Future of Science and Technology Q&A (August 2, 2024)

Hello, everyone. Welcome to another episode of Q and A about future of science and technology. I see we have a whole bunch of questions here. Saved up. Let's see. Oh, this one looks interesting from Teo. How do you envision the future of physics? Informed neuroscience in particular? Do you believe that despite the brain being a warm environment, quantum effects such as entanglement and superposition play a role? Do you think the concept of quantum cognition will remain more philosophical than scientific? Okay, that's kind of interesting. So several different directions to go. One is something quite fundamental to quantum mechanics, which is the following thing. In a sense, the core idea of quantum mechanics is that there are many possible histories that all in some sense, happen. In classical physics, you kind of throw a ball, goes indefinite trajectory. In quantum physics, the idea is there are many possible trajectories that the ball could follow, and you somehow only get to know the probabilities for different trajectories. At least the theory only tells you those probabilities. Although we tend to think that definite things happen. So there's sort of a core phenomenon there, the interplay between sort of the formalism of quantum mechanics and our perception of what happens in the world. The formalism of quantum mechanics is all about many paths of history being followed and all about the weighting of those paths and quantum amplitudes and all these kinds of things. And the idea of superposition is that there can be states of the world where several of those paths are simultaneously being followed. There's a superposition of paths. Famously, the cat can be both alive and dead on different paths. Both are kind of the state of the world is the superposition of those possibilities. But the thing that has been a mystery for 100 years is how does it come to be that people think definite things happen? Because the formalism, the mathematical formalism of quantum mechanics says there are these many paths of history. They have certain weightings and so on. It doesn't say how that resolves into. And then we think definite things happen. And in the formalism of quantum mechanics, it's the measurement process, which is mathematically perfectly well defined, but the actual mechanism by which that works. And things like how long does it take? Is. Is not really addressed by the formalism of quantum mechanics. It's something that our physics project has ways to start addressing, ways to start understanding. How do these different threads of history get knitted together for purposes of our perception of what's happening? And let me probably not go into too much of the physics of this, but kind of the idea is we have this notion of branchial space, the space of quantum branches, so to speak, these different branches of history. And kind of the idea is that our minds, our brains, are extended objects in branchial space, just as our brains are extended objects in physical space. And our brains, our eyes and so on, the details of every single atom, every single photon and so on, we don't know. We only see with our eyes the aggregate of lots of photons arriving. We, we're sort of averaging over some region of space, for example, how much light there is there, et cetera. Well, it's the same thing, we think in branchial space that kind of our brains are averaging over several different paths of history. We're kind of spanning multiple paths of history, and our perception is somehow the aggregate of those paths of history. And that's kind of what's happening. This measurement process is about knitting together these parts of history so that we can conclude something definite happens. There are different kinds of measurement. You can imagine making one thing, there are certain kinds of measurements, like if you have a weighing balance, in the end, whichever side is heavier is going to go kathunk and go down. But it's a very yes or no type of thing. It either goes down or it doesn't, it doesn't sort of wobble in the middle. And that's a case where just having a little bit more left than right means it's going to the left, it's going to have it, so to speak. Whereas other cases where you say, is it gray, is it black? Well, it's a bit gray, it's a bit grayer, it's a bit blacker, and so on. It's a continuous kind of thing. But in any case, I think the key sort of phenomenon of quantum mechanics is that there's many parts of history. The key feature of our experience of quantum mechanics is somehow our brains knit together those paths of history so that in our perception something definite happens. So for example, inside a quantum computer, its big point is that it can kind of follow all those different paths of history. So, you know, you're trying to factor a number. You can try, does it divide by this 2, 3, 5, et cetera, et cetera, et cetera. Each path of history can try a different division computation. And so that means you can do all those division computations according to the mathematics of quantum mechanics. You can do them all in parallel. Well, you might say, okay, it's just a mathematical operation to decide what the answer was given all these different parallel results. Actually, I think that's a thing that has to be sort of physically happen, and it could take a long time. I've thought for 40 years or so that quantum computers are not going to live up to their kind of quantum expectations, because this measurement process really is a process that takes effort and time. And the advantages you get from this parallelism will be more than compensated for by the time it takes to kind of knit all those threads of history together. Okay, so the point there is one of the surprising things about brains is that they believe that definite things happen. And it is surprising because we got 100 billion neurons. And presumably as I'm sitting here yakking with you, some of the neurons in my brain are doing one set of things, another are doing another set of things. But somehow in the end, I'm going to say a definite word. Even though one population of neurons might be kind of up in arms talking about something different and so on, we believe we experience a single thread of history. And that's a very non trivial fact about brains. When people lose that ability because of damage to their cranial nerves and so on, then they are. Most people would say that they're no longer conscious or capable of being conscious. If they don't have the ability to kind of integrate together these different sort of sensory inputs and so on, and to get a single thread of history out of it, then they seem to have lost kind of the essence of what one might think of as consciousness and so on. It's this ability to pull out a single thread of history that seems to be pretty important to the way brains work. It's worth realizing that just as our 100 billion neurons have to kind of decide what to do, that it happens at a larger scale. If you're an ant colony or something, there are definite sort of collective things that happen in the ant colony. If you're dealing with human society, there are at least some level of definite things that society goes in this direction or that direction. Even though there are individual people who kind of are doing what they feel like doing, so to speak, but all sort of connected into the society. So this process by which one sort of reaches consensus in some sense and a definite thing happens. It may be different in different kinds of systems, but that's a critical thing that happens in brains. So now the one question would be, will people say, well, a quantum computer inside the quantum computer, all these parallel threads of history are being followed, and that would be just fine. You would translate that directly to brains. If inside brains things were happening quantum mechanically, if it was the case that we could have those different threads of history actually experienced within our brains and we never pull them together into this single thread of experience. So in some sense, sort of the quantum brain is a brain where those multiple threads of history are all being followed. But it isn't a brain like the way we experience our brains, which is that we experience definite things as happening. And exactly how that sort of, that kind of collapse from many threads of history to the one thing we experience, how that happens, it seems to be important, sort of important story in the dynamics of how brains work. And a certain amount is known about how that happens at the level of nerves that connect in with this and that aggregate, this and so on. So there's then the question, well, okay, how important are quantum effects in the brain? The question of what's important in the brain has been somewhat clarified in very recent times by things like LLMs, because it was before that. One could say, well, there's a lot of stuff that's going on in brains that we've never been able to reproduce with computers. There must be something going on in brains that is new physics, that is something far beyond anything we know. That's been the story of how far will AI get? How far will computers be able to get in doing brain like things? And the sort of advance of ChatGPT and so on said, well, actually this thing