Space Summary
The Twitter Space AI Trends You Should Be Paying Attention To hosted by KIPprotocol. Delve into the realm of AI trends with a focus on $KIP, the Web3 AI base layer that garnered acclaim as the winner of the 2023 Chainlink Hackathon. Partnered with industry giants like @AnimocaVentures and Tribe Capital, and serving as the official AI partner of @opencampus_xyz, $KIP stands at the forefront of innovative AI solutions within the Web3 landscape. Learn about the cutting-edge developments, strategic collaborations, and educational AI applications that define $KIP's pioneering role in shaping the future of AI technology.
For more spaces, visit the AI page.
Space Statistics
For more stats visit the full Live report
Questions
Q: How is AI evolving within the Web3 space?
A: Projects like $KIP showcase the advancement of AI technology within Web3 environments.
Q: What does $KIP's win at the 2023 Chainlink Hackathon signify?
A: The victory highlights $KIP's innovation, competitiveness, and industry recognition.
Q: What significance do partnerships with @AnimocaVentures and Tribe Capital hold for $KIP?
A: These partnerships signal strong support, expertise, and potentially increased adoption for $KIP.
Q: How does being the official AI partner of @opencampus_xyz benefit $KIP?
A: This position offers $KIP extensive visibility, credibility, and opportunities for educational AI advancements.
Highlights
Time: 00:15:42
Introduction of $KIP: Web3 AI Base Layer Exploring the innovative features and potential impact of $KIP in the AI landscape.
Time: 00:25:18
$KIP's Success at the 2023 Chainlink Hackathon Discussing the significance of $KIP's achievement and its implications for the AI industry.
Time: 00:35:55
Partnerships with @AnimocaVentures and Tribe Capital Analyzing the strategic collaborations and support systems backing $KIP's AI initiatives.
Time: 00:45:29
$KIP as the Official AI Partner of @opencampus_xyz Understanding the role and benefits of $KIP's partnership with @opencampus_xyz in educational AI advancements.
Key Takeaways
- AI innovation in the Web3 realm is rapidly evolving with projects like $KIP.
- $KIP's victory at the 2023 Chainlink Hackathon underscores its potential and recognition.
- Partnerships with @AnimocaVentures and Tribe Capital indicate strong support for $KIP.
- Being the official AI partner of @opencampus_xyz positions $KIP at the forefront of educational AI solutions.
Behind the Mic
Introduction and Context
It. Every time. Make it do it makes us. Better. Faster, stronger, more than power hour never ever after work is over before. Better, faster, stronger, working ever. Over. After hour. GMGM everybody. Welcome back to another banger show with Kip Protocol, web free base layer for AI. Kip, winner of the 2023 Chainlink Hackathon backed by Animoca Ventures. We have Julian, the CEO and co founder of Kipp Protocol, with us today, and a number of banger guest speakers. We have momate AI consumer platform. We have Azeeb with us today who also works on bolts Net. We've got the Bolt man network with us also on the show, building the truly decentralized and private AI platform. We also have Jen with us today, the co founder and chief AI officer of Kit Protocol. And we have in with us today the first modular AI native data pre processing layer backed by Binance Labs. I'm so happy that I was able to get through that without tripping over my words. We also have Jay with us today, the founder of Boltsnet. We got basically just a number of incredible speakers on the show.
Show Structure and Audience Engagement
Me and Julian go on a co host by co host basis on this one, so it'll be a bit of a different show if you had some of the ones in the past. And for the listeners, let's get the likes and the retweets out of the room. And what we will try and do is, if anyone has any questions during the show, hit us in the comment section in the bottom right hand corner, and we'll try and get to those during the end of the show. And if you do want to hop up and stage during the end of the show, we also allow that for any questions you might have for us or any of the incredible guest panel that we have. Julian, are you excited for today's show? How is it going, homie? Hola. I'm doing. I would say, well, I have a terrible neck ache, but otherwise I think today it. Oh, we lost you there, brother. The projects. Oh, did it? Did you? I'm. I'm in a hotel with, like, a very, SaS connection, so cut me off if my connection proves to be like this.
Projects and AI Discussions
Okay. Yeah. I was very excited that we have so many great speakers on space today, and I would love to hear more about their projects as well, because we are now actively in the phase of trying to link up more and more AI related projects in the space, both from a tech perspective and also from a PD perspective. So, yeah, it's as much I'm going to gain as much from space as hopefully all of our listeners. So I was wondering, why do we start the show off with the introduction by our speakers about themselves, a bit about themselves, a bit about that project. So let's kick it off with Mo mate. I know Ahad is behind the mic from Moemate. So ahad, why don't you kick us off? Hi, Julian. Thanks. Hope you're feeling better with your next situation. So ahad here, founder at Momate, we're basically, you know, building an ecosystem. Focused. We are trying to. You guys can hear me, right?
Introductory Challenges and Backgrounds
I can hear you, Momi. I think, Julian, I think your rug of a Wi Fi in the hotel may be doing you some disservice here, homie. But Mo mate, we can all hear you aside from Julian. So I will bridge the gap and hopefully bring Julian back into the loop on this one. Okay? Yeah. So quick introduction. We are just working on the future of entertainment with Aihdem where we're working on sort of a UGC platform, you know, making things like, you know, chatbots and then, you know, going up to like, you know, full fledged simulations, sort of like a mini game engine and along with that, you know, a couple of other projects as well. So that's a quick introduction about us. As for my background, you know, I've been in product for a while, shipped, you know, software as well as hardware products across like, you know, startups, big tech such as Apple, etcetera as well. That's about myself.
