Democratizing Artificial intelligence by Democratizing Data Access with Dan Gailey, CEO of Synapse
Synapse CEO Dan Gailey is interviewed by Gabe Colors for the podcast Coloring Crypto. Decentralized and Democratized Artificial Intelligence, what it means, how you can participate, and what the impact on the world will be.
Gabe: If you hear the word “crypto” do you generally think it means currency?
Dan: It depends on context. Right? I grew up in the hacking phreaking scene. Traditionally, crypto meant cryptography.
Dan: Synapse is basically the division democratizing artificial intelligence by democratizing data access. You can get an allowance from talking to a computer. The binding aspects of both data and machine running models through a decentralized distributive service like if they use blockchain.
Gabe: The blockchain, not just crypto currency. By crypto, we mean cryptography, all of this, including Bitcoin and other crypto currencies, like Ethereum with contracts and all this stuff, but also the whole greater picture of the concept of the block chain.
Alright, if I’m getting over your head, then you’re in trouble because the next podcast with Dan Gailey, the CEO of Synapse.ai, it just goes way over my head.
Dan: Yeah, sure, so … I’ll start with kind of what we’re doing and why it matters. So, Synapse is basically the division democratizing artificial intelligence by democratizing data access. And then providing access to both data and machine-learning models through a decentralized and distributive service like Ethereum and blockchains.
So, we’re basically democratizing artificial intelligence by democratizing data. And giving … I could explain that right there if you want before I move on.
Gabe: Yeah, let’s do that. Let’s unpack that.
Dan: So, I think philosophically for me, I grew up a hacker and phone phreaker and I believe that information should be free and people should be compensated for how they participate, and ultimately right now, what we have are people who are talking about artificial intelligence and what it means to humanity and what it means to humanity’s future. And they’re also talking about democratizing artificial intelligence. But ultimately what they’re talking about is democratizing access to it, and not necessarily making it open, available, and fair to the participants contributing that data and contributing to training those models.
So, in order to create models that represent people, that represent humanity as a whole, you have to kind of open the opportunity for contributing data up to everyone. So, it can’t, in essence, be just owned and monopolized by a business who has a particular directive and audience that they care to serve.
Gabe: So, this is where, like … We would understand the big companies like Facebook and Google, right now, are collecting all this information and they’re running that as massive corporations and you’re saying, “Let’s democratize that.”
Dan: Right, so that’s the big first step, right? Right now, if you look at Jaron Lanier, or Lanier, it depends on how you want to it to sound, with Who Owns the Future? Which basically describes this local maxim that we’re in where you have what he calls Siren Servers, and it’s basically companies like Facebook and Google and Amazon that monopolize this data and use that data in this walled garden in order to train their own models to use it in their products, right?
So they’re basically supplementing the costs of their service to acquire your data and then using that data to create recommendation engines, advertisements, anything on their own platform. And what happens after they train that data, or they train their models, and use that data, is that data kind of goes away, right? No one else can use it, and it certainly is not leaving Facebook.
Gabe: And, the question on that even … I mean, this is interesting stuff … I’ve always wondered … It’s my behavior Facebook is tracking, but I don’t even have, seemingly, the rights to see what they’re recording of my behavior. Do you know, is that true?
Dan: Right. So, you don’t have an opportunity to see, kind of, exactly what they’re tracking and how they’re using that. And that’s kind of also something that we want to surface … By building on top of the Ethereum and blockchains, we’re basically creating providence, or what’s called data lineage. And when somebody contributes data, or your data’s purchased or used in a model, or however that works, you can go back and you can actually track exactly the graph of who has access to what and how they’re using it inside the network.
Gabe: I get the concept, and I think, who wouldn’t be behind the concept? Like, I think we’re all a little frightened of how much these corporations know about us and how much they’re … What they’re doing with the data … I mean, I understand what artificial intelligence is doing, where it’s coming up with insights that we might not already see, but you’re attacking not just, it sounds like … So, getting digital behavior on the blockchain, but then also attaching the artificial intelligence piece to it. Is that right?
Dan: Correct.
Gabe: What are the challenges, in terms of building this?
