AI and Open Compute - How to Solve the World's Big Problems - TOKEN2049 Singapore 2023

Ai and open compute AI is a Hot Topic right now I'm hot it is it can solve The World's hard problems but it can't do it by itself it needs decentralized open compute so let me explain a little bit about what that means first what are the biggest problems they're the ones that are most complex right climate monetary policy energy disease these are relatively an obvious list so how are we doing in solving these problems as you know a race you know every climate forecast has been wrong massively the U.S federal reserve with thousands of employees and tons of phds can't figure out when to raise rates certainly gets it wrong we have all kinds of data points on health that we can't use genome is sequence but we're not allowed to compare it to phenotypes so we are a lot a lot of terrific Innovation is is latent and sitting there unused um there are all kinds of clean energy options but countries like Germany are increasing coal and the pandemic modeling was wildly off right so there's a lot better we can do in all these problems so and why are they hard they're hard because they take a lot of compute power price matters the location of the data matters and it helps to be on the edge but the biggest problem is relationships are not stable over time they're always evolving and so you need a solution that doesn't analyze data in a static way but keeps evolving and keeps innovating because the problems change you know looking at something can change it certainly true with monetary policy and as we all know diseases evolve as well so AI to the rescue right AI can solve this you know llms can dig into huge and incomplete data sets to pull out inferences that traditional programming cannot uncover and won't uncover and that's that sounds like and could be a terrific solution and for many things it is but AI isn't perfect it's not verifiable right you don't know for sure what data your model was run on and you can't verify it it's certainly subject to censorship I think people here are very aware of the discussion and topics around answers and particular answers it can't be answered the right way and questions and how they're addressed and what date even goes in data refresh you have to constantly be refreshing the data in order for it to be relevant and also AI itself is going to continue to innovate so even if you're refreshing the data you've got to keep changing your model to make sure your model is evolving and is the latest and greatest option for evaluating that data so open compute and compute Solutions have evolved to try to solve some of these problems but and they've been work people have been working on compute solutions for years but none have worked and there's a whole number of reasons why compute hasn't worked in a scaled um uh seamless manner but there have been coin economy flaws clunky user interfaces you know poor architecture decisions and also very little true compute decentralization the exception of that is the Smart contracts obviously smart contracts are truly decentralized but they don't have the scale for um for compute and that's why compute is hard that's why open decentralized compute is very hard because you have to be generalized verifiable and scalable and while smart contracts are verifiable they are not scalable and we can talk as much as we want about layer twos about extra sharding Solutions but generalized compute is hundreds of orders of magnitude more demanding than the most optimistic projections of actual smart contracts so we need a different solution and so you have to build decentralized and verifiable without blockchain and that is difficult and so we go back to decentralization and if everybody remembers I'm going to reference the next top next speaker is from Napster right Napster is the first application that was decentralized at scale way predated blockchain so there's ways to do decentralization pre-bitcoin pre this universe and so what that is I think if you look back we actually can get a better solution to these problems and so fluence which I'll talk about in a moment isn't Bound by these blockchain constraints and can run any type of computation and so how does open decentralized compute work it's got Limitless scale and it's Unstoppable you've got large capacity thanks to crypto incentives it's Enterprise grade you know Enterprise grade data centers aren't going to rely just on coin economy they need Fiat payments they need to know that their expenses are going to be covered with a high degree of certainty so you can't just do a crypto economy and expect to have a durable enduring scaled Enterprise grade system you have to be it's Edge enabled and it's verifiable we have a proof of compute so you can actually validate that the computation has happened and with subnets and fault tolerance you have true decentralization user interface follow the um you know well-known user interfaces and so it's familiar to developers Best in Class and so what does that mean you get open compute decentralized benefits that you get verifiability you can verify that the compute was run on particular data it's also censorship resistant because any host can run applications if a host goes down your application shifts and the data the compute shifts to another host so super super important it's cheap because when you basically can shift from host to host the compute can move from provider to provider there's no pricing power and with no pricing power there are very thin margins and within margins you see the customer saves money and so you end up with very inexpensive compute you know potentially 80 percent cheaper than centralized Cloud versions and it's buildable and this is I think the most kind of interesting part and this is the part where it ties back into AI or applications and data can be hosted and as long as there's demand that data those applications