Discussion on Momate and Wi-Fi Issues
Yeah, definitely love that take. I think. Look, to compete with the hotel Wi Fi issues currently going on with Julian, Jen, you've just been promoted to co host of the show, so we're gonna keep this one cooking. Jen, what's your take on momate's reaction? And yeah, welcome to the show. Maybe intro yourself a little bit and then we'll get the intros from the rest of the team. Yeah. So I am Jen, I'm in charge of AI and related matters at KiPP. So that involves language models, video models, hooking them up to front end pipelines and so on and so forth. And yeah, just very happy to be here with all of our great guests tonight. Me too. Me too. And look, we've got bolts, man. Network with us too. I'd love an introduction from you guys. What do you do? What you're cooking, what you're building. And then we can hear from Razeeb and Jay as well.
Overview of Boltzmann Network
Yeah, I think Jay can take it, but yeah. So at Boltzmann network, we're building a decentralized AI platform. That we're the only platform that splits AI models themselves among a number of different decentralized nodes. What that means is that far more devices than traditionally can contribute computing power to the network because the actual needs of the device are much lower if you actually split the models into different layers as well. It inherently protects the privacy and ownership of your data and information because we're not harvesting it, using it to train our models, giving you real sovereignty there. And then as well, from a censorship and usability perspective, it also greatly assists with model verification. And I think Jay as well, can kind of go into it additionally.
Privacy and Decentralization of AI
Yeah, I mean, I'm happy to be here. Like a great idea and really glad to be here. Yeah. So we're building something that I think is really unique. We're focusing on privacy first, AI and privacy issues only going to become more evident as time goes on. And everybody knows if you ask questions OpenAI, they log everything. So, yeah, that's what we're building. We're also decentralizing compute, and I think this is a much needed effort. I love it. Jay, can you give us a little bit more context over the decentralizing compute? What do you mean by that, though? Yes, absolutely. So, largely. Hi, everyone. Because I got some lagging from my side to hear you guys speaking. Yeah, if you guys can hear me, I can make a short introduction from.
Communication Issues in the Discussion
My side, I think. I think. Yeah, we've got Jay talking at the moment, so I don't know if you're not able to hear some of our other speakers, which was the same issue that Julian had earlier. So maybe hop back down and back up, and that should solve the issue for you, and then we'll get right to you. Yeah, sure. I can make a quick Jay. So. Hello, everyone. This is Jenny. Okay, I think Din will go. Jay next. Hey, Din. Yeah, sorry. I think there's a couple of mic issues on your end, so you're not able to hear all of our speakers. We did just have Jay speaking, but if you want to do your quick intro, and then we'll throw the mic back over to Jay, because I think there's a few connection issues. Once you've done your intro, maybe hop down from stage and hop back up, and it should solve some of the Internet issues.
Technical Backgrounds and AI Trends
We've got some of the speakers not hearing other speakers. All fun and games when we're on X people. Elon is not paying it the most attention, it seems. Jackie. And then hopefully we can hear from din afterwards. Absolutely. Yeah, so, yeah, so large language models allow for some sort of structuralism, and that is allowing for models to be split along layer wise and parameter wise. And that allows for different computers to do partial computation of a inference task and pass on the result, the intermediate result, to another computer. So that what that means is that now we can have computer, one computer doing some layers of computation and then another computer doing another layers of computation and then passing on the results in a privacy preserving manner.
Privacy Issues and Final Thoughts
So that way what you're asking can really be easily reconstructed. I think privacy and AI, just imagine data breaches that happen all the time and what kind of things that people are asking to their AI agents or chat. GPT privacy is going to become a major issue in coming months, if not weeks. Okay, loving this, loving hearing this so far. So for the listeners today, likes and retweets, let's blow this one up. As we dive into the main topic of the conversation today, AI trends you should be paying attention to. And as you've just heard, we've got some incredible guest speakers with a lot of experience in the field to really back this one up. And I think it's one thing that Kip do really differently, actually, is, you know, we're not just talking about this broadly. We've got real builders in the space, in the AI ecosystem contributing to the conversation today, and Jen and Julian, who are absolute gigabrains when it comes to this stuff.
Introduction to AI Trends
My technical advice for you, my qa. Yeah, exactly, Jen. So I think for me, I'd love to get on to the subject matter now, which is AI trends you should be paying attention to. I know you're deep into the field and Julian would normally have a witty remark to call you a nerd right now. And based on my past experience of the show. So I'm going to throw the mic over to you and give, let's just start this conversation like what AI trends are you paying attention to right now that you believe our listeners should?
Current Exciting Trends in AI
Yeah, so obviously the, I think the huge thing this week has been in the whole entropics thing on Twitter. So it's this guy who is doing, trying different sampling methods to minimize hallucinations in LLMs and. Yeah, I, I think I, I've been kind of tracking that as everyone in AI has this weekend, some interesting papers about reducing the energy costs of LLMs by replacing the floating point multiplications with tensor additions. So that is kind of the less appreciated discovery this week. But I think it's extremely cool and could potentially have even more impact in the longer run. But those are kind of the two that sprang to my mind. But I would really like to hear what Alpha the other guests have to drop on this.