language that we have thought was a sort of vaunted characteristic of brains not replicable by anything else, could actually in a first approximation be replicated just with these simple neural nets implemented in a digital computer. So sort of the evidence that there's more going on in the brain than what we kind of can replicate in our digital electronics with our simple models of neural function and so on. The evidence that there's more going on really isn't very convincing at this point. And even a question like, okay, with 100 billion neurons, how much memory can you store? Is it plausible that you could store your lifetime of memory in that, just with memory being encoded as it is in the standard artificial neural nets, by things about the synapses, the connections between nerve cells and so on? I think the answer is yes, it's perfectly plausible. I mean, I think that the amounts of storage, which are much larger still than the amounts, I mean, the number of weights in a sort of high end LLM is approaching a trillion 10 to the 12 in a human brain. The, let's see, the number of neurons is around 100 billion. And the number of connections between neurons is surely at least 1,000 to 10,000 times that number. So it's at least hundreds of times larger than the number of weights in these neural nets, even though the neural nets have sort of broader knowledge, in some sense, of different things in the world than people typically have. So I think it is perfectly convincing that there's nothing sort of mysterious and more molecular about the ways that things get stored in the brain than that there is a certain amount of information stored at each synapse, at each connection between nerve cells in the brain. But there are consistent theories about other things that might be going on. I mean, the ones to mention are microtubules, dendritic spines, RNA in synapses and in nerve cells and so on. So, I mean, one of the possibilities is that the microtubules, so all cells have. Well, they have a lot of. The surface of a cell is not a barren place, the surface of a cell. Cells have these lipid. Have these membranes. Lipid membranes around them, but sticking through the cell are all these pores and all these different kinds of objects for different types of cells and so on. Although there's some commonality, a pretty common thing are microtubules, which are little things, things that stick out of the cell and that can move. And you can imagine that there are ways in which sort of whole fields of microtubules can interact with each other. It's an old theory from, I think from the very first conference about cellular automata that I organized back in 1983. The main person who's been interested in this came there and was talking about cellular automata as being something implemented by microtubules on the surface of cells. And that's one of the places where sort of more detailed stuff could be going on than we currently know. And some of the things that happen in microtubules could be importantly affected by quantum mechanics. So similarly, one can imagine that individual molecules are used as information storage. We certainly know that when memory is consolidated, synthesis proteins are produced. Exactly how that works. And whether the actual sequence of, let's say, base pairs on a piece of RNA or residues on amino acids on a protein or something, whether these are actually used, whether we're basically recording a tape of our memory on the base pair sequence of a piece of RNA or DNA, more likely, rna. Is that a possibility? And could it be the case that with sort of reading the RNA by maybe pulling it through some membrane, just like modern sequencing machines often do, that kind of nanopore type approach of pulling a piece of, I guess, RNA through something and seeing the effects of that sort of. As the different base Pairs come through depending on which base pair got pulled through next. Sort of a different, I guess it's voltage, but certainly atoms are moving around as a result of those different base pairs going through. And is that relevant? Could we store our memories that way? There doesn't seem to be much evidence that we are, but certainly not impossible. And one of the things one learns from looking at biology a bit is that whereas in physics the simplest model is often right, in biology things are always more complicated than you can possibly imagine. If you say memory is stored in synapses, there'll always be a footnote that says, except for this and that and the other thing, when it really matters what the cell body is doing, or it really matters what these extra glial cells are doing, or whatever else, it'll never be a clear cut kind of thing. It doesn't tend to work that way in biology. Biology, like a lot of engineering, is kind of hacks upon hacks making things work and sort of progressive polishing of that. So there are other effects. People talk quite a bit these days about quantum biology and about the possibility of, of quintessentially quantum mechanical effects doing things in biology. So a very common case people discuss is the spins of nuclei and of electrons actually where that's a quantum mechanical phenomenon. And can that have an effect on sort of the chemical processes that are going on in the brain? And for example, people say pigeons are sensitive to the Earth's magnetic field. That is one of the sort of use cases of kind of quantum devices is quantum sensors, where there's much, much higher sensitivity to magnetic fields than has ever been possible before. By looking kind of at the effects on quantum systems of those magnetic fields. And maybe pigeons directly. Well, pigeons do seem to directly sense the Earth's magnetic field. Maybe us humans directly sense the Earth's magnetic field through some kind of quantum process that, that is linked to things that we can actually sense in our brains. So I mean, there are those kinds of things. I don't think that the temperature as such of our brains is so much the issue because I think there are plenty of settings in which you can have an atom or an ion trapped in a protein cage where really not a lot affects it from the outside. I mean, the cage might be wobbling around as a result of it being hit by different things from the outside at the temperature that the brain operates at. But I don't think the stuff, the thing inside the cage is not as likely to be affected by that. So I think going back to my statement about biology that things are always more complicated than you imagine. I'm sure there are things where you could say yeah, that's a quantum mechanical effect. I mean, for example, people talk about the operation of chlorophyll, the energy producing protein in plants, that people say that the actual operation of chlorophyll depends deeply on various quantum mechanical effects about electrons and the production of photons and the production of electron cascade from them. And, and the way that particular. What is it, magnesium, Is that right? Is that the ion that's critical? It's either that or manganese, I can't remember which. That's critical to chlorophyll. There's a bunch of quantum mechanics associated with that. So there will be. It's not the case that you can throw out the quantum mechanics textbook when you think about biology. But if you say is the core operation of the brain dependent on it? The evidence these days is no, not really. We can see sort of brain like things going on perfectly well without having to invoke sort of quantum mechanical physics. And I mean, so this question about quantum cognition, there is this point that there is this phenomenon that causes the different threads of history to be kind of knitted together by things happening in our brains. And could one imagine being able to sort of operate in some kind of quantum trance where you are separately sensing those different threads of history? I don't know. I don't think it's totally impossible. I think that it's one of these things where whether one can sort of build up enough, whether one can have enough conclusion about what's going on to rather than just saying, I've got an individual neuron that's affected this way and I've got another individual neuron that's got affected that way. We can't sense, I don't think, the kind of, the states of our individual neurons. We only sense these very collective states. So exactly how you would tease that apart is not really clear, at least to me. Let's see, there's a question here from Veruliad, that's a fun handle. Are microtubules like electrochemical transistors? Not a bad suggestion? I mean, I think that the key thing about a transistor is that it operates as a switch. In the kind of transistors that exist in microelectronics, field effect transistors. The basic setup is you've got a source, a gate and a drain, and electric current is flowing from the source to the drain. But if there is a voltage at the gate, it can prevent or allow the current to flow. So it operates as a switch. By just changing that voltage on the gate, you determine whether the current can flow or not. And you can use that to amplify things, because if you have a little effect on the gate, it's having this big effect on whether current flows. So you can kind of feed that back to make an amplifier and so on. Now the question is, can an arrangement of microtubules