Software Huddle - Powered by Neurons with Ewelina Kurtys

Episode Date: September 16, 2025

Today we have Dr. Ewelina Kurtys on the show. Ewelina has a background in Neuroscience and is currently working at FinalSpark. FinalSpark is using live Neurons for computations instead of traditiona...l electric CPUs. The advantage is that live Neurons are significantly more energy efficient than traditional computing, and given all the energy concerns right now with regards to running AI workloads and data centers, this seems quite relevant, even though bioprocessors are still very much in the research phase. 

Transcript
Discussion (0)
Starting point is 00:00:00 Our objective is not to understand, really, how neurons work. We want to make them work. So it has to be a product. So that's the most important. That whether we understand or not, it's secondary. But what is the most important is that it has to work. What do you see is the biggest, like, scientific or engineering hurdles to be able to reach this, basically being able to scale this technology beyond, you know, where people are using it to then?
Starting point is 00:00:26 So the hardest is programming neurons. nobody really knows how they encode information. So that's why it's so difficult. You know, are there sort of competitors in the space or other people doing similar types of work? Yes, luckily we do, and we are happy with this because it's such a novel project that if we would be alone, we would worry.
Starting point is 00:00:46 Can you talk a little bit about, like, what one of these bioprocessors actually look like and is it, you know, are you talking a single neuron or are these like multiple neurons kind of like working together? Can you just like get into, explain a little bit, how that stuff works. Hello and welcome back to software huddle. We have a pretty special episode today,
Starting point is 00:01:07 a bit different than our usual back-end infrastructure, deep dives. We're talking bioprocessors with Dr. Evelyn and Curtis from Final Spark. Admittedly, this is an area I'm certainly not an expert in, and arguably there might only be like 10 experts on the planet in this, but it's a very, very cool concept. Final Spark is using living neurons for computation instead of traditional electric, CPUs, the advantage is that wide neurons are significantly more energy efficient than traditional computing. And given all the energy concerns right now with regards to running AI workloads
Starting point is 00:01:39 and data centers, this seems quite relevant, even though bioprocessors are still very much in the research phase. I really enjoyed my conversation with Dr. Hughes, and hopefully you do too. And with that, let's take you over the episode. Elina, welcome to software huddle. Hello, happy to be here. Yeah, thanks for being here. So I think we, we, We're talking about a really fascinating area today about bioprocessors, biocomputing, something I would say I have very little knowledge about, but I'm very interested in learning more. So you'll have to bear with me a little bit.
Starting point is 00:02:12 And I think you'll probably have to do a fair amount of explaining of the basics today. But hopefully you're prepared for that. Yes, I'm happy to start from the scratch. So we try to build computers from living neurons. So we use the same neurons as we have in our brains. and this is quite common in biomedical research that you have different type of cells in the lab but usually you do this for drug development
Starting point is 00:02:36 for understanding biology but we have totally different purpose we try to build computers and we want that at the heart of the computer are living neurons because they are one million times more energy efficient than digital hardware so what we do
Starting point is 00:02:52 we take living neurons we put them on the electrodes in the laboratory and we try to send them electrical signals and we also measure the response from them so that we try to program them in such a way. And actually the same way, neurons are processing information also in our brains.
Starting point is 00:03:11 They process electrical signals and chemical signals. And we try to do the same in our lab. We send them electrical signals, we measure electrical response, and also programmatically we can release to the medium to the liquid in which neurons are placed. We can release neurotransmitters, such as, for example, dopamine. So this is also the way to influence the behavior and learning of neurons.
Starting point is 00:03:38 You said the heart of the computer. Should it actually be the brain at the computer since you're using neurons? No, actually not, because we only use the same building blocks as in the brain, but we absolutely don't try to reproduce a human brain. That wouldn't make sense also because it's too complex. It's created through evolution. We wouldn't be able to reproduce this perfectly in the lab. So we just use the same building blocks.