Introducing Din and Their Work
Dean still isn't in. Oh no, Dean is in. Sorry, Din is in. Let's do the intro. Shall we finally get to do Din's intro? If you can talk right now, I would love that. Din, if you are available, would love an introduction from you, who you are, what you're building and all of the ability and why you're on the show today around AI trends. Yes, you can. For crystal clear. Nice, nice. Nice. Hello everyone. Hello. Thank you for inviting us to join this space because we're very close partner of Kape Protocol and this is Jenny. I'm the head of marketing from Ding and we're building this AI native data pre processing layer right now, basically on BNB chain ecosystem.
Future Goals and Insights on AI Trends
And yeah, starting from this year, our main goal is to launch our token at the end of this year. So recently we're more focusing on our own node cell and also node mining. And in the next few weeks we're more focusing on building some big campaign with binance, web three, wallet and other ecosystems. So just stay tuned on our latest update. Regarding to the topic today, I'd like to share some insights from our project from our perspective, because we are building this AR native data privacy layer. So we like to empower everyone to join the AI data collection, validation and also the vectorization process, because in some developing countries all the people are being used to get involved in this data training and also the data collection process of the AI development.
Empowering Data Collection through Decentralization
So we want to use the decentralized way and also the crypto help to empower everyone from those countries to join this AI data process and development. So I think for Ding, for us, our main goal is to provide this kind of network and also a platform for everyone to join the AI data collection, the validation and also the vectorization process so that they could really on their data they collected and also they can earn from the whole process. Yeah, that's what I want to share first. Thank you so much.
Panel Engagement and AI Concepts Clarified
Thank you. I love that and it was worth the wait. Thanks. Jen and Jenny, that's not going to get confusing. Look, I think this has been really great getting to know the panel, getting to know who the experts are on the show as we dive in. And Jen, you gave a really good little breakdown to begin with on like the areas of AI and the trends that you're looking into. One thing that I think we do a little bit on this show, though, I'm going to call us out for this because I think we're all definitely, you know, susceptible to this is LLM. Like we start using terms that maybe not all of our audience understands what they actually mean. So large language models for the people up in the listener room.
Explaining Large Language Models
Jen, do you want to maybe go into a little bit more context on the trend side on these LLMs that you were speaking to? And then we'll get to Jay, we'll get to Boltzmann network as well, and we'll get to din for a take on what they're paying attention to right now. Oh, yeah, sure. So LLM is a large language model. So that is something like GPT or Claude or deep seq or llama, the models that do your homework and pretend to be your girlfriend in real time. And, yeah, so the trends that I mentioned, I think I mentioned, too. The first is to do with the way that the model picks what is likely to be the most probable sentence completion, how it does its autocomplete, because the newest GPT, they seem to have come up with a way to kind of get the model to think more carefully about the answers that it's going to give.
Exciting Developments in AI
And everyone was sort of rightfully very impressed about that. But then there's also this other guy on Twitter who's just been kind of playing around with it by himself, XJDR. I couldn't remember his name earlier, but yeah, that's his handle. And he's got a GitHub repo out now just demonstrating how he's using what is called entropy based sampling to kind of replicate this in an open source way, which is really exciting. And then the other one that I mentioned was getting really deep into the maths of how you pick the most likely next word. So the traditional way of doing this is via floating point multiplications. And some people have worked out how to do this using addition instead of multiplication, which uses a lot less energy. So that's a much more economical way of doing it.
Looking Ahead: Trends and Developments
And it's quite interesting. So that's my kind of two hot tips for this week, and I think I shared, I haven't shared the entropy thing on Twitter yet, but I've shared the other one on my Twitter today. So, yeah, if anyone is interested in this, go and take a look. But I really want to hand this over to the guests because I have been so busy this week and I only really noticed the biggest news items, and I'm sure I've missed huge amounts of alpha from Twitter and elsewhere. So, yeah, please drop your hot tips on what is the next new thing in AI?
The Insight from Jay
Love it. Jen. Yeah. Let's go over to Jay, then we'll go over to Boltzmann on this one. Exact same question to you. Like, what has caught your attention in the AI space? What trends are you paying attention into right now? Jay, over to you. I mean, so from our end, like, distributed computing is a trend that I follow a lot, and there's a project called Exo that's been doing essentially what we want to do, but better so we're always catching up and trying to catch up and do much better ourselves. But yeah, they already have a local LLM running on consumer grade hardware, and you can just run, for example, MacBook or series of them with really high end, high fidelity models.
Local AI and Distributed Computing
That's got billions and billions of parameters. Distributed computing side is coming along nicely. I think that you will see more and more consumer grade hardwares, even cell phones being used to host part of the computation. Local AI is something that is near and dear to my heart, but to get to a point where we can host local AI, we're going to need a lot more computational power and individual machines. So that's why distributed computing is necessary. So, yeah, those are the trends I follow really closely.
Emerging AI Trends and Developments
There has been a lot of more recent sort of like, you know, visions, statements that are coming out from various projects about how like, all, every phone will be an AI, like, compute node or something like that. So that's what I'm following. Yeah. There was some news this week that I'm not sure on the distributed side, but that, like, moving forward with each of the new iPhones, that Apple is trying to do a lot more to actually integrate AI into every aspect of the way that you personally use their phones.