act as a switch? And I don't specifically know. My guess would be that the answer is yes. One of the general principles Of a lot of science that I've done is it's easier to make computational devices than you might think. You don't need very sophisticated raw material to be able to assemble something that's capable of doing computational kinds of things. I mean, another area of great interest is with light, can one make a switch, an optical switch, where you purely. Where you have different A stream of light, and you switch that stream of light with another stream of light? We know that with very intense streams of light, that's possible using nonlinear effects and absorbers. But can that be done with kind of little amounts of light without expending lots of energy there? If that can one day be done when we'll have optical transistors and probably electronics will switch to photonics rather quickly in terms of this sort of electrochemical version of that, I don't specifically know. It would take a little bit of thought to think through whether the kind of the field of microtubules interacting with each other, how that will be used as a switch. But my guess is it can be. And that's an interesting suggestion. Let's see. RBS is asking, could the Boolean arithmetic functional devices in our brains Be affected by temperature, or is temperature one layer above that? Well, I mean, most likely temperature. It's like, is a microprocessor affected by temperature? A microprocessor has electrons that represent data and represent operations on data. Typically in a modern microprocessor, it's about 100,000 electrons per bit. Those 100,000 electrons, they're operating at typically room temperature for most electronics. And so all those electrons are running around in this way that is affected by temperature. They have the kinetic energy associated with being at room temperature. But nevertheless, even though these electrons are running around in the sort of random way, it's still the case that we manage to treat that aggregate of 100,000 electrons as being one definitive digital bit. And we don't really care about the fact that those electrons are running around in their own separate ways. So this was a thing people in the very early history of computers were very confused about. They thought that there would be enough sort of imperfections in the operation of computers that you really wouldn't be able to run any long program because it would always sort of topple over as a result of just a series of small changes, effectively making the whole thing collapse. But that didn't happen because digital electronics, it's possible to make it incredibly reliable. Like, you can go trillion trillion steps and operations, and not a single one will be wrong. It will be even though the electrons that make up each bit are running around in random ways. And so I think it is with the brain that there can be all sorts of chemical things that are very sort of randomly running around, but nevertheless, the conclusion can be something definite. Now, the brain seems to make use of a certain amount of randomness in its very operation. That's certainly what we've seen in neural networks and so on. And it may not be the case. It may be the case that its way of information processing actually likes randomness. It uses randomness to prevent getting stuck. And so it's a little different from digital electronics that way. But one shouldn't conclude that the fact that there is sort of kinetic energy associated with temperature and the individual chemical components of brains, that doesn't have an implication really, for whether the brains can do definite things or whether they'll be sort of randomized, because digital electronics doesn't get randomized even by that kind of effective temperature. Now, there's another question, which is when randomness is produced in the brain, how is it produced? I mean, when you see a little fly buzzing around and it's flying in different random directions, how did its little brain decide that, oh, now I should change direction and pick a random direction? It isn't really known. I mean, it's known things like the sort of central pattern generator that makes critters walk and periodically put one foot in front of the other. There are actual brain circuits which one can identify that are associated with that kind of periodic driving of legs or whatever. But a similar sort of circuit, that is the thing that makes randomness. We don't know what that is, whether that is excavating temperature or whether that is more a pseudo random generator that's just essentially an algorithm that produces something that appears random. We don't know. My guess is more likely to be the second one than the first one, but we don't really know. And I think there's a different question, which is when things like LLMs, large language models operate, like ChatGPT and so on, they have another notion of what they call temperature that has to do with the following things. So in a large language model, you take the text you've written so far and you sort of take that, convert it into numbers, ripple that array of numbers through this whole neural net system, and at the end the neural net will tell you, here are the probabilities for each of the possible words you could write next, or possible tokens you could write next. And then the large language model has to say, well, okay, I know those probabilities. And there's a most likely word and there's a slightly less likely word and there's a really, really unlikely word. Which word should I actually write next? And so there's the notion of zero temperature. It's not a real physical temperature, it's a sort of algorithmic temperature. Zero temperature means always write the word that the LLM estimated had the highest probability. Always use that word next. And what tends to happen is the kind of text that LLMs produce is very flat. If you do that, always use the most popular word next. Sometimes it's good for writing code and things like this, but it tends to lead to very flat kind of English prose. And so what was found when ChatGPT was being developed at a temperature of 0.8 gives one good kind of sort of plausible, somewhat creative, but not off the wall kind of English text. What does that mean? Well, the way this is set up, a temperature of one means pick those words according to the probabilities at random, but according to the probabilities that the LLM said they would have. So instead of just saying every time, just pick the most probable one, say, do a random, roll a dice at every step, but weight it according to the probabilities the LLM said those words should have and generate words based on that. 0.8 means go a little bit closer to the kind of flatness of always pick the most popular word and use that as the mechanism. But so far as we know, that kind of temperature has, so far as we know, that has very little to do with any kind of physical temperature in the brain. Now I have to say, a few years ago I was wondering about this and I asked a friend of mine who's sort of leader in computational neuroscience about this and he immediately responded with a paper which I've now forgotten the details of, that claimed that there was some kind of temperature like phenomenon in the brain that doesn't have to do with physical Temperature, but has to do with kind of randomness happening in the brain. That was something very analogous to what the LLMs are using. And the sort of implication was that's how we manage to do kind of creative things. That's why we don't all just do the same standard thing we always did, because it was the thing that everybody would conclude was the best thing, the most likely thing to do. Let's see. Hybe asks, which do you think will happen first? Repairing brains naturally through natural science, medical science, and so on, or having the first computer brain transplant for those who suffer brain traumas? That's an interesting question. It's kind of a race right now. So, you know, stem cells, okay. So, you know, we all start from one cell. We start replicating, we replicate many times. That initial fertilized egg cell eventually differentiates into all the different types of cells that make up our bodies. But it's not that many kinds of cells. It's maybe 500 kinds of cells, give or take, that are skin cells and muscle cells and heart cells and brains and many different kinds of brain cells and immune cells and so on. These are what make us up. And when we go from that initial cell, that stem cell, it then differentiates into all these different kinds of cells. And what people for a long time had no idea how you would get a stem cell other than by starting with one the From a fetus or whatever else. What was discovered about. When was it, 15 years ago, give or take, was this series of things you can do to a cell that will make it revert. Even if it's a skin cell or something that's differentiated into. I'm specifically a skin cell that will make it revert back to being a pluripotent. Could do anything. Stem cell. It's a. This thing's called Yamanaka factors, which are really a very weird sequence of. I don't know what it is. 24, I think maybe different things that you do to the cell. And then somehow it gets back to being a stem cell that could differentiate into any other kind of cell. And