Starting point is 00:04:06 It's like you take the bricks, you can build a house, or you can build a bridge or something else. So it's just the building blocks. And actually we would like to build structures, which will be much, much bigger than human brains, so we hope it will give us some superior computing power. But of course this is the future, because today we are just doing the experiments on this small scale.
Starting point is 00:04:26 scale. And so we have such a little blobs of cells in 3D, such a little bulbs. And they are around 10,000 neurons each and half millimeter diameter. So that's just for experimental. But in the future, we want to make them much bigger. So when I was in university, I remember hearing about DNA computing. And obviously this is something different. But at the time, I believe someone had shown that DNA can be used to solve like a traveling salesman. I guess is there any relation in terms of like the origins of, you know, some stuff that was happening in DNA computing to this idea or these kind of like just completely separate, you know, frames of reference? No. The only thing in common is that they are both based on living matter because, okay, DNA is also from living organism, but not it's totally different mechanism because in the DNA you try to use nucleotides, the bricks of the DNA, which are just four. For computation, I actually don't know exactly how it's done, but it's totally different mechanism because in living neurons, you have neurons which are processing electrical signals and chemical signals.
Starting point is 00:05:39 So this is actually a much bigger structure because they are living cells, they are whole, they have also DNA inside, but they are much bigger, and they have all structure, the membrane and everything what living cells has. and DNA is just a strand of nucleotides, of tiny pieces. So I guess it's much simpler. But I think maybe DNA analysis is computation is maybe more advanced. I'm not sure. Maybe it's a bit longer than because biocomputing on living neurons is possible already only since 15 years maybe because we have ways now to produce a huge number of living neurons. without any sacrifices.
Starting point is 00:06:24 In the past, it was not possible to do without sacrifices. You had to sacrifice animals and take the, usually you took neurons from the heads to have real living neurons. So that's why such a researcher biocomputing wouldn't be possible in the past. Yeah. I mean, I have no idea where DNA computing is today. I just remember, you know, like reading a story like 20 years ago about it. So, like in traditional computing, you have.
Starting point is 00:06:52 kind of this hierarchy of like hardware, firmware, there's operating systems, application software, you kind of like go up this like level of traction in the stack. Like when we talk about, you know, bioprocessors, you know, is there sort of an equivalent or are we just kind of throwing out this is the, maybe these like concepts that we have from sort of traditional computing and we have to create like a whole new conceptualization of how you would do computation in this domain? So many things work differently. But I think that this is what you said about operating system, I would say it will be the same because every computer needs some system, some high level also programming.
Starting point is 00:07:33 So, of course, today is not possible. Today you have to understand very well neurons, so you are at the really hardware level. But in the future, we imagine there will be high level programming and it will be possible to do a lot of stuff without understanding exactly how neuron is working. So yes, I think it will be the future. Although they process information in a totally different way, because we know they process information in time and space. So this is different than digital computers, which are coding zero and ones. So for sure, that processing information is different, but the layers around, I would say, probably will be the same or similar. So in traditional computing, sort of the base primitive is like zero to one.
Starting point is 00:08:18 You have like on and off essentially at the transistor level. So what is that base primitive when we're talking about neurons? So in neurons is much more complex because we have a frequency of the signal. You could say that. And there are many ways how you can interpret the signal. So you can interpret activity of neurons as spike trains. So these are just yes or no. These are just moments in time, which you usually depict as a dots.
Starting point is 00:08:45 So there is spike or there is no spike. So neurons was active or not. So that's a simple way. On the other hand, you can also measure the whole shape of the spike. So whole shape, how exactly the charge was changing on the membrane. And you can also make a lot of conclusions regarding what is the distance between the spikes. So how often neurons here is active or after the stimulus, what was the time until the neuron was active. So there are many ways how you can interpret this.
Starting point is 00:09:16 And also, we know that brain can actually process both analog and digital, kind of like digitize, digitize information, like discrete and analog. So, yeah, there are many differences. And that's why actually neurons are so efficient, because they are differently processing information in a way, which is much, much more energy efficient than a digital system. I guess, like, does the energy efficiency come from the fact that it's, since it's more, more complex, it can handle more sort of complex inputs. It can do more sort of computation for, I guess, less work.