Centralization vs. Decentralization in AI
It's interesting with the ability of a large centralized provider to harvest a lot of that information, because I suppose one of the benefits of Apple historically has been their encryption technology and that very difficult to break, and that there won't be as many people spying in and having access to your data on this. You know, if they're going to be rolling out like a large, you know, AI like suite as like a huge part of their services, might be difficult for them to be able to bridge those two things as well. I guess. There were two Nobel prizes for AI that occurred, I believe, within this last week, or at least that involved AI.
Nobel Prizes in AI and Genomics
I mean, one of those. I'm not sure if Razeeb is at the mic currently, but much appreciated being on the stage to help boost our space to all of his followers and everyone in our ecosystem. One of them actually has to do with the prediction of various protein sequences. Talking about human genomics and health care and all the potential benefits that can be done there. I believe it was. See here? Yeah, it was awarded to. It was a bunch of. And Jay could speak to this as well. It was a bunch of the guys from DeepMind who were granted these awards for the genomics, and then in recognition of some previous accomplishments that had made the current AI boom possible, really but I think those were from years ago.
Reflections and Future of AI
Absolutely love this so far, I think so many follow up questions. I think to set the stage for both our speakers and our listeners, we have some of the foremost experts in AI on the show right now, which means you do get those giga brain takes. I am here playing the guy who asks the stupid questions so you don't have to. So for the listeners, if you're like, what? Don't worry. I will probe. I will try and get these takes a little bit more dumbed down for both myself and for anyone else in the audience who maybe isn't as intellectually set up as some of these absolute giant brains on stage.
AI Developments and Expectations
I think we haven't heard from a hard yet on momate side. So momate would love the take from you on this. What's got your attention? And, yeah, be kind to me and potentially anyone else like me in terms of, you know, breaking this down for. For the everyman or maybe, you know, slightly lesser than the everyman in my case. I mean, of course, I think the biggest news is the Nobel prizes, right? Like two consecutive Nobel prizes. You know, funnily enough, one in physics, the other is in chemistry. So I do think that, you know, it's definitely foreshadowing.
AI's Impact on the Future
What's to come is that I think anybody who is in the doubter scamp of, like, you know, what AI is about to achieve or unfold should understand that. I think we really are talking about intelligence at a scale which is going to solve much bigger problems than just chat bots or chat GPT or image generation. I think those are just the technologies we get to see on the consumer side. But I think the impact of. Of this technology is far and beyond. And even the science community sort of, like, you know, realizing it and sort of trying to, like, adopt it as well as, like, claim it.
Concluding Thoughts on AI Trends
Right? So that was interesting. But overall, trends, I mean, I'm just. Just like, to echo what everyone said is that, like, you know, it looks like that these models keep on getting better, faster and cheaper. You know, like jen mentioned about, you know, the XDJR experiment. But other than that, I think very interesting development which has been happening recently is by this company called Liquid AI, and they basically, you know, made, like, non transformer based models. So, like, just for everyone's understanding, if, like, GPT, like, the t in GPT is, like, transformer.
Exploring Beyond Transformer Models
And one of the limitations of these transformer models typically is that they are limited by context size. So these new models, they are non transformer. Based, which means that we might be soon going towards infinite context. And if we go towards infinite context, I think a lot of these use cases which currently we think are not feasible or this is going to open up. So for me, that's something very interesting happening in the field of AI, because right now all of our scaling laws and everything is still basing off of transformer models. And one of the biggest limitations there seems to be context size.
A Glimpse into the Future of AI
And as soon as the context size limitation goes away, I think, yeah, we can start dreaming even much bigger than what we are already dreaming right now, what is possible with. Yeah, yeah, that's both fascinating, terrifying, and also completely, it's just mind blowing, right, that this, you know, this world and like the rapidity of AI at the moment, the speed of which it develops and how quickly it just seems to be supporting all different ways of life, like Nobel prizes, guys, like, basically the award for the thing that we identify as, like the thing that's pushing humanity forward in its given areas as much as possible, and AI is now contributing to that.
Understanding the Impact of AI Advancement
So I just, I think it's wild to think about that and then to think that there's alterations that are being made right now that are actually going to open that up even further, allow AI to develop even quicker and to deal with way larger volumes of information, data and manage all of that, I think, yeah, you genuinely can't comprehend where this is going. Razeeb, I can see you've just come off mute. Welcome to the stage. Great having you on. How is it going? Are there any trends, anything in AI that, you know, you're specifically paying attention to?
Razeeb's Insights on AI Trends
And then we'll get over to Julian, as I saw the hand raise there. Yeah, sure. So can you guys hear me? Yes. All right, so, you know, I'm here partly because basically, you know, in my field, the field that I came up in genomics, you know, we've had an explosion of data, we have structured data, we have an explosion of various analyses to do, you know, and obviously, machine learning and artificial intelligence are the only ways that we can really scale our understanding of biological processes, biological mechanisms, and also give people biological insights that they need.
The Evolution of Biological Data Analysis
So just give you a concrete example, let's say the year 2000. If you were taking a graduate level biology course, your data analysis, maybe you would do some stata or something, but really a lot of it is just Excel spreadsheets and maybe a couple of hundred data points.