are hosted which gives people and developers the ability to build further on top of that data and they can keep building and know that they cannot be turned off and so that leads to applications that aren't stoppable and so let's think about Innovation for a second every company has a product roadmap of features right but that road map has to be constrained it's constrained by the team by the skills by the budget by priorities by internal politics pick your company pick your Google pick your Facebook they still can't do everything at once all the time they have to prioritize and so what is more powerful than those companies the global developer community and so when you have applications and data that's open and buildable the whole world can be out there working to innovate and build on top of those applications in a basically Limitless product development role and you say that sounds great does it work well it actually works so well that we have proof the biggest companies in the world can't compete Facebook Twitter and Reddit all had open apis but the developer Community globally kept coming up with applications that were so good it was drawing users away from those core core applications and so those companies largest in the world shut off those apis because they wanted to control the data and users themselves and didn't want things sucked away they effectively used the developer Community as a um as a product development testing ground found stuff that worked and then shut them off but what does that mean it meant that that's the level that Innovation got to so they've continued to innovate but we have no idea what else could have been built if those apis had had stayed open and more um and and we've been you know a decade later with what's happened um and so what is open computation enable you know no closed ecosystems you can build on top of applications if you want um that we think is a creativity accelerant you basically get a transparent cloud with vast offerings um and leads to Unstoppable applications where applications can't be turned off can't be censored can keep running and can continually be built on and that's why we think that decentralized serverless compute is the next generation of Internet Computing we all start no we started off with mainframes and desktops 1.0 Kicking and Screaming over 20 years you know moved over to the cloud that has been a resounding resounding success whereas that transition still continues but decentralized serverless compute is certainly the next if not the last stage where applications exist free of centralized companies free of centralized ecosystems free of single points of failure that and that is what fluence enables and so what is fluence is the decentralized serverless compute platform it replicates cloud services and adds decentralization benefits so you can build and scale applications easily with off-chain compute importantly we've replicated the cloud stack so that you can build native peer-to-peer native decentralized applications which we know people use a cloud for a reason it's easy the stack is simple and it is is easy to build that's why it works right so the only way we can compete is if we have built a stack that mimics that but works for decentralized applications back to AI that allows you to run AI models cheaply censorship resistant and with the ability to compose on top of them easily and innovate on top of those models and continually innovate those models there's something like I think hugging phase I just heard has 340 000 models on it and they're adding around 40 000 a month all right and to give you a sense chat GPT cost about 700 000 a day to run so if you want to run a model and you think you're going to have users you're not anywhere near chat GPT a tenth of it that's super expensive you want to run it on centralized Cloud your only shot is to run it in a system where you've got an 80 price savings and you also want to be able to innovate and build on top of existing models so that is the tie-in between open compute which saves you money allows Innovation Nai and together you really have something that's very powerful and so the product that fluence has is this decentralized Amazon Lambda that market the serverless market is 9 billion today growing to 22. I'm sorry to 55 billion and that is an enormous Market um which is what we think um we can grow even even beyond that and so take a step back and imagine what it looks like when you've commoditized the back end Services you've got Unstoppable applications all the effort of development can be spent on user interface on innovation of things that matter not the back end services that people continually have to rebuild the global Community developer Community is constantly looking for how to find better products how to innovate and when you haven't default open ecosystem that Innovation is rewarded and can continually evolve and so we think we enable a constant transparent and Relentless Innovation and that you know we don't solve all the problems immediately but with both Ai and open compute we can do a lot better than we have so what's our mission we want to empower the next wave of Internet Innovation and going right back to the founder of the internet Tim berners-lee you know his goal the goal of the web was to serve Humanity he didn't know exactly what the HTTP protocol would and what hyperlinks would allow but he knew it would be big and he knew and he couldn't imagine it but he knew there were things out there that would be terrific and it's been an incredible Boon to humanity that's the type of thing we think open compute especially with AI can enable so so sign up learn more about fluence website is fluents.network get this QR code join our Twitter and telegram channels and stay in touch we're going for mainnet by the end of this year um or by say a winner and so we're very close and everyone's cheering so thank you appreciate it