Starting point is 00:09:53 Yes, absolutely. Is that where this coming versus, like, you have to combine so many ones and zeros together to do something with your virtual computing? Yes, and also neurons, there are many, many connections, I think, more than in artificial digital neurons. Also, you have a lot of recurrent connections, which are existent on digital, but are much more difficult. So, yes, so neurons are much more messy, much more connected. Actually, on our website, in some of our blogs articles, you can find a picture of our neurons from electron
Starting point is 00:10:26 microscopy. So it's a very high resolution, and you can see how messy it is, the structure of the neurons. So where did this, you know, idea sort of originate of being able to use live neurons for our computation? Well, I think. I think it's kind of made a natural conclusion because we see that our brains are working. So yes, so there must be something. So with Final Spark, how it started in generally. So Final Spark was founded in 2014 by Fred Jordan and Martin Kutter. These are two Swiss entrepreneurs and they are much more definitely on the engineering side.
Starting point is 00:11:06 And they had a dream of building thinking machine. Which is a dream of many engineers, very ambitious of course, but actually we can see see there is a lot of progress in artificial intelligence. So you see that AI is more and more intelligent. And engineers will always dreaming about this general AI. So they tried, final spark tried this in digital systems initially. So they did a lot of fundamental research in artificial intelligence. But what they realize is that it costs a lot of energy. And indeed, we can see today that the success of chat GPTs in the big part due to the huge spending on energy. And they realized that final spark that for a small startup is too difficult to compete.
Starting point is 00:11:50 You know, it doesn't make sense. Huge companies, it's huge funding is what you need for such a project. So they realize that if they want to stay small, independent, they have to come up with something totally different. And Switzerland is a perfect place for coming up with ideas because there is high density of many innovation. there, a lot of universities, private companies, I think it's a perfect environment. And our first employee, Jean-Marc, he had some experience already with neurons. He had a chance to do a project with Professor Markram, who is a very famous neuroscientist. He actually initiated the Brain Blue project.
Starting point is 00:12:32 It was a huge project in Europe, which I think took around 10 years. And Jean-Marc had a chance to work with living neurons. So he had some kind of experience already with this. And this is why Fred and Martin, the founders decided, okay, maybe that's a good starting point. Maybe let's try neurons because they are so energy efficient, we know. And this would be something totally new to revolutionize artificial intelligence and to solve this problem of high energy usage.
Starting point is 00:13:02 So this is how it started. So they set up a lab. They found advisors locally and they learned everything from the scratch. So it's also impressive because, you know, when you have already established some expertise for 20 years in the field, it's very difficult sometimes to change because you have to be totally ignorant again. So it's always a bit stressful, but it can be a lot of fun also. So all final park is like that. They are either biologists or engineers who had to go out a little bit from the comfort zone
Starting point is 00:13:37 and to start totally a new topic because engineers had to learn biology and biologists have to learn engineering. So that's always stimulating environments. But this is how it started. Yes, to make a thinking machine and to save energy. Yeah, my grad student, or my grad supervisor used to say that the worst thing about people with PhDs is that they now are convinced
Starting point is 00:14:04 that they know something. So they're like afraid to learn their, afraid to, like, learn new things, and they think that warning stops. But clearly, uh, these guys, uh, were, were very ambitious to go beyond just, you know, the things that they knew and start from scratch and, because they, they cared about this idea. Yes, absolutely. Yeah. The energy consumption thing is, uh, um, I think it's something is really sort of on the forefront of many people's mind. Like, I heard recently that the, like, the, the, the data center cost, like, overall energy costs is, like, gone from, um,
Starting point is 00:14:37 It was previously, you know, one to half, two percent was from data centers. Now it should lead to like 6 percent or something like that because of all these AI models essentially running on top of it. And it's like, you know, where's that end? At some point, like, you're going to run out of energy to run these things. So how do you solve that problem? And there's a, you know, a bunch of different approaches that people are kind of looking at to try to solve that.