Dan: So, there’s a couple of different challenges. So, the first part is the marketplace. And the marketplace basically allows anyone to participate as a data provider, data seller, data buyer, as well as a machine learning model seller and buyer. So we’re basically connecting these two services in our marketplace, and it’s the first time anything like this has ever been done. So, if you look at machine learning models, you know that they require data to be trained. So, connecting this on the marketplace, we’re kind of delegating responsibility to the original owners right now. That’s kind of how we’re kick starting this all.
Gabe: And who would be the original owners?
Dan: Whoever collected the data. So right now we are partnering with … And we’ll announce these partnerships … We’re partnering with people that sell financial data, that sell sports data, that sell … Really just-
Gabe: Okay, so it’s pre-existing data. It’s data that already exists now?
Dan: Right.
Gabe: Okay.
Dan: So data marketplaces exist, right? They just don’t exist in the one place, or in a decentralized way. There’s no marketplace for data right now and there’s no decentralized marketplace for data.
Gabe: And what’s their incentive to hand over that data to you? Are you buying it from them? Or what’s in it for them in the partnership?
Dan: So the partnership, basically, gets them more distribution. They’re really excited to see how their data can be leveraged through this way, which is exciting to them because they can use it as a case study. It basically helps promote their business, as well as give them an opportunity to participate in something that’s kind of really exciting.
Gabe: Let’s kind of back out of this, and give me a sense of your background and how you initially got into the whole idea of cryptocurrency, the blockchain, and then came up with Synapse.
Dan: Yeah, sure, that’s an awesome story. So, we’ll start off with, kind of, who I am and philosophically what this means. So, my background is I grew up … My father was an attorney, and my mother was a nurse. He had his own law firm, and they invested in personal computers before a lot of people did, and I got to hang around, and the eventually we got the internet. I asked somebody how to hack, and they taught me how to code. So I was programming at around ten years old.
Gabe: What code were you programming when you were ten years old?
Dan: C.
Gabe: You learned C at ten years old? Awesome.
Dan: Yeah.
Gabe: Okay.
Dan: So, there was BASIC and C, but I liked the syntax of C a little more. From there, I grew up in the hacking and phreaking scene, where they have their own ideologies about information and how it can be accessed. Obviously, there’s a kind of sci-fi and things like Snow Crash and Diamond Age and William Gibson and just all these really great ideas that are finally, perhaps, going to be realized with technology.
So, I grew up there. My background is, I didn’t immediately go into college. I worked in the ISP industry, telecom industry. My first job out of high school was research and development for a telecom company, basically developing the precursor to what cloud computing is today.
Gabe: Were you coding for them, or doing something else?
Dan: Yeah, coding and Linux and system administration and basically helping set up their entire network. So, full stack stuff. So, from there, worked at an ISP, did the network security, ended up in network security at an oil and gas company. This whole time, coding is just a part of who I am. It’s kind of just a tool I grew up with. And decided to go back to school for electrical engineering and computational chemistry.
That’s where I kind of got introduced to … I sort of was introduced to machine learning and bio-informatics at a genetics company that I consulted for. But I really didn’t dig into the models, themselves, until I did computational chemistry and started thinking about how to apply those because my advisor in electrical engineering was researching machine learning and autonomous robotics for deep space exploration.
Gabe: Just to understand, how is computational chemistry different than chemistry? It’s just all the mass part of chemistry?
Dan: No, so what’s different between computational and regular chemistry is, regular chemistry is what they call wet chemistry, where you have theoretical models, but ultimately you go into the lab to test them. So, if you’re running all your experiments, you kind of need a lot of grad students, right, to run those experiments for you, and what computational chemistry does, is say, “Hey, we have pretty good mathematical models of how these molecules exist in space,” so you can run virtual experiments and get a prediction of some outcomes. So you can run hundreds of thousands of experiments simultaneously and then go into the lab and test what might be some likely outcomes to confirm the results from the computational model.
Gabe: Okay, so what time period is this? Is this, like, early 2000s?
Dan: No, well, there’s this whole part of my life where I started a punk rock band and hosted a poetry night.
Gabe: In the Bay Area? Here in San Francisco?
Dan: No. It was in Texas.
Gabe: Oh, cool.
Dan: Yeah, and so I guess from 2005 to 2010 is when I was in university.
Gabe: And what happened in 2010 after school?
