, my own feeling on that is soul slash consciousness is imminent in the universe, and it will pop up in whatever system you build, right? And I mean, this more the Asian perspective, I think, like, yeah, the soul, whatever it is, is not in the meat of your brain, it's in the universe, and your brain kind of amplifies that universal consciousness in a certain way. And I don't see why a digital computer that's highly intelligent, can't kind of amplify universal consciousness in its own similar, but perhaps different way. But that's another another frontier, as you say.

Ben Greenfield

Yeah, yeah, Ben, I want to because I know you're involved in a lot of different companies, I want to kind of subscribe to a feed or a newsletter to stay up to date about the developments in terms of what you're doing, because I'm pretty fascinated. I'm sure some of my audience might be as well. So where is the best place to follow your research? Do you have a blog or anything like that?

Ben Goertzel

So I do I do have a Substack. And if you just search my name, Ben Goertzel Substack, you'll find that if you go to singularitynet.io, that that's my AI blockchain project. But that's best organized in terms of having email mailing lists and sort of everywhere to go. I mean, Matt Morgan, who you mentioned, we're working together on a new AI company called BGI Labs. And there's a there's a website bgilabs.ai as well.

Ben Greenfield

But OK. By the way, do you say BGI or BGI?

Ben Goertzel

BGI beneficial, beneficial general intelligence.

Ben Greenfield

Got it. OK, cool. Cool.

Ben Goertzel

Building general intelligence for everyone's benefit, including extension of everyone's life or so we hope.

Ben Greenfield

Yeah. And great initials, too, with the BG. I'm going to put all this at bengreenfieldlife.com/AGIpodcast. I'll link to the show notes, other podcasts that are similar to this one. Ben's Substack BGI website and plenty more. Ben, thanks so much for doing this, man. You're you're a fascinating guy up to some very fascinating work.

Ben Goertzel

Yeah. Yeah. Thanks a lot. It's been a fun conversation.

Ben Greenfield

All right, folks, thanks for tuning in. Have an incredible week. To discover even more tips, tricks, hacks and content to become the most complete boundless version of you, visit bengreenfieldlife.com. In compliance with the FTC guidelines, please assume the following about links and posts on this site. Most of the links going to products are often affiliate links, of which I receive a small commission from sales of certain items. But the price is the same for you. And sometimes I even get to share a unique and somewhat significant discount with you. In some cases, I might also be an investor in a company I mentioned. I'm the founder, for example, of Kion LLC, the makers of Kion branded supplements and products, which I talk about quite a bit. Regardless of the relationship, if I post or talk about an affiliate link to a product, it is indeed something I personally use, support and with full authenticity and transparency recommend in good conscience. I personally vet each and every product that I talk about. My first priority is providing valuable information and resources to you that help you positively optimize your mind, body and spirit. And I'll only ever link to products or resources, affiliate or otherwise, that fit within this purpose. So there's your fancy legal disclaimer.