then you have the forward problem of, given that you have the stem cell, how do you coax it into. Into differentiating into this kind of cell or that kind of cell. And again, it's a very weird. Kind of. Seems like alchemical magic to get it to turn into a heart cell or a brain cell. And I think, you know, there are processes for getting. Making heart cells that take only a few days from the original stem cell. There are processes for making brain cells that take, you know, three, two and a half months. There are processes for turning it into a pancreatic beta cell and important for curing type 1 diabetes that take, again, several months of tend your cells every day and feed them different kinds of things to coax them into turning into these different kinds of cells. We don't know for all the different cell types in the body, we certainly don't know procedures like that yet. The ones that have been concentrated on are kind of the ones that have the most medical impact. But among the ones that have been successfully done are many kinds of brain cells. You can make from that original stem cell. You can differentiate it into a whole collection of different kinds of nerve cells. You don't get the overall architecture of the brain. You get the jumble of nerve cells. Now, if you print a scaffold, like, I know people who are doing this for lungs, where you print kind of the scaffold using 3D printing, and then you kind of fill the scaffold with stem cells, and then they differentiate into actual lung cells, but with the actual structure of lungs. So the things will operate presumably just like actual lungs that were grown with cartilage and so on in them, all those kinds of things. But so for brains, that sort of overall structural thing I don't think has been done yet. I think it's a little bit. One's a bit squeamish about that, because when you still have a jumble of nerve cells, you can kind of tell yourself that's not really a brain there. You have something that has the kind of overall morphological structure of a brain, and you actually have actual brain cells filling in there. It really becomes a lot more brain like. And at this point, people have made organoids that are millions of nerve cells, you know, not up to 100 billion yet. But there doesn't seem to be any reason why you couldn't do that. Although right now, they're all sort of just jumbled up now. So then. So that's kind of path number one is you go from the. You make stem cells, you make them, let's say, for a particular person. So their immune system wouldn't reject anything that was made. The immune system would say, yep, part of me, I'm all good. And you wouldn't get rejection of any kind. And then you kind of go forwards and you make these different kinds of cells. Now, I mean, one of the things not very well understood yet is the phenomenon of genetic instability, that when you go back, when you use these Yamanaka factors. You reprogram the cells, people back to being stem cells. It's not obvious that the thing you get at the end has really lost all its memory. It's not obvious that you didn't need like another 20 steps or something to really erase this aspect of its memory as well. And there's a certain tendency for the things that grow from those stem cells to somehow retain a memory of the type of cell that they originally were before they were reverted back to stem cells. And one of the problems there is that it's tumor cells are something else you can get if you have a sort of genetically unstable cell. So that's kind of one of the risk factors if you're making things from stem cells and so on. But in any case, sort of one branch is you make nerve cells from stem cells. Maybe you find a way to make something that's more architecturally like a brain rather than like a jumble of nerve cells. People have certainly found electrical activity in organoids and brain organoids that where even though the nerve cells are jumbled, they can still be electrically active. Like in brains. One could sort of imagine, can you talk to the brain organoid? Well, it's hard to know how it will know what you're saying and how you would sort of train it to do anything. I sort of imagine you could feed it something like dopamine as a kind of a global signal of did you do the right thing? Brain organoid, it gets very ethically squiggly, I think, as you get into that kind of domain. But one of the things, so that's sort of branch number one, is you grow sort of replacement tissue and you can for brain tissue. And then the question of how that gets sort of populated and how like memories and things and processing start, how that gets knitted into the existing brain, not clear. But one of the things that seems to happen in biology, more than you would expect is, is if you put in the right kind of cell, it'll find out if it's a muscle cell for the heart or something like this. It's not so hard for it to go and migrate to the right place and kind of knit itself into what's already there and just start working. It seems to be the case that biology, in a lot of cases, it's just like throwing the right kind of cells that organize themselves in the right way. So it seems not out of the question that you could have kind of stem cell derived nerve cells that would just sort of knit themselves into the brain in a useful way. And I think, you know, there certainly are, well, diseases like Parkinson's disease, where I guess one can imagine. I don't think this has been done yet. One can imagine some sort of growing cells that produce dopamine and things like this, which is what's lacking in Parkinson's disease. Yeah, that's right. And then that you could use that kind of method. So in other words, that's kind of replace it with neural tissue. And that neural tissue initially will be sort of untrained. I mean, when one does this in artificial neural nets, you initialize the network with randomness. And then as you train the network, you keep on trying to adjust the weights in these neural networks and these neurons to correspond to what you want the neural network to learn. And presumably that same kind of thing would happen in a brain. If you start with these kind of random, sort of randomized sort of freshly grown neurons, they would presumably knit themselves into the brain, as the brain already exists. Quite what it would feel like if you replaced a piece of some number of cubic centimeters of brain tissue with this kind of new and sort of randomized, as if you were a newborn type thing, a piece of brain tissue. I have no idea what that would feel like and how bizarre it would get. And whether if you have a brain that is partly X number of decades old and partly like a newborn brain, how does that connect? How does it work? So that's branch number one. Branch number two, by the way, in that branch number one, I think the expectation would be that the newly added neurons would knit themselves in with existing neurons pretty easily, that they would start having electrical activity from the existing neurons. You wouldn't have to go to a lot of trouble to guide the neurons to exactly the right place. Okay, branch number two is to use electronics and to have something where you have an interface, a brain computer interface, where you are having a little electrode basically that can excite a nerve cell and that can read the output from a nerve cell. And where the nerve cell, which usually would be electrochemically sort of interacting with another nerve cell, instead is electrochemically interacting with that piece of electronics. That's sort of a different plan. The feature of electronics is that it runs a million times faster than nerve cells. It takes a nanosecond where a nerve cell takes a millisecond to do things. And quite what, you know, whether one would be able to have a sort of broad interface between nerve cells, I mean, that's what people are trying to do now is to have lots of little tiny electrodes that are sort of touching lots of different parts of nerve cells. Another approach is optogenetics, where one puts in different genes into the nerve cells so that the nerve cells literally flash when they are active. And then you can imagine having some pure optical mechanism for reading out the brain. And you could even imagine the other way around. Maybe you use two different frequencies, one for the transmit frequency from a nerve cell and the other for the receive frequency where you have something that's flashing light at the receive frequency to send data into the nerve cells. And that that would be a sort of two way communication mechanism with nerve cells. And then the question of sort of building. I think I've talked about this before, kind of the neural user interface of that sort of electronics with things in the brain. That's its own separate challenge in terms of which will arrive first. I don't know. I think it's a very close race actually. I think for different purposes like these more collective things like Parkinson's disease, which has been sort of approached with deep brain stimulators, which are more on the kind of electronics meets the brain side. That's something where you can also imagine stem cells and actually adding tissue being relevant. And cases where you have for example, issues of motor control, where you have lost. Where something where the brain is communicating