Dan: Oh, I just moved out to the Bay Area. So, I grew up in the hacking and phreaking scene, and I had a hacking group myself and all of my friends lived out in the Bay Area, and I’ve always known that this is kind of where I need to be. In a very circuitous route, I managed to end up here. So, in 2010, I just decided, “You know what? I’m just gonna pack up everything in my car and move out here.” I didn’t have a job, I didn’t have friends. I didn’t even have a place to stay. I had, I think, $4000 in my account from selling my car after I moved out here.
Gabe: No way.
Dan: Yeah, I was-
Gabe: Where did you stay the first night? You must have had some place to go
Dan: Yeah, so I had a little bit of cash, and I think I stayed in a … Like, a Craigslist ad, right, out in Outer Sunset or something.
Gabe: So, this is a couple years after Bitcoin had launched. Were you already interested and involved in the whole blockchain cryptocurrency space?
Dan: No, this is … So 2010 was probably when the coin just launched and no one really knew about it. I didn’t hear about Bitcoin until 2012, maybe. It was a friend of mine … So I ran the San Francisco Hacker News meetup here in San Francisco … So just surrounded by awesome nerds and Bitcoin came up. As we already do, we sit around, drink beer, and talk about the impact that this might have on society, on culture, on the economics, of a nation. It was a pretty big, new thing.
So, I started getting kind of involved in that scene. I was also doing that full time, and someone reached out to me that wanted me to be an EIR at their venture capital firm. And it just so happens that they were just now starting to look into cryptocurrency. So, I worked at E.ventures, which is a pretty big firm here in San Francisco, and they invest in everything from small startups to big multi-million dollar rounds.
So, my job was more merchant tech, so looking at the crypto-space. I got to go out to the … I think this was 2012 or 2013 … The cryptocoin conference. So, prior to this, I had met up with Brian Armstrong, who now is the CEO of
Coinbase, and we had met because he had posted on Hacker News that he was looking for somebody to build a cryptocurrency exchange. I had been thinking about it, so I got my friend, who was a journalist at the time, to come out there and just be there because it might be a really crazy moment if we decide to hook up and build this exchange.
I decided that I wanted to build something with a friend, and the friend kind of puttered out. I remember this to this day. They were like, “Bitcoin won’t even be around next year.”
Gabe: So, wait, so you turned down the opportunity to be a part of building what later became Coinbase?
Dan: Yeah.
Gabe: What were your thoughts, man? If you look back at that time, can you recall what the sort of conventional wisdom, or your own thoughts, about the space was?
Dan: I thought it was an open playground for nerds. Like, that’s … And it had the unlimited potential to impact entire industries. So, at the time, it just felt really heavy, but at the same time, you also, kind of, were considered crazy by a lot of people. You could go out and you could evangelize about the new cool thing that’s called blockchain and Bitcoin and it’s new currency and people would just be like; No one cared at the time, or they didn’t get it.
So, I’m working and I’m trying to figure out … I’m at the venture capital firm. I ended up leaving there and looking to do my next thing, and that’s when I came across this idea. Because we have this immutable chain that basically is a record of everything that happens on it, it would be really cool and something that hasn’t existed and probably doesn’t exist to this day, is an open point of interest market where you have all these, kind of, agents that are communicating to one another, exchanging services, like negotiating for services …
Here’s what happened. I threw a hack-a-thon called Hackendo, and the theme of it was integrate. So, we had all these IOT services and all these sensors that were coming out like the Fitbit and all these things that would basically transcribe physical, real-world sensor data into some digital form.
Gabe: Dude, I think I’m not so much a geek, probably, but not a hacker to the degree, which you are. Why’d you call your hack-a-thon Hackendo?
Dan: So, I started a site called Techendo, actually, that was tech news-
Gabe: What does that mean? Is this funny-
Dan: So, “endo” means insider, but it’s prefix. But, it sounded better as a post-fix, right, to tech. So, Techendo was born, and then I was like, “You know what? This is us. We’re gonna do Hackendo.” And we ended up throwing a hack-a-thon. It was just me. I was running the whole thing. It was pretty intense. We had it over at Tech Shop, but I managed to get all these companies that had all the sensor data out, and I wanted to-
Gabe: What do you mean sensor data?
Dan: They had all these devices that would transcribe real-world signal into digital signal, so you know-
Gabe: What’s an example?
Dan: Fitbit, your things that would plug into your car and basically take your mileage, your geolocation stuff, your Fitbit. We had EEG sensors out. Anything that would measure something in the real world and basically create some kind of signal, some kind of data that you could analyze.