Starting point is 00:14:59 In terms of, you know, what the final spark is doing, you know, where, what's some of the status today. Like, this feels like a very ambitious thing. You've been, the company started in 2013, so you've been working on it for, you know, 12 years now. Like, what is, what is the current sort of capabilities and how far are we away from, from some, something that's, you know, maybe potentially, you know, consumer-facing to be able to solve this issue?
Starting point is 00:15:26 So actually, it's a long-term project because it's really deep tech. So such a project, it's like with quantum computing. It doesn't get built in three years. So we expect to be able to build a biocomputer in 10 years. That's our estimation, assuming we find an investor because so far we are self-funded and by the founders who have another company, which is in digital systems, and it's very profitable. But we talk to investors. We are seeking 50 million Swiss francs, which is around 50 million dollars.
Starting point is 00:15:58 And this amount of money can change a lot in our timelines. And if we will have this kind of fundings, then we can be faster. And then we assume to be able to build a computer in 10 years around. Okay. Can you talk a little bit about what one of these bioprocessors actually look like? And is it, you know, are you talking a single neuron or are these like multiple neurons kind of like working together? Can you just like get into explain a little bit how that stuff works?
Starting point is 00:16:29 So there will be many, many neurons. Actually, we want to build something what is much bigger than human brain, maybe even 100 meters long, because we can make it much bigger. And we imagine bioserver, which will be central biocomputer, available remotely in the same way as today
Starting point is 00:16:47 you can access cloud computing. Because we think that it makes more sense because neurons are very fragile, so you have to keep them in very special conditions. Yeah, I mean, I guess it would be similar to, like, quantum computing. Like, you're not going to have a quantum computer in your home so you can play
Starting point is 00:17:03 Unreal Torbent. You have a quantum computer in like a lab somewhere in a control environment and then maybe you access it remotely sort of through a conventional computer even. Yes, absolutely. That maybe it will be easier, yes. And then what is it
Starting point is 00:17:19 that, what is sort of the neural platform that you're developing? So this is our laboratory and this is kind of side effect of our work. So during COVID, our lab get totally remote
Starting point is 00:17:33 so our engineers they developed everything remote so you can basically access the lab through the internet browser by just writing Python codes from anywhere
Starting point is 00:17:44 in the world and you have full access and everything in the lab is automated so you can run experiment you can collect data everything and because of this we decided to use this opportunity
Starting point is 00:17:55 and invite others to collaborate so initially we invited nine universities selected out of 34 proposals. We selected nine universities who have free access to our lab, to do research. And the criteria is some novelty. They have to come up with something which we didn't think about before. And there must be a chance for a publication.
Starting point is 00:18:20 So that's our criteria. And then we also surprisingly got requests, started to get requests from an industry, from big companies, from startups, even from individual, who are ready to pay us to get access to our lab and to play with neurons. So now we have a subscription, you can rent the lab for one month, and we call it Neuro Platform. And we also publish the results, I mean, the description of our lab, scientific, in the peer-reviewed journal Frontiers.
Starting point is 00:18:54 And this publication is actually in top 1% in publication. read in this journal, and we describe everything what we have in our neuroplatform and what you can do with this today. But it's mostly for, yeah, it's definitely for fundamental research in signal processing in neurons. Yeah. What are some of the things that, you know, these universities or even these private institutions are trying to do with the neural platform? So for the private is confidential. They don't have to say even to us what they actually do. And for universities, is actually everything can be found in our paper in Frontiers. We had made a list of current projects, which are a description.