Understanding the Human Genome
As a lot of you guys know, the average human genome, well, not the average human genome, just the human genome has 3 billion base pairs, okay? The average human has 5 million single nucleotide polymorphisms and hundreds of thousands of structural variants. Data is just on orders of magnitude scale, like multiple orders of magnitude scale larger than it was a generation ago. So we can still do the things that we could do a generation ago, target these specific genes, like look at the bricket gene, you know, breast cancer gene. You guys have probably heard about that, but there's a lot of other things in the genome that we need to look at, but we don't know we need to look at. And part of that is because there's so much to look at. And so automating that process, prioritizing it, allowing artificial intelligence to do a lot of that is actually going to be pretty essential for us to get the real diagnostic insight that we've been talking about for a generation that hasn't really been here.
Security Challenges in Human Genetic Data
Another issue for human genetic data, and apologies to my friends who were working in drosophila or maize or whatever, human genetic data has particular security needs. And so when we're talking about distributed computing, distributed AI, there's, I think, going to be some worries that if you just put it on AWS, if there's a security hole, well, you have one copy of your genome and you're never going to get it, like a new Social Security number, right? So that's going to be a problem. So that's why I'm actually interested in Boltzmann AI solutions. You know, like in terms of security. Instead of locking things down now, you're going to trade some efficiency for it, you know, but I think that's, most people say it's worth it. Also, legally, you probably have to do something like that to cover your ass. Now, to be candid, the laws are a little outdated, but, you know, I think they're going to catch up. Europe has GDPR and stuff like that, and there's a lot of problems. A lot of, you probably know what GDPR, you know, human genetic data, human medical data, all these things, they're in a different security category.
Commercial Models and Data Protection
So they probably have to have a different solution than, for example, when you're looking at your marketing data for like what you purchased. Okay? So, you know, I know people who've gone into this bio artificial intelligence biological space who come out from other, come in from other parts of big data. A lot of it has to do with financial transactions, right, but financial transactions, I mean, I don't think they really define who you are. If there's a security break and, you know, people know that you buy Danielle steel novels or whatever. I mean, who really cares, right? I mean, a lot of people will care actually, but you know what I'm saying? So I think that's what I was going to say. I'm excited. We're very early on and, you know, it's not just fun and games in my domain in terms of biological science, of medical, health and technology.
Sales Trends in Data Security
Humi, note to self, don't take the mic after Razeeb talks. It's a very hard act to follow. But interesting take. So one of the reasons why for my rug problems with Twitter is I'm on a business trip. We are selling to universities. One of the interesting trends that has become very clear to us on the sales side is that KiPP, we have always been very focused on rag solutions. That's one of our core emphases, because we believe in keeping the data separate as far as we can, which gives the potential and the possibility for keeping it secure, keeping separate, at least keeping it out of the hands, the people who would take it to train their models. Right. And all of that, a lot of that has already happened.
Proactive Data Protection Awareness
So the trend that I see very clearly is that few months back, just even, just like three months back when were talking to the same universities, the same clients, academic institutions and whatnot, right? And first question they ask is, okay, so what is your whatever your reg thing, right? How is it different from chat GPT? What do we need it? We already have a subscription in chat GPT. Now the questions that are coming, is that, okay, how do we prevent chat GPT or OpenAI from making film, stealing our data, essentially. So they are asking us proactively for data protection, knowledge protection, which I think it's increasing awareness. The fact that the, a lot of these large commercial models, they didn't come magically, they came built on the collective knowledge of all of us.
Implications of Data Ownership
So that I think will result in more, I would say, well, awareness will result in a lot more customization in terms of the entire pipelines and the frameworks that are being used to deploy AI. That's one trend that is very relevant for us. And there's another trend I'd like to start because, and picking on, not picking on, but following on to what Ahad said just then about the Nobel laureates. So I don't know if many people caught this, but Jeffrey Hinton in his acceptance speeches is one of the proudest moments was when his student fired Sam Alan, you know, so I thought that was pretty funny in the beginning, but I thought that, hey, you know what, this is a trend we should start. Everybody should be looking more closely into what Sam is actually doing.
Discussions on Genomics and AI
He's done a lot of good things, sure, for the AI space, but everyone should look at it in its entirety. So this is a trend I would like to start. So who's with me? Everyone raise their hand here. We should all fire Sam Altman on a semi weekly basis. Look, I know I gave Elon a little tough love earlier about how spaces is falling apart at the moment. I don't generally like to make enemies of some of the most influential, crazy powerful people on the planet. So I'm just going to take a slight step outside of Julian and Jen's take here and say, guys, I don't know how I'm in this space, I don't know how I got here, but these thoughts aren't my own.
Exploring the Importance of Genomic Data
Or whatever you have to say to protect yourself both legally and from the powers that be and whilst hopefully also not incurring the wrath of Jen and Julian. And what this you are doing now you're pushing out. Come on, come on. It's good publicity. Okay, do more. I can just, I can literally picture how chat GPT the next time I ask her for any sort of advice, it's just going to be like, well, you know what really helps is if you walk in front of a train, then that will do something to you that literally, I can just see it happening now. If you don't hear from me next week, this is Julian. This definitely means Julian's got me killed by Sam and the overlords of AI.