Gabe: Was your intention to attract that type of … Companies that had sensor data? Or it just happened that they showed up?
Dan: 100%.
Gabe: Okay.
Dan: So I was sitting there, and you kind of imagine this connected world coming back from networking, from understanding how computers work. So, you can imagine just taking those nodes, those PCs, that are sitting behind closed doors and start wearing them and bringing them around and then start with the knowledge that they’re gonna start tracking things. How can it make your life cooler and better? And at this point, you had all these people cropping up that were very domain-specific. And they all kind of existed in these kind of walled gardens with the APIs that you can get access to and play with your own data.
And I said, “Cool. Now that’s our first tier. Now we have these devices that have access to the data. What’s the second tier?” And that’s integrating these things. If the past was me and my devices, the future is going to be me and my devices, you and your devices, and how they work together. And at the time, that’s a radical idea that I think a lot of people only just now in hindsight are really starting to get.
Gabe: Let me hear you say … I’m not sure I’ve got it exactly, but I’m kind of hearing … You’re working on producing a blockchain that includes all of this personal and digital behavior data and the immediate concern is, “Why would I support that? Whether it’s Facebook or Google or anybody, even if it’s open source why would I want more and more of my private information recorded?”
Dan: So, it’s up to you. We don’t actually use the blockchain to record public data. What we use the blockchain for is access control and data lineage, providence. So you don’t actually see … Like, if you were to acquire like venture did from sensor data, from your ISP device inside your home, you won’t actually see that being transacted on the blockchain.
What happens is the contract is created on the blockchain and because it’s decentralized, meaning every note observed what happens on the chain, tt can be a fulfillment contract, meaning, “I need this type of data to be sent to this location or this service or this server or this port,” and the people listening on the network, the agents, everyone, I call them the nodes, can decide for themselves whether or not they want to submit that data.
Gabe: Is that an actual decision? Is that an opt-in type of choice? Or is that something just happening?
Dan: Correct. It’s opt-in, but also, I think, could go to supplementing your experience with devices in the future. Like, one day you may get a free cell phone because you’re participating by default on this network.
Gabe: Right. I also have an iPhone and I just recently bought a … It was Boost Mobile, and I was like, “Why is this phone so cheap?” It was full of ads, like, right off the bat. So, that’s obviously why it was so cheap. They were selling advertisements and they’ve got a lot of information on me. And you’re saying, so this could be used as a way that by allowing some of my personal information to be shared on the blockchain in this model, it might possibly offer me various discounted things?
Gabe: So, what is the business model you foresee with this? How does that work?
Dan: So, we actually take a percentage of the fees transacted, or transactions that happen. So, we just charge fees. So, whenever somebody buys the cell data, we take a small nominal amount. But ultimately it’s about creating a circular economy, and I think that’s kind of how the network will perpetuate itself. Ultimately, it’s about getting access to data, training a model, someone using that model as a service, and then because there’s providence, you can create royalties for the people that contributed their data to train those models.
That’s, I think, the coolest emergent feature of this network, is it’s literally the future of work, right? Like, if automation is coming, and you are training models to recommend things and create intelligence and insight to products that companies use, and are essentially the company’s IP, if we can do it in an open way that maps your participation … That if you start training a machine today …Let’s say you use an application that allows you to start training voice models so that some voice agents can understand what people are saying, that’s all captured on the blockchain now. So, in 100 years when that model’s still in use, you’re still getting royalties.
Gabe: Let’s talk to our listeners here. What would be a consumer’s experience that would be impacted by this?
Dan: I totally 100% believe that we’ll be based into every operating system. So, we have a developer’s fund that we’re launching with our token, where we take 33% of the value of our token and we give it to developers working on cool stuff that’s either leveraging our technology or complimentary to it. One of those big ideas is a blockchain OS. And so-
Gabe: An operating system for the blockchain?
Dan: Right, and this is brand new and it’s just started, but it’s kinda been something on our minds for a while. You see all of these services that are distributed like ITFS for distributive compute, where you have all these decentralized participants distributing both file store and compute, and basically what you’re talking about is redefining the stack that has existed locally, but now all people are participating on it, on your mobile device or your computer at home.