through the spinal cord, for example, with the rest of the body has been lost. Whether you can pick that up with electrodes in the brain, that's one way you need some way. You either need to repair the connections to the outside world or you need to intercept things in a more electronic way. My guess is that one will see approaches based on both those different directions coming in in the not too distant future with different particular issues being addressed by each one of those kinds of things. Let's see. Dougie says I've heard that AI should be able to develop treatments for cancer, but it will take decades of machine learning. What do I think could accelerate this learning process? You know, I'm a little unconvinced by the, you know, AI will cure cancer idea. I think one has to sort of take apart what is that idea. One idea is already in the literature of biomedicine is the cure for cancer. We just haven't noticed it. And if you put this and this and this together, you'll suddenly get by putting, by having this process where it's 10 different things put together, then you'll sort of have the cure for cancer. Now there have been proto cures for cancer for a really long time, like immunotherapy is one that's been sort of out there for, oh, nearly 100 years, I think. And it has all these different reasons why it was tried and this didn't work. And then people say, no, but it really is worth trying again. And there's a phase now where it's really worth trying again. It seems to make intuitive sense that it should work to get the immune system. Normally, tumors have a bad habit of switching off the operation of the immune system and convincing the body that. That the tumor is really not an alien thing and shouldn't be attacked by the immune system. If it was attacked by the immune system, it'd be gone really quickly. And so the challenge is, can you sort of reprogram the immune system to really pay attention to the tumor, or can you mark the tumor in a way so that the immune system doesn't ignore it anymore? But that's an example of something where it's not like people haven't known that was a possibility. But the devil tends to be in the details of can you actually implement that? What, what happens to the actual drug? Does the drug turn out to produce toxicity to the kidneys or something like this? That was something completely out of left field that you weren't thinking about when you were thinking about the drug as a way of marking tumor cells or something like this. So this question of can one, for example, read the literature? Can an AI read the literature and sort of connect A to B to C in a way that will lead to progress? I think the answer is possibly yes. I mean, lots of progress was introduced many, many years ago when people started to be able to search the literature of the scientific literature. I mean, that happened long before the web. There were online database services. I certainly used them in the 1970s, where you could search for all abstracts of all scientific papers quite straightforwardly. Those things had started to exist in printed form, I think, in the late 1950s. Certainly they were robustly there. By the 1960s, it was a little bit. I used the Science Citation Index, for example, where you would have this so called. They called it the permuterm index, where they basically took the terms in the abstract and they would sort them in all possible. They would permute them in all possible ways, and you could look up in an index, sort of every word that was kind of like a keyword search, but done with big books of thin paper and so on. So that stuff existed and that stuff surely advanced biomedicine, significantly, to be able to look things up and not just sort of repeat the same experiment, because you didn't know about the previous version of it that had been tried. And one can do more of that. I think with AI Knitting things together in terms of the idea of AI is going to have a brilliant idea that we haven't had. Just because it's AI it doesn't really ring true to me. I mean, the fact is it's very easy to just sort of say, let's invent a molecule that's never existed before. You too can type in a random, or you can generate with Wolfram language or something, a random sequence of amino acids. Even actually in version 14.1 that just came out, we have the capability, you generate a random sequence of amino acids and we'll tell you at least roughly what the protein that corresponds to that sequence of amino acids, roughly how it folds up and what shape it is. So you can easily generate a protein that nobody's ever talked about before. Now, is that protein useful or it actually fold in precisely the way you think, what will its interactions with other things be? That's a long, complicated story. And even knowing, well, what kind of protein shape is worth having, kind of what target should I attack with this particular drug or something, what do I want it to do? That's a challenge as well. And I don't think there isn't a magic bullet there in terms of, oh, we'll just teach a machine learning system do it and it will magically do it. This and the idea of will it be creative in a way that humans are not very easy to roll the dice and produce something that's never been produced before. The question is, how does that thing fit into the other things one knows and how does one make it useful and so on. RBS is commenting. I've heard 10 plus times that they found a way of curing cancer. The thing to understand though is, is a lot of progress has been made in cancer. I mean, there are, you know, cancer is a generic issue, so to speak, where sort of, you know, you have kind of replication without bound of tissues. You usually have genetic instabilities of various kinds where the genome is messed up in all sorts of ways. But you know, and the thing worth realizing about it, it's sort of a return to the most primitive form of life. I mean, we in our, you know, for critters like us, in our normal operation, we grow to a certain size and then we stop growing and our different tissues they, you know, our organs grow to a certain size and they stop. They're very well controlled in the early history of life on earth. It's almost certainly the case that, you know, that cells just replicated without bound. You would get these stromatolites and things like that, giant mats of material that could be a kilometer across or something that might have been effectively a single organism, maybe a colony of organisms, depending on how you look at it. But it's a thing that's just growing, growing, growing. The goal is to grow. And tumors are, in a sense, a reversion back to that very primitive form of life. And the question is, but that's a generic statement. Each individual kind of tumor has all sorts of different characteristics of how does it masquerade as a particular kind of tissue? Does it just hang out somewhere being very different from that kind of tissue, et cetera, et cetera, et cetera? Is it a solid tumor? Is it something that's circulating in blood, et cetera, et cetera? And there's been tremendous progress made in the last few decades on lots of kinds of tumors, which it used to be those kinds of tumors would be really bad news. But now they end up being treatable and have very high cure rates. So, I mean, there has been plenty of progress made. I would say the area where perhaps there's been even more progress made is early detection, because with mri, with now genetic tests and so on, it's increasingly possible to detect tumors when they're really small. And it's much easier to get rid of a tumor when it's really small, perhaps just by removing it, but otherwise by attacking it with drugs and so on, than when it's gotten really big. Because, among other things, by the time tumors have got really big, they built all kinds of fortifications around them to prevent themselves being attacked. And those are hard to break through. But yes, it would be unfair to say that there hasn't been progress made in curing cancer. But cancer is a very generic issue. And so it's curing every tentacle. It's like saying, can we cure every disease? Can we cure every bacterial infection? Well, antibiotics have gotten a long way in curing bacterial infections, but you've got to have different kinds of antibiotics for this and that. And then there are all kinds of different issues and different ways that you can deliver an antibiotic to different tissues, et cetera, et cetera, et cetera. So it's not, you know, the one could say antibiotics cured bacterial disease. That would not be an unreasonable thing to say. And we haven't had a similar kind of broad spectrum kind of cure for cancer. It's been More specific kinds and so on. But certainly progress has been being made whether there will be eventually a slam dunk. This is just a cure. This sort of goes to the heart of how it works. I mean, in a sense, bacterial infections, bacteria are prokaryotes. Our cells are eukaryotic cells. Prokaryotic cells have different things like cell membranes. They have different, a bunch of different apparatus for replicating the cell. So if you can kind of target those particular kinds of the thing that makes a prokaryote different from a eukaryote, then you can make a drug that will blow up a bacterium and will leave an ordinary cell of our bodies unaffected. So