Topics for Future Exploration
Look, guys, this has been fascinating. I do want to, I do want to open up the conversation and Jen, Julian, I'll throw the mic back over to you two first and then you guys can decide how to open this up further. There's been so much and that's been opened up on the Nobel side, on the AI side more generally. But also I think the genome side is very interesting and will be interesting to the audience as well. But we talk about data protection, we talk about security. Why is it so important that something like your genome doesn't just end up freely on the Internet? Like just in case, just playing devil's advocate here, why not? Why can't everyone have access to my genome and would love your take on that?
Considerations on Genomic Sharing
And then also what the implications could be on the positive side, as well, because I don't think we've touched necessarily on the extent of how important this could be for really for the improvement of society and the healthcare system for basically like we've never seen before. I think this is Webcare 3.0 we're coming across right now. Yeah, I mean, if, honestly, if someone wants to go out and make a million clones of me, I hope they do. I think the world needs more copies of me, infinite. But, yeah, I guess it is kind of an open question right now. We rushed into a lot of this sort of 23andMe stuff, and now 23andMe seems to be going bankrupt and they're selling off their stock and the borders resigned and, oh, suddenly now your DNA is going to be sold to a.
Exploring DNA Ownership and Ethical Concerns
A holding company in the Cayman Islands or something. I don't know. Yeah. So this is kind of something that we're still exploring right now. And I'm really interested to see what the guests takes are on this because I think this was a subject that came up in an earlier space that we did that. Yeah, it's great to have the power to say no to this stuff, but you also potentially want the power to say yes to it. Like, if someone wants to make copies of my DNA, go nuts. I'm really curious to see what you do with it. If I get royalties on that, great. But if not, yeah, whatever.
Privacy and Consent in Genetic Data
But then other people will be equally opposed. So, yeah, I'm curious to see what the guests have to say about this. I don't know if we want to. We want to bring Din back in on this. Din has been quite quiet. Yeah, I can't currently see Din as a speaker. I have been trying to invite them back up on stage. Din, if you are a speaker, then. Then do hop back up. But otherwise, yeah, I think we can maybe get to. I've seen Jay with a bunch of reactions, so maybe we go Jay first and then hopefully give yourself a reset. Again, sorry, Jay, over to you on this one.
Concerns Regarding Genetic Discrimination
Oh, I mean, the negatives are things like, I guess, you know, genetic discrimination type of thing. You know, I don't know if that's a big thing right now, but, you know, your insurance company could look at it and say, hey, we don't want to. We will have to charge you more, know, premium or something like that. So I don't really know the full social implication of this. I mean, genetic screening. And it's done, you know, in family planning quite frequently nowadays. So, you know, I could see that being an issue. And also it's just, I think just generally for people, like factor, right. It's like, it's as though someone else knows, like inside of your house kind of thing, you know?
The Need for Privacy Regulations
So, yeah, I see, like a big push for privacy regulations surrounding this. But I also think that, like Jen said, you know, there ought to be an opt in if, you know, if you want to share your data, I suppose, and maybe even allow for some way to monetize it. And I think that might open up a quite a, like, ample, you know, source of data for genetics researcher, but only if, like, someone consent to it. And I think that's the important part. Yeah, I do have this dystopian idea in my head is like, we've talked about this a bit on the show already, but it's how quickly AI is advancing.
Concerns Over Future AI Developments
And we've talked about, you know, ways in the advancements become even bigger than what we've seen already, which is a thought that can't wrap my head around. They literally can't comprehend how AI could improve any faster than it currently is. Because if you literally think of even a year ago, most people weren't even using GPT, like a part of my life. And I'm not the biggest, you know, I'm not the biggest, like, tech guy. I'm not like the first mover on some of this stuff. Imagine if it develops in a way where it can understand the genome so well that all of a sudden you can stop being discriminated against from your genome in like, the political sense, in terms of, is your vote worth as much?
Ethical Implications of Genetic Understanding
If you have this genome, then it necessarily means you have this IQ. And if this IQ or, you know, we just don't know where it's going to go. So I think that's also my sort of semi dystopian outlook to this, is it's not so much what the genome can tell you right now, or what we understand of the genome. AI is getting increasingly better at understanding the nuances with the genome. And what if it gets to a point where it starts discrete and against it, or giving people an opportunity to understand and discriminate against it, which is obviously a lot more likely. I'm thinking of, like, the Cambridge analyst situation.
Navigating the Environment of Genomic Data
But like, on a genome level, I think that's where my head goes, which I know I'm the least educated person on this stage by some degree, quite a few degrees. I imagine these guys are doctorates, masters of the whole thing. But anyway, pun, outside of that's just where my head goes. Maybe get over some of them already. I like looking at the pfps on stage. I definitely don't hold that crown either, Jenkin. So, yeah, I'm just on a lose streak here, but let's maybe get the mic over to Boltzmann and get your take on this.
Optionality and Privacy in Genetics
Am I completely out to lunch to the advancements and the dystopia that I can see with genome side? And are there any benefits to AI understanding the genome, which I know there's numerous, but would love for you to get into some detail around that, too. Yeah, I think that the most important thing is creating optionality, creating potential. You know, basically, if someone wants to be in such a system, they can do that. And if they want to have a little more ownership and sovereignty over their data and their activity.