So, what I think is going to happen is you’re gonna have an operating system where you have all of these decentralized services based in, part of the stack, and right next to your battery meter on your phone, you’re gonna have how much money you’re making by just being alive per hour. Like, the data that you contribute, the resources that you contribute, how you participate on the network. So, right next to your battery meter, it’s gonna be, like $50/hour.
Gabe: What token would that be?
Dan: So, I think for the data part, it’s going to be [inaudible 00:26:01]. But cumulatively, you’ll have a wallet that accommodates all tokens simultaneously because locally, you’ll have your own off-chain transactions that happen locally. So, you’ll have to switch from one token to another in order to go back and forth locally between different services. I think the tokens are gonna be exchanged up and down that stack.
Gabe: So, I kind of heard you say … Let’s say, I own in my wallet, 20 different tokens. It doesn’t even really matter which token is being activated. There’s just some intelligence that’s just putting up whatever token is most relevant to my own combined token that is what’s active out there.
There’s two questions here. How do you separate your token from this mass noise of new tokens? But also, this idea we just talked about where, at some point, it could be internally managed and it’s just sort of like I wave my phone at a transaction and it just picks the most appropriate token. It’s invisible to me.
How does Synapse Ai fit in, play a role, in that? If I’m hearing you correctly, it sounds like you’re taking on this hugely ambitious overall operating system take on this. Am I getting it?
Dan: Yeah, so we would basically … Where Synapse would … We would be like a service, a part of that stack where there is distributed file store, there’s distributed compute, there’s your data service, and your intelligence layer. We’re talking about the future. It’s great, it’s exciting, and it’s a lot to take in, but there’s a lot of great people working on it.
Gabe: You had your pre-ICO, and how much money did you guys raise?
Dan: So, we did altogether about $200,000.
Gabe: Who was buying in, in this first round?
Dan: Who we really resonated with were the technical people. People that have been in tech, really understood what was going on. You have to understand the fundamentals to understand what could emerge from them, the possibilities. So, it was either zero or one. It was binary for us. We could either go out … And we would talk to people and they’d be like, “I don’t get it, and I’ll never get it.” And those are the people that, they’ll get it when it’s in their hand.
And then you talk to the people that understand what’s going on, and what’s coming up next, and they 100% get it. And they jump into the conversation and then they jive, and they’re just like, “Whoa, this has unlimited potential,” and those have been the people, historically, that have bought in early to crypto.
Gabe: If you hear the word “crypto,” do you generally think it means currency?
Dan: So, it depends on the context. I grew up in the hacking/phreaking scene. Traditionally crypto meant cryptography, but now when I’m talking to somebody who isn’t a die-hard hacker, when it say “crypto,” it means cryptocurrency. So, things change, but hackers do complain … And I’ve been a part of this conversation many times. Hackers do complain that they hate that cryptocurrency has taken over the crypto word.
Gabe: Okay, if you could give me a final, sort of, sentence that my five year old niece could understand, what you’re creating and how to jump in.
Dan: To your five year old niece … Okay …
Gabe: Or twelve. Twelve years old.
Dan: Okay, I’d say you could get an allowance from talking to a computer. That would be it. That would be like the biggest high-level to a five year old that I could get.
Gabe: You can get an allowance, you can get paid, by being on your computer?
Dan: By talking to your computer. I think that’s the coolest Diamond Age-esque reference I can give.
Gabe: Awesome. Okay, and for folks that want to find out more about you or get involved, share some access points.
Dan: Yeah, sure. They can visit us at synapse.ai, and they can follow us on Twitter at @AiSynapse. And they can follow me on Twitter at @dpg, those are my initials.
Gabe: Oh, I had one question from a listener. How did you get a three character Twitter?
Dan: You know, you just gotta know someone, I guess.
Gabe: Hey man, Dan, thank you so much. I appreciate it. I’m looking forward to looking back on this conversation in a few months when I probably know a little bit more and can put it all into context, but I really feel like you’ve got some exciting stuff going on, and I hope to keep in touch.
Dan: Yeah, cool. Thanks, Gabe.
Gabe: Thanks a lot, Dan. Bye.
Dan: Bye.
Gabe: Coloring Crypto is now a weekly podcast. And are releasing new episodes every Sunday for newbies who are curious and experts who think they understand. Tweet us at @ColoringCrypto and leave us a review on iTunes.
That concludes the Coloring Crypto Podcast interview with Dan Gailey.