I think it's. And you know, is there a way of saying this cell is a cancer cell, I'm going to blow it up? That's been hard to do because the mechanisms of what's exposed on the cell surface are messed up in cancer. And so it's harder to tell that. And that's why the immune system can't tell, et cetera, et cetera, et cetera. So, yeah, IR says maybe not a cure, but a control. Micro monitoring and cancer killing nanobots. Yeah, I think that's a. I mean, look, that's happening right now. Our immune systems, probably people always say, and I think it's true, that the immune system constantly is surveilling for tumor cells. Even within every cell. There are. Oh my gosh, what are they called? There is a mechanism for finding misfolded proteins and destroying them. So that's sort of an early stage. I mean, you know, the way we sort of deal with sort of things going wrong, like with the immune system, you know, we have the innate immune system and the adaptive immune system. The innate immune system just uses certain patterns to decide. Yep, that's. That really doesn't look like a thing that should be getting inside. Let's zap it in some way. And that's, you know, that's the set of heuristic patterns that have evolved over the course of history, course of biological history, to be things that you could innately immediately say, reject that. Reject that. The adaptive immune system in each one of us is trained in our early years to notice what things are part of us and have the genetic sequence of us versus things that are alien and should be attacked. So there are those two different mechanisms. But the idea that the immune system in its different forms kind of monitors for what we're trying to do with drugs and other things is to find aspects of tumor cells that are different from other Cells. And that will allow us to attach the killer explosive or something to the tumor cell. That when the thing's attached to the tumor cell, it's going to blow it up in some chemical way, probably. But just like in a cell, when a cell has programmed cell death, when a cell decides that sort of it shouldn't be around anymore, it releases lysosomes, I guess, which release chemicals that destroy the cell. And so the question is, and you know, the question is, can you target things that way? The immune system is already doing that kind of targeting. In the case of the immune system, there are killer T cells that attack cells, and eventually, gosh, it's always so complicated. And I have to remember the whole tale of woe of how the immune system works. But they're marked by antibodies, and then there are things that come along and macrophages that come and attack them, and T cells that interact with cells and can destroy them and so on. But the basic bottom line is at a cellular level, there's continual surveillance. We could think of the things we have in our immune system as like little nanobots. They happen to be sort of biological nanobots. And will we have more technological nanobots? Well, probably those things will be built in a way that is very biocompatible. So they might very well be built with things like proteins, because those are things that sort of are compatible with other things in our bodies. But they might be designed in a way that could be really like a little machine, like a little, you know, we don't yet have a fully artificial cell. There's all sorts of efforts to find out what sort of the minimal set of apparatus that you need to make a viable biological cell and perhaps a viable organism that can replicate. We don't quite know that yet. There are claims about what that minimal cell might be like. I think they're a little bit funky because the basic statement is if we take anything out, the cell will die. But that doesn't tell you you got to the minimum. That just tells you you're at a local minimum. And that from the things you have, if you take anything away from the thing you have, you die. But that doesn't mean there isn't a much better minimum somewhere else that you wouldn't get to without sort of adding stuff in and then taking stuff away and so on and going down a different pathway. But I think that the. And the question of surveillance, increasingly there's a question of whether you take cells out of the bloodstream and you do you check for the things like Methylation of DNA, kind of modification of DNA after it's been, when it's in a cell. And the way that different parts of the DNA, not their sequences being changed, but the way the DNA molecule is sort of the external coating of the DNA molecule more or less that affects sort of how that DNA molecule can be transcribed. And those kinds of things have, it seems, a pattern that corresponds to the appearance of tumor cells. And one of the things that's also convenient about sort of the way critters like us are built is the blood kind of goes everywhere. Blood goes within 2 microns of basically everywhere in your body. Otherwise you wouldn't be able to supply oxygen to those tissues. So there's capillaries that have the blood go everywhere. So that means if there's something funky going on in the body, it's probably going to drain kind of, you know, whatever's there, those cells are going to drain into the blood and the blood is going to get all mixed up. And if you just take a blood sample, you're probably going to get a few of those kinds of cells just like, you know, you'll get, oh, I don't know. Any woman who's had a child will probably have some cells from their child circulating in their bloodstream that are, that is sort of a reverse transfer type thing. So it's kind of a, you know, the bloodstream kind of sees everything. So that means that if you're trying to kind of surveil your body just like the immune system does, you can sort of just check what's happening in the bloodstream. And so that's a way that you can expect to do that externally. You can have the little bots going around as the immune system does within, within the bloodstream itself. Or you can have kind of, or you can take the blood out and go put it in a machine, so to speak. There's a question here from kk. Will we ever perfect the human immune system? It's an interesting question. It's an interesting question what that would mean. I mean, we have, you know, how does the adaptive immune system work? Well, we've got these two different paths, the B cells and the T cells, but they're both, they're little pieces of DNA which produce these hypervariable regions that are sort of randomly chosen and which produce like 10 billion different kinds of antibodies, let's say that are all slightly different shape. And that different shape will, when some sort of alien substance comes in, it will be a certain shape. And the question is which antibodies are the complementary shape, and will therefore stick on to the thing that comes in. And the whole point is that in our early life, our antibody population is kind of controlled so that the antibodies that are kind of the complementary shapes to things that are in our own bodies, those antibodies are removed. So we're left with ones which are if something matches this antibody, it's an alien thing. And then when the alien thing comes in and when the antibodies start to bind, there's kind of a feedback loop that produces more and more of those antibodies, then recruits other kinds of cells and so on. T cells are very much the same way. They have T cell receptors, they have a few trillion, I think, different possible kinds of T cells. But the question is, okay, so we've got all these different shapes. First question is how come there is universality in the shapes of that you get in antibodies or T cell receptors. That isn't really known yet. In other words, you've got all these different possible sort of sequences of amino acids that make proteins. And the claim is that by changing that sequence of amino acids, you can make kind of close enough to any shape for any possible intruder. And you can kind of be far enough away from the shapes of things that are actually in your own body and molecules that are in your own body, proteins that are in your own body. And so how that universality maps out the space of shapes is not really well understood yet. And then when you have things like, when you do things like vaccines, what you're doing is you're saying, okay, we're going to put in this antigen this alien substance that is going to cause the body to enhance the production of, for example, antibodies that, and eventually T cells and so on that react to that particular antigen. But that's a tricky thing to design the right shaped molecule that masquerades as a piece of the thing you're trying to vaccinate against. Like, you know, you pick some random virus and the virus is usually folded up in a capsid or something like that, and the virus also has actual viral protein in it. And you say, well, I'm going to pick these pieces of the capsid protein, these pieces of the viral protein and so on. I'm going to put those together, going to say, this is what our vaccine is going to be. Now try and get everybody, when the vaccine is when they're exposed to the vaccine, get them to produce antibodies to those shapes. Now it's a tricky thing because the kind of, the difficult side of vaccines is just putting a little tiny drop of something in doesn't necessarily get your immune