Data Security in a Digital Age
Because even going outside of, like, genomics, like, we think about all of these like, phishing scams that just occur constantly, the financial impact, the bad stuff that's going on, and how that interacts with a lot of our legacy systems, and if it's already possible to like, basically go online and like, harvest your Social Security number and a bunch of other personal information, if you're uploading all of your personal information to a service that is not secure, that is not protecting all of your data to the extent that you want that protected, then it can be very easily, like, hacked or breached with the massive amounts of information that were providing to a lot of large centralized providers, and then the type of scams and disruptions that we see now.
Futuristic Concerns and Privacy
Imagine that, like 100,000 x. I mean, unfortunately, that's kind of what we might be facing in the future. And that's why with Boltzmann, not only does it enhance the scalability, so the actual usability and experience of something that you're using, and something like a chat GPT style platform, but also, by default, providing you with a higher level of privacy, and then providing you with a lot of options. If you want to have your requests run through Tor style onion routing, we provide you the ability to be able to do that.
Introducing Razeeb and the Discussion on Genomics
I think maybe Razeeb might be probably the best person on the stage and probably one of the best persons on Twitter to ask if he's available. We've got Razeeb in here. So we definitely need to do sort of a couple of minutes on genomics and privacy and extrapolation from results. Definitely, yeah. Yeah. So let's break it down, like, in terms of the present and the future.
Personal Genome Sharing and Risks
So as an individual, many of you know, I actually have my whole genome public. You can just google it so you can look at it. So I'm not, like, super stressed about it personally, but some people, they do worry about it because they know that they have genetic diseases in the family. So, for example, like, I don't know if you have, like, an uncle with autism, your probability of having autism, you know, or you having a child with autism is like eight x or something like that. Right. Eight times higher. Right. So if you have children, you might not necessarily want people to know what their risks are when they're young. That might have some consequences, possibly employment discrimination.
Worries About Genetic Information Disclosure
These are things people worry about. So I would estimate that about 10% of people actually have something that they probably wouldn't want to get out there. Okay, so that's one thing. Or, for example, if you were a womanizer and you have illegitimate children, although with 23 me, that doesn't really matter. But anyway, there are people that have things to hide. Asymmetric information is one reason that long term healthcare insurance in the United States actually collapsed, because the Genetic non Discrimination act made it so that insurance could not demand that genetic information. So what was happening was people with family histories or suspicions would get tested.
Healthcare Insurance and Asymmetrical Information
They would know exactly what they, you know, what was going to happen to them in terms of, you know, they're going to get Parkinson's when they're 55, they would buy long term healthcare insurance and, you know, obviously use it, whereas other people that were worried because of family history, they would get the test and showed that they wouldn't get the disease, they wouldn't buy the insurance. And so these asymmetric informations crop up. You know, people obviously want to know more about themselves, and they want other people to know when they have those situations. The other thing is future.
The Role of Genomes in Research and Future Discoveries
One reason I do have my genome out there is researchers have looked at it and given me tips about various things that they've seen in there. The genome is an appreciating asset in terms of, we know more and more as we get more genomes out there. They've been probably on the order of 1 million, somewhat more than a million whole genomes sequenced in the world right now. Probably around a million. Obviously, there's 8 billion people, there's hundreds of millions of Americans, et cetera. We're still at a stage where there's things like rare variants, etcetera, that we're going to discover a lot about, that we don't know about.
Implications of Emerging Genetic Knowledge
That's good and it's bad. It's bad insofar as well you might not want to know. Have people know eight years from now when there's a new gene with a high risk susceptibility that's discovered, and it turns out you have that variant. So all of a sudden you've gone from the 90% of people that's like, oh, whatever, clone me. You can look at my genome too. You know, I'd rather not. You look at it. Right. James Watson, fairly fair, famously, he was one of the earliest people sequenced, Jim Watson, the cricket Watson guy.
The Need for Larger Data Sets in Research
And, he didn't want any results returned from APoe, I believe, which is the Alzheimer's, locus. he does have that history, and he didn't want to know. Okay, so that's one thing in terms of, you know, upside, I would say the more we get people into the system, the more information we can get, the training sets bigger. I'm, I mean, this is just like simple math, right? Right now we have bottlenecks in terms of data storage, management, all these other things. You know, biologists are not trained to be big data people.
Challenges in Genomic Data Management
So there's like a skill issue, I think, that's going on here. But aside from the skill issue, there's also just like infrastructure issue. You know, people are trying to repurpose frameworks that come from other fields in data into biology, and it's not like a perfect match, you know, so, you know, we're going to be able to give more insights, more returns. 23 andme, I'm not going to talk about like the whole thing. I know some, I know the co founder of 23, me, she's a friend of mine, not Anne, but Linda Avey.
Recollections About 23andMe and Market Dynamics
Ann was the third co founder. There were two earlier co founders. They were both pushed out by 2009, okay, or left. So anyway, you know, a lot of what's happened with 23 andme is partly deserved. Zero interest rate, period. A lot of business decisions that weren't advised. And, you know, I would say, like, frankly, a little bit of egotism there. That's what's happened to some of these biotech companies in the genomic space where they've expected zero interest rate to go on forever.