Starting point is 00:19:35 An example is a project from University of Kodazur in France, in South France. There is also our PhD students there, who is also part of our team, Gregorio. And he works on the connectivity, because our neurons are run 3D. structure, we put them all electrodes, and we measure electrical activity from the surface of this structure. But he tries to figure out mathematically what are the connections between neurons inside, this which we cannot see. Because today, these 3D structures of neurons is really real black box. It's not like digital, because digital, actually you can figure out what is inside, but in real neurons you really can't. So he's trying to make a
Starting point is 00:20:27 some models so that he can calculate more or less what are the connections between neurons inside. And this is connectivity, which is actually a hot topic also in human brain. So people try to make imaging studies, try to figure out the connectivity of brain. It can be a hallmark of creativity. For example, if your brain is more connected, so it's actually interesting topic, also in medical research. But we do this connectivity on our little blocks of cells in our biocomputer prototypes. What's it mean to, I guess, like, write code for neurons? Like you have this Python API, but I'm just trying to imagine this as somebody is like
Starting point is 00:21:07 written code before, like, what am I doing? And then how does I get translated down to the neuron level? So, yeah, it's actually cool. You can write a digital system, everything. So it's a job for engineers. And you decide how you stimulate the neurons. For example, you decide when you send electrical signals. after when you collect the data, so the signals from neurons.
Starting point is 00:21:30 So basically this bilateral communication between you, your hardware, and neurons. And you can design this communication using Python script. Okay, so it sounds like it's fairly low level. It's kind of almost like assembly programming where you're like you're turning on certain things and in reading a signal. Is that right? Yes, but you have to, yes. And this is very important because it's really low-level programming.
Starting point is 00:21:58 In this case, I mean, you have to know how neurons work. Otherwise, you can kill them or, you know, maybe you collect the data which don't make sense. So, yes, so you have to understand the neurons. You have to know which kind of signals send them and how to interpret the data. Okay. How do you accidentally kill a neuron in this process? How you what? How do you accidentally kill a neuron?
Starting point is 00:22:23 Oh, what if you send too much electrical activity? It's actually the same of human, actually. You can die if you get too much electricity. If you get to contact with electrical stuff in the wrong way. So actually, yes, it will just explode, kind of like a cell will explode. Okay. Yeah, that would be bad. I mean, is there any, you know, are there sort of competitors in this?
Starting point is 00:22:53 space or other people doing some more types of work? Yes, luckily we do, and we are happy with this because it's such a novel project that if we would be alone, we would worry if it makes sense. But there is also Cortical Labs in Australia. They are a startup. They actually have different strategy because they have built a prototype for portable computers, so they try to, this different direction in biocomputers. And there is also Koniku in the US. Oh, yes, we have. We have competitors. To our knowledge, these are the two most important. So in, like, traditionally I, there's, we of course now, like all these transformer models, like a big part of that is deep learning based on neural networks. So you have this concept of like a digital
Starting point is 00:23:41 neuron. How do you see some of those concepts maybe translating into like basically like like an enliad neuron architecture? Well, we think that actually when you program neurons, it will have to, you will need totally new framework. So I think everything has to be figured out from the scratch, how to program one, and for sure that will be different. This low level, low level programming will be totally different. Then in digital, because it works differently.
Starting point is 00:24:15 We believe that everything what is today done on digital neural networks, will be much better, much more efficient on the real neurons. Do you see a certain set of applications being the best fit for this? If you reach your goal of n10 years, is it like AI applications, these kind of very energy expensive uses of computers today that are sort of the use case that you're going after, or are there other things that you're interested in? Yes, we believe it will be great for generative AI.
Starting point is 00:24:49 AI. And generally, you know, when you look at the human brain and compare this with a computer, you can have idea that brain is not so fast. It doesn't have great memory, but it's very efficient in complex tasks. So we think that kind of things, that kind of task, maybe generic ideas, solving problems, doing scientific research. That would be good task for neurons. How do you actually, like, observe and measure the outputs of the neurons? So we measure a electrical activity. And this can be seen also on our website on FinalSpack.com. We have section
Starting point is 00:25:25 live. And you can see actually in real time the readout from our neurons from the lab. Okay. Is that something that's like in quantum computing one of the challenges is that the signals are like really noisy? Yes. And do you have a
Starting point is 00:25:41 similar problem in this world? Absolutely. Biological signal is always very noisy. So also So when you use like Apple Watch or stuff like this, that's also a very noisy signal, any kind of signals from our body. And yes, neurons are also very noisy. And so we need to always filter the real data from the noise. So how do you do that? How do you do this kind of like error correction to take it?