system terribly excited. So you have to put in these so called adjuvant molecules and so on, which do things like say, hey, immune system. It got injected into your arm and it's really stuck there. It's a muscle. But get the immune system to get really hyperactive about things and send a bunch of immune system cells over to that site, parade lots of immune system cells past that site so that it will have sort of an outsize effect and actually make the immune system pay attention and produce antibodies and so on. But how the precise design of vaccines is a difficult matter. Like in the COVID vaccines there were various efforts to pick different proteins and they didn't work. They had all kinds of problems. And it's sort of an art to make them work just this well and not too well and not cause other tissues to be affected and all sorts of other things. So knowing exactly how to do that is a challenging thing. And I think there's sort of fundamental biology that one should understand better in figuring out how to make that work. But if you say a perfect immune system, what does that quite mean? It means presumably an immune system which can tell self from non self and blow up anything that's non self. But that's a tricky business because, for example, let's say. Well, let's say that there's. Well, the microbiome is an interesting area. Actually there's another question there about that. You know, how do you not attack certain gut bacteria and so on. I think that is because the immune system is not exposed to those things. I think that's the reason that that doesn't happen. But you know, how can you. For example, one of the issues of the immune system is things like viruses mutate and if you've said okay, I've got this vaccine and it's really given people this intense blast of antibodies of these particular sequences and then the virus mutates a bit and then it's like, well actually we didn't get. It could be that in the natural operation of the immune system there are various theories about this which have been either proposed, credited, discredited and so on. A nice theory that I don't think is quite right, but maybe there's something to it. There's a network theory of the immune system that says you make all these antibodies and then to prevent there being too many antibodies, you make anti antibodies and you make anti anti antibodies and so on. And it's a big dynamic equilibrium of all these different kinds of anti anti antibodies. That you're maintaining. And one feature of that model is, is that it means that over the course of time, as you go through these cycles, you kind of broaden out the region of shape space that you're making these antibodies to. And if that is happening at the same time that viruses in the environment are mutating to change their proteins a bit, you might land up in the right place at the right time, so to speak, with that. But so making the perfect immune system has to deal with issues like mutation of the viruses in the outside world. Maybe you mean by a perfect immune system that it can say, just perfectly nail the question of self versus non self. And I don't know, you know, it's an interesting question whether, you know, you're down at the level of molecules. So you're trying to tell these proteins. Can you tell a self protein from a non self protein? Remembering that proteins are flopping around all over the place. So it's not like, despite what you might think from a protein folding computer program, even crystallography on a protein, the protein isn't just sort of all locked in one shape, it's flopping around all the time. And exactly what it means to say, well, it's self versus non self. And this piece of shape space is okay, and that piece of shape space isn't. These are tricky issues, but it's an interesting question. So Nikolai asks, do you think, let's see. Well, do I think that the relevance of the microbiome in medical science will increase in the future? My guess would be yes. I mean, it's always amusing when you talk to people who are experts on these things. They'll say, well, you have, whatever it is, 10 trillion ordinary cells in your body, but you have a lot more bacteria. And if you talk to people who are experts on viruses, you say, and you have even more viruses in your body than bacteria. But yeah, you have lots of bacteria and there's some kind of symbiotic relationship between you and the bacteria in you. I mean, for some organisms, they couldn't digest their food without microorganisms, without bacteria in them. I think we can, but I think so. I don't think we need bacteria in our gut to be able to digest food. But certainly there's a giant ecology of what's happening in our gut as the bacteria sort of digest things, redigest things, produce different substances, and there's sort of a complicated equilibrium, just like in the ecology of an island or something. There'll be creatures eating creatures, and one creature will produce this substance and secrete that substance and do this and that and the other that's happening in our gut, in the microbiome there, and microbiomes in our mouths and other places and so on. There's bacteria all over the place. And a typical person's microbiome, a given person, it seems, will produce a very definite microbiome. And even if you zap it with antibiotics and kill a bunch of bacteria by taking antibiotics, I've become very squeamish about that because it's like my microbiome is probably an important part of me, and I really don't want to, you know, kill it off with antibiotics if I can at all help it. But even if you do that, the microbiome will, I think, typically come back in the same form in a person as it had before you zapped it, so to speak. So we all tend to have some kind of microbiome that is affected by what we eat. If you change your diet a bunch, you'll change the concentration of different kinds of bacteria in your microbiome. And there are all kinds of details that are just emerging. Like the human appendix, for example, the might be kind of a place where you are sort of a farm for the microbiome that's accumulating different kinds of bacteria and such like. And that's important for kind of the ecology of the microbiome. But I think there's increasing evidence that sort of, you have an effect on your microbiome and your microbiome has an effect on you, just as, I mean, again, it's kind of biology is always more complicated than you can possibly imagine. There's nerves, which clearly have. From the gut to the brain, from the brain to the gut, from the gut to the brain. There's interaction between nerves and the immune system. There's interaction with the microbiome, et cetera, et cetera, et cetera. All these different systems seem to be connected in some way. Whether those connections are incredibly important or not is unclear. Whether they just happen to be there because of the way evolution worked is also unclear. But my guess is that microbiomes will be increasingly kind of paid attention to now. I, because I do every one of these kind of weird personal analytics types of things. I studied my microbiome years ago now, and you get this giant list of, here are all the bacteria that you have. And it's like, so what? I don't know what to do with this information. And still we're in a position where we really don't know what to do with that information. Perhaps in time we will know. And certainly it's interesting to see. If you change your diet, how does your microbiome change? Whether it matters, I don't know. I mean, I heard about a slightly funky scheme recently to change your mouth microbiome to prevent tooth decay, because apparently you can change the concentration of different kinds of microorganisms there to prevent the ones which are more acidic, I guess, that lead to tooth decay. I have to believe. My immediate question is, but does it leave a nasty taste in your mouth type thing? Because I don't know to what extent the microbiome in one's mouth affects the background taste sensations that one has in taste buds on the tongue. I have to believe that that's a cellular level phenomenon because the taste buds are sensing, well, perhaps even individual molecules, but, you know, aspects of cells and so on. So I don't really know how that will play out. Ivy is asking, maybe not an artificial brain. What about artificial hearts? Would those be easier to have a technological implant or a natural one? Livers and kidneys, you know, people are working on a bunch of these things. My guess is that kidneys will be comparatively, you know, okay, so there's a big effort. People I know involved in the effort to make artificial lungs seem to be going quite well. And the claim is that similar technology can be used for kidneys. I think the liver is a little bit more complicated, but not fundamentally. They're fairly uniform tissues that if you can print sufficiently high resolution, kind of scaffold, and you can sort of coax the stem cells to turn into the right types of cells, really, in the end, those are just blobs of cells that are doing chemical kinds of things. And. Well, the kidney is more complicated. It has all sorts of ducts and so on in it. But that, I mean, my impression, what I've heard is that those are more sort of more accessible hearts. That's complicated because the fundamental problem of things like artificial hearts