The Future of Genomic Businesses
And so they were throwing around cheap money. They actually do have businesses, they actually do have products quite often. It's just that the addressable market is that big. Yet the addressable market will get bigger as more people get sequenced and more results come in. Right. For those who aren't aware, you were, well, the first person, and by extension your childhood, one of your children was the first child to be genotyped in utero, right? And then he's got whole genome sequence.
Pioneering Genetic Sequencing
He's the first human being who has whole genome sequenced in utero who was born. So you can just like Google. Razeeb Khan Sun I think the media results will come in. So that was 2014, right? So, you know, I am a enthusiast. I'm very big into sequencing and stuff like that. I'm actually frustrated where we are and where we are in terms of, you know, not leveraging that information. Partly it just has to do with fear.
Addressing Fears in Data Usage
And something like decentralized AI with high security would actually allay that fear. Obviously, distributed computing and AI, that's what you need to scale. You need to distributed computing to just scale in terms of data storage management, but you need the AI to actually scale in terms of insight generation and discovery. As the training sets get bigger, as the data that goes in gets more informative, we will be able to pick out signal out of the noise.
Potential Applications of Genomic Insights
So what can we do with it? I mean, obviously prediction, diagnostic, that's one thing. It's like pretty straightforward. I don't know how much juice that's going to add for a lot of people. It is going to add a lot of juice for people who have, say, family history, schizophrenia and autism. I'm just going to put that out there. That's not like 1% of people. That's more than 1% of people. Right? So it's already non trivial there.
Leveraging Genetic Data for Drug Discovery
But I think another thing that 23 me was focused on and a lot of the other companies were focused on is how we can use genetics and artificial intelligence on this big data to more intentionally narrow the search space for pharmaceuticals, right? So as I'm sure everybody knows, pharmaceutical, the drug discovery takes a long time. Part of the long time is due to various regulatory issues.
Understanding the Drug Discovery Process
But aside from the regulatory issues, you know, it is. I don't know how to say it. A lot of it's trial and error. It's kind of unintelligent. It's just kind of iterating through the search space. And so the whole idea is how we narrow the search space down. Right. And genomics can help narrow the search space down. This just recently, and it's really interesting how the, how people are working on this.
Conclusion and Future Directions
I think were talking about Athena Dow and we had a couple of others that were kind of jumping in and discussing the deci projects that they've got. And I know it's. Oh, I just saw the time and we're nearly at the time limit tonight, so can I bring each of the guests up to quickly let us know what they have in the pipeline coming up? I think Din is still missing in action. Otherwise I can.
Exciting Developments and Guest Contributions
I had, you have 1 minute to tell us what you have coming in the next few weeks that we should get excited about. We are working on like launching a new product which will basically make it easier for people to build AI based applications and launch them. That's sort of what we are working on right now. Once again, more focused on entertainment and creativity, but then we'll branch out into other things as well.
Investment in AI Applications and Community Engagement
So excited about that because I think that will enable a lot of people to learn to, you know, build custom apps, sort of like WordPress for AI apps. Oh, yeah. Yeah. Well, I know Kip has been collaborating with various aspects of this, so. Yeah, very excited to see where that goes. And Boltzmann let me know what you have to get me excited tonight.
Innovations in Language Model Processing
Yeah, so we just proved with the internal demo that were able to split large language models layer wise. So we have it working with reference models and we're extending it to a bunch of different models now. So we should be able to show the community like the distributed nodes processing separately and then recombining the results and then doing a general inference task. So this will be really exciting.
Plans for Open Source Demonstrations
And then we'll open source, sort of like proof of concept and demo, and then we'll move on to sort of integrating the commerce layer, which is like actually paying the distributed nodes for their compute time. So, yeah, in next few weeks, it's going to be busy couple weeks for us. Oh, very nice. Yeah. And I guess last of all, I will jump to Julian.
OCU Launch and Community Initiatives
Let's know what Kip is up to. Very intense next couple of weeks. In terms of products, we are launching OCU imminently likely to be the largest app on the new chain and probably one of the largest decentralized AI plus decentralized education apps ever, which is very easy when we are the first. Okay, so we will be the biggest because we'll be the only one.
Commitment to Education and Community Building
But we are getting very excited about that. There's a lot of, we've got ten in the first batch of professors who have onboarded their materials and their courses and the students. So that's, I think, setting a good trend for what to continue. For OCU, we have products. I think Hart mentioned a little bit on what he's working on.
Upcoming Community Engagement Activities
We are involved with that as well. Hopefully we'll get to launch it together and announce it together. Community wise, we have a lot of community activation campaigns and initiatives, which actually, a lot of it started today. We have one that launched today with gate. We have one more coming up with another exchange and airdrops.
Anticipation for Airdrops and Engagement
I hear from Jack that there were lot of questions on when airdrop soon. So that's why we are doing all of these campaigns. So watch out for it. Okay. A lot of things to. If there was a time to keep track of Kip, the time kind of starts now. Now is a good time. I agree.
Closing Remarks
So, yeah, I think that about wraps it up for this evening. I just want to say thank you. Thank you to everyone for joining and for being here and especially to our guests. This was a great space for me since obviously the AI side is where I came from originally. And I love just sitting on a space and chatting about it with some very intelligent contributors.
Final Acknowledgments
So, yeah, thank you very much, everyone.