Starting point is 00:26:04 How it does apply to filter? So it's signal processing, mathematical formula, which is just taking away anything which is below some threshold. Okay. So beyond reducing energy costs, are there other novel uses of bioprocessors that might change the way that we think about computing? So for now, we aim for the same algorithm as are done in digital. We think if this would be achieved, that would be already great success. And of course, advantages of neurons is also that you can produce a lot quite easily.
Starting point is 00:26:41 So it's very scalable. And that is proven to work because we can talk. So that means our neurons are working. Definitely neurons can process information. What do you see is the biggest, like, scientific or engineering hurdles to be able to reach this, you know, basically being able to scale this technology beyond, you know, where people are using it to then? So the hardest is programming neurons.
Starting point is 00:27:07 And so nobody really knows how they encode information. So that's why it's so difficult. And we still don't know and we do a lot of trial and error. And of course, it's also a challenge to keep them alive for a very long time on electrodes because it's not natural environment. So you have to kind of mimic what they have in nature. So it's still also, it's a big challenge. We actually have some patents, how to make neurons live longer,
Starting point is 00:27:34 but we are still working on further improvements. Yeah, so it sounds like you have to kind of like, there's a lot of reverse engineering of trying to understand how neurons encode information just to be able to make this viable? Maybe I would say it's more trial and error because we have engineering approach, actually. So our objective is not to understand really how neurons work. We want to make them work. So it has to be a product.
Starting point is 00:27:58 So that's the most important that whether we understand or not, it's secondary, but what is the most important is that it has to work. So there's a difference there, it's, you know, as long as I understand that when I give this signal in, I'm going to get the signal out, I don't really care how. it's interpreted on the insides, like a black box, essentially. Yes, although it's very difficult, yes. But in a sense, that's what we aim for, yes, that we can have some meaningful relationship. So we can maybe input some images, maybe it can recognize the image.
Starting point is 00:28:31 So that's what we want to achieve. So the neurons are actually alive. Like, what has to be done to keep them alive? I'm assuming they have to be fed in some fashion. they consume something, how's that work? Yes, absolutely. Yes, so they need to have nutrients, a steady supply. It's all in the liquid surrounding neurons, which is called medium.
Starting point is 00:28:57 They need oxygen, of course, like every living cells also in our brain. They need temperature. Everything has to be perfect, actually. Even the small vibration when you move the laboratory equipment, this already can change the activity of neurons. what's the lab actually look like like how big is this so lab is actually small it's a room where you have a special kind of like cupboard you could say with the open window when there is a septic environment so you need to every time you open neurons in the let's say open air it has to be without any bacteria because they don't have immune system
Starting point is 00:29:38 So always when you have any kind of cells in vitro in the dish, you always have to keep them sterile. So we basically put our hands inside such a cupboard where there is laminar airflow. So air is constantly filtered, so you have zero bacteria there. And you work also. So you work there with your cells and we keep them in incubators. Incubators look like a fridge, but is hot inside. 36 degrees around and some CO2 to ensure pH of the neurons on the of the medium. So, yeah, it looks quite gray sometimes.