is clotting of blood. If blood is ever stagnant, it will clot. And so you have to kind of. The heart is nice and soft and squidgy and pumps in just the right way so that blood is continually flowing. There aren't stagnation points and so on in the heart if there were stagnation points. And that's when, if you have funky hot valves and things like this, you can end up with these places where the flow is not nice, laminar flow, but it can have kind of eddies and stagnation points and so on. And those are bad news because that can lead to blood clotting and then the blood doesn't flow and then the heart doesn't work and it's bad news. So I don't know, I have not heard as much about sort of using scaffolding and stem cells for hearts. I have heard about replacing heart muscle tissue with stem cell derived cells and I actually have to say I heard about that like five years ago or more and I haven't heard about that since then. So I don't really know what's happened with that. But the surprising claim was that if heart muscle is damaged, if you have a heart attack or something, heart muscle dies, then you can potentially feed in more heart muscle as you know, derived from stem cells and so on, and that those cells will sort of migrate to the right place and, and knit themselves into the rest of the heart muscle. Let's see. Well, RBS comments In future, hopefully we'll have a machine detector can detect every atom or molecule in our bodies and we can simulate solutions in a fast computer. Well, you know, there are a lot of molecules in a body and I think that computer is going to have to be, you know, it's going to have to be pretty close to being like us to be able to do that simulation. So I think what will come first is sort of broad surveillance of for example, chemicals in the blood and so on. And I think it's been a sort of holy grail for a long time to do Raman spectroscopy through the skin of what's in the blood. Raman spectroscopy is a form of spectroscopy that is sensitive to kind of molecular vibrations and that pretty much allows you to tell from the spectrum you're getting out to tell what kind of molecule you have. So sort of the idea becomes, you know, you, you have, you wear your, you know, instead of just a, a PPG sensor on your, on your smart watch type thing, you have a sensor that's doing revance spectroscopy and that's basically measuring the levels of everything in your blood. So you eat something, it can measure glucose, go up, it could measure this, go up, that, go down, all those kinds of things, things. Right now blood glucose sensors still have to actually poke inside, at least into the intercellular fluid to be able to sense glucose concentrations. But it's sort of been a holy grail for a long time to do that completely non invasively purely by looking at the Roman spectroscopy of the glucose molecule in the blood. And when that happens, probably fairly soon, around that time, it will be possible to sort of surveil a large number of different molecules. And what does that mean? I don't know. I mean, one will get. This is one of the challenges that there is starting to be a lot of data, for example, genetic data, microbiome data, things like this, where it's like you get all this data and it's not clear how you interpret it. Now, genome data has the feature that pretty much whatever genome you have when you're born is the genome you're going to have for your whole life. There are in principle retroviruses and things that could edit it, but that's exotic. And I think whereas these other things like the microbiome, like blood levels of different kinds of substances, those are things that you can fully expect to change. Maybe they could change even on timescales of order minutes. And so that's something which sort of gives you much more kind of much more information that whole time series. And how do you interpret it? Well, that's a challenge. And my kind of assumption is that in time, kind of the typical sort of pattern of understanding one's health or whatever will be like the pattern of understanding a stock portfolio or something. You can continually watch all these different numbers go up and down and so on, and you can ask the question, well, is that going to be a big problem? Is it not going to be a big problem? What can I do in the case of, not that I do this, but perhaps I should hedge this thing based on, in a stock portfolio and so on, based on this or that happening. You could imagine doing that like, oh, I'm seeing my level of potassium is going down today, right? I should make sure I have a banana for dinner type thing in really seeing in real time the adjustment of those types of things. And I can see that being sort of more, let's watch the health of things by watching the sort of continual daily kind of ups and downs. I mean, to some extent people are doing that. I like to walk my 10,000 steps a day and have managed to do it pretty much every day for the last nearly a decade now. And that I don't know whether the precise number is right, but that's what I've kind of aimed for and achieved. And it's one of these things where again, because we have the data, we can make something of it. And when we have internal data like blood levels and so on, we'll be able to do more things with it. But I think untangling what it means and one of the issues is let's Say you see that some level of this or that is doing what it shouldn't be doing. Well, what do you do about it? Sometimes that'll be, well, just look it up. It's a known syndrome and you can just go say, eat more bananas or whatever else it is. Sometimes it'll be like one doesn't know why that's happening and one doesn't know what consequence it will have, et cetera. I mean, I know for myself because I'm a data crazy person from a long time ago. I found some blood levels of different things are not in their normal range. And I went and really tried to figure out why. And at least one case the answer is I have a mutation in some transmembrane protein that leads to more efficient transport of this or that thing. So what? Who knows? Maybe that's good for me, maybe it's not good for me. Hard to tell. It's not. So it's one of these things about. It's sort of an interesting thing because you can say, well, let me always be in the normal range for everything. Nobody should achieve that because the normal range is a statistical kind of thing. And if you have enough hundreds of things that you measure it better statistically be the case that some things are out of the normal range. But then figuring out what does that mean? What do you do about it? That's a challenge. And it could be that in the 100,000 kind of diseases that have been cataloged, that the thing you see is one of those diseases and you just go look in the medical textbook, more or less, and it says, well, the thing you do about that disease is, is this weird kind of solution that was discovered 75 years ago. Nobody really knows why it works, but it seems to work type thing. And you might be, well, with this particular thing going on. Nobody's seen that before. One doesn't really know what that means and what to do about it. And having a more kind of this deterministic model of how biology works and how you can figure out from first principles what, what to do about something that you've never seen before. That's a much bigger challenge. That's really a challenge that's not been met in biomedicine. In other words, biomedicine works by precedence. We've seen this disease before. We discovered this happens. We did a clinical trial, we discovered this was a good thing, et cetera, et cetera, et cetera. The first principles are, we've never seen this before. What do we do about it is really not what biomedicine concentrates on, but yet all of us, no doubt, all the time, things are happening in us that have never happened before in the history of the 100 billion humans who've lived. They're just things that are molecularly, chemically different because you ate this or that or this particular protein flopped around in this particular way. And knowing what to do about those things is just something that's out of the domain of current biomedicine. And I think, you know, having. I tend to think that one needs more foundational knowledge about how biology works, that before one can really expect to have any kind of global approach to this kind of thing. It's something I'm pretty actively interested in actually, is sort of what kind of global theories could you have in biology? What kinds of theories? One of the things been, one of my starting points is what kind of thing is biological tissue? It's not a solid, it's not a liquid. A solid has all the molecules nicely lined up. A liquid, everything is sort of randomly bouncing around. It's something different from that. It's something where molecules are moving around, but in a very orchestrated way. How do you understand that? What is sort of the theory of bulk orchestration? That's my current kind of conceptual approach to sort of thinking about sort of the big picture of what's happening in biology. But I think you need that if you're going to have any chance of making kind of ab initio statements in biology or biomedicine. All right, I should wrap up here, go back to my day job, but thank you for joining me and I'll see you another time. Bye for now.