Starting point is 00:30:23 But yeah, this is how the like typical cell culture lab, I would say. And in terms of getting to a place where you have, you know, these like, bio processors, like, do you think that those will always be, like, hybrid systems where you're using traditional CPUs and some combination with the bioprocessors? Yes, I definitely believe in the hybrid system in general, because you can see there are a lot of technologies being developed, a lot of alternatives to CPU and GPU. Some are very crazy, like quantum computing or biocomputing. Some are more traditional because you see also electronic chips, which are optimized for some
Starting point is 00:31:04 specific tasks. So generally, I see the future as a hybrid, and I think these all technologies will complement each other. Yeah, so do you think that in terms of the role that something like quantum computing might play, do you see that playing a different role than biocomputing? Yes, definitely, because biocomputers will be optimal for different tasks. I'm actually not an expert in quantum, but I think it will be good for cryptography, that's for sure. Maybe for some repetitive fast operations. So actually that will be opposite of biocomputers of living neurons. Because neurons are slow and they don't have great memory, but they are very energy efficient for complex tasks. So I would imagine there will be just different areas of expertise. Biocomputing for sure
Starting point is 00:31:52 will not be good for cryptography because that's kind of opposite where you have a lot of fast computation, which is believed will be perfect for quantum. So the computation, in my right, understand it's slow, but is energy efficient? Is that fair? Yes, absolutely. So we imagine for generative AI, for example, it may be okay because, you know, you don't need this immediately answer, but you have cheap computation power. Like today you use chat GPT, but you know, we can expect that in the future these things will become much more expensive. So when we will have alternative hardware, AI will be cheaper, so it will be also more accessible to everyone. So by slower, like, what would be a comparison?
Starting point is 00:32:35 Like, how, I guess, like, how slow are we talking? I think this is difficult because today it's only speculation because what we manage to do in our lab is to store one bit of information. So it's very early. But we can have some idea looking at the human brain that, you know, how fast we can think and what can we do with neurons which we have inside our heads. and then what is your role within final spark so i work on the commercial site i actually talk not like today on the podcast also with a journalist we talk with investors because we are
Starting point is 00:33:14 fundraising as i mentioned we have a scientific collaboration so i also search sometimes or establish the collaboration um what else and actually we have also clients so that's also I care about this. So any kind of external affairs. And sometimes I also take part in some R&D discussions, but I'm not working on
Starting point is 00:33:40 R&D anymore. And in terms of like cultivating these like living like neurons, like is there I guess like
Starting point is 00:33:52 push back from certain pockets of the world that I don't know, using some sort of living representation of a species is, is, you know, bad in some way. Like, what are sort of the general reaction to this? Most of them are positive and enthusiastic, but yes, we get some questions, is it okay? And we are, of course, not competent for to talk about ethics, but we reach out to philosophers. We put a lot of effort to promote this topic among philosophers so that they can help us to
Starting point is 00:34:24 answer these questions. Because, of course, we want that our work is accepted by society. and every new technology is bringing new questions, it's normal thing. So we want to prepare for this and make sure that our work is accepted by society. And if there is anything we have to do to make it ethical, then of course we want to make sure we do this. Yeah, I mean, do you see any challenges around? If you got like a really big biocompator with a lot of neurons that was essentially relatively mapped to, you know, how many neurons were going in the human brain.
Starting point is 00:35:04 Like, is there, would there be questions around things like consciousness? And, you know, is this a living organism versus a computer? Yes, definitely. Consciousness, I think it's now a hot topic. Many people like to talk about this. I think it's a bit exaggeration. But, of course, I'm not an expert in this. So that's question for philosophers. And definitely that's some of the, one of the big.
Starting point is 00:35:28 questions we need to address, of course. There are also questions about digital AI if it will become conscious. And I think these questions are coming when AI is intelligent, when it has some superior skills, then we question, oh, is it like us conscious? If it can talk like us, maybe it's also conscious. Even some people already look for some qualia,
Starting point is 00:35:51 like they call it, some traits of consciousness or human traits in the chat GPT. So there are also some, already some papers on that. So I think it's a normal thing which people are thinking about. I don't know. I led this to answer philosophers. I think they are much better, much more competent. Yeah, I mean, that's fair.
Starting point is 00:36:15 I think that, you know, people have been comparing various technologies to, like, humans and behavior and human brains for very, very long time. Like, even going back to when, like, telephone operators used to, take the plugs and plug it into other things to route telephone calls manually. There was analogies of how that map to human brain functionality and so forth. And now it seems kind of silly. But I think where this gets a little bit more challenging to sort of dismiss is the fact that there is some sort of living component to this.
Starting point is 00:36:49 There is something that is sort of rooted in biology, whereas even a neural network is just electrical signals over, you know, GPUs or CPUs, depending on how you're running in it. So I think it's easier to argue that it's, you know, not necessarily human. But there are, of course, all these questions are in AI in general of like, you know, like, does it matter? And, you know, are you able to mimic human intelligence? And it's a big, big topic of the conversation, of course, right now because AI has, you know, sucked up all the oxygen in the world in the last couple years. Yes, absolutely.
Starting point is 00:37:23 Do you think that, you know, the work that you're doing, the final spark and the idea of, you know, biocomputation, is it changing the way that we think about things like, you know, the intelligence, learning, or, you know, other things either in technology or even in sort of the biological world? Like, is this helping inform our understanding beyond just the goal of, hey, we want to create more energy efficient computation? I think maybe it will not change the way of thinking, but definitely it can help us also to interact with human brain. We had such ideas that if we are able to teach neurons in vitro, then maybe we can also teach human brain by sending some electrical signals. And who knows, maybe we can upload, for example, new language in your brain. So yes, there are definitely such ideas. And we think that our work can benefit to anyone who's doing brain machine interface or, Maybe this futuristic ideas which don't exist yet, but about brain stimulation.
Starting point is 00:38:27 Or maybe on the other hand, stuff like Neuralink, which is trying to remotely control stuff with your thought, that also requires understanding what brain activity means. So we think that our insights from learning in vitro can be helpful for that, because we believe that the general mechanism will be the same. So it doesn't matter if neuron is in the dish or in the brain, the learning mechanism should be similar. Great. I mean, so is there anything else that we should talk about that I didn't bring up?
Starting point is 00:39:05 No, I think we quite covered everything. Maybe it's worth to mention that if anyone would like to get in touch with us, it's very easy because we are open for communication with anyone interested. So at finalspark.com, our website, it's possible to fill in contact form. Also on the team section on our website, their email address of everyone from us. So it's very easy to contact us, also LinkedIn. We are very active. I encourage everyone to follow us on LinkedIn or X.
Starting point is 00:39:39 And we also have nice Discord community, also for technical people, for our users, for enthusiasts, Many people interested. There are a lot of discussions there about our work also from the technical side. Who are, do you come across that are like very sort of generally interested in this topic? Is it people coming more from the biology side or is it more from the technology computing side? I think it's more from the technology side because to work on a, for example, neural platform, it's necessary to code. So this is definitely a job for a programmer
Starting point is 00:40:20 and I really think that the barrier to entry is much bigger on the programming engineering side. But I'm from biology so maybe I look like that way but I think it's relatively easy to learn some basics about neurons you just have to read about this but whole programming, you know, data analysis, visualization, this is a long story.
Starting point is 00:40:41 So yeah, definitely But what I also notice is that people who are really like computer scientists are not always interested by our work, because they are so much focus on digital. So people who really like us, as mostly maybe physicists or mathematicians, so people who are able to code, but they are not computer scientists. That's at least the trend which I have seen from all these connections we get, although we also get questions from computer scientists, definitely. Okay. And then what's next for you guys besides the fundraising? Learning. Learning in vitro. That's our big challenge. That's the most important problem on which we are working now. And we are, of course, making small changes because, you know, it's really long-term projects. So, you know, it's not such a fireworks every month, a big update. But we have newsletter, which we are sending every six months.
Starting point is 00:41:37 So it's also possible to subscribe on our website. And we are making always some small improvements. For example, now we included serotonin, so we don't have only dopamine treatment, but we also also serotonin. So we are constantly improving our hardware on such a small technical level. So there is constantly something going on. So what's that give you, you know, that change that you're making from dopamine to the other ways? No, it's not a change, it's addition.
Starting point is 00:42:06 So actually, we would like in the future have a whole variety of different. neural transmitters because in the brain we also have full cocktail of different substances which are affecting neurons so we would like to achieve this also in the lab so we are adding new and new slowly well great dr velina curtis thank you so much for being with us thank you so much for nice questions and it was a pleasure cheers thank you

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.