Owning Your Own AI

Wade Erickson (00:15):

Welcome all to another episode of Tech Leaders Unplugged. We're getting unplugged today with Reza Rassool CEO of Kwaai, and we're going too be talking about the topic is Owning your own AI. And res is actually a returning guest we had him on last year when he was with Real Networks and was the CTO there. We'll talk a little bit about the transition from Real Networks into this new venture he's got going on. And but yeah, Reza, thank you for joining us today and sharing with us a little bit about AI and, and, and the world you're trying to build over there at Kwaai. And yeah, just tell you about, tell, tell the audience a little bit about yourself and your background and, and then we'll get right into the topic.

Reza Rassool (01:04):

Wade, thanks so much. Thanks to everyone in the, in the LinkedIn audience and the re streamed audience. Thanks. Your, your tech leaders audience. Yes. second time on your show. Thanks so much. I'm Reza Rassool former CTO of Real Networks. And now I am chairman and CEO of an organization called Kwaai, Kwaai is a nonprofit, very different from my previous mode of, of operating. It's a nonprofit with a mission to democratize ai. So I'd love to tell you more about what, how we implement that mission. Thanks so much. We, we, you know, we're, we're exhausted and exhilarated just having come from the scale conference where we programmed their AI track. So they, our Kwaai summit, our personal AI summit was co-located at the Southern California Linux Expo, the scale conference. And it was a great audience. The room was packed out and we had people sort of lining up in the hallway to, to for MTCs to come in, and we had a great list of speakers keynoting the five hours of Summit. So Wade thanks so much for inviting me back to Tech leaders Unplugged.

Wade Erickson (02:39):

Yeah. Yeah. So we'll probably talk a little bit about that show later that you guys were at and you know, we'll put in our notes, the links to those, those videos. 'cause I, like you said, it was like five hours long of speakers and going in all kinds of areas of AI. So definitely a, a great opportunity for people to come across some more content on what you guys are working on, some of the, the tracks of areas that you're trying to resolve and solve for the AI community and, and trying to keep this in the open domain as they say. So, yeah, so tell me a little bit about then why, what you guys are trying to accomplish in the space. And then, you know, we'll talk a little bit about the open source side of things and, and yeah, just go ahead and start sharing the topic.

Reza Rassool (03:36):

Look, I, I retired from Real Networks late summer last year, and I've been, I'm now busier than ever. I could have taken up golfing like many other people, but hey, that's not me. I'm actually busier than ever. And what I did was I formed a movement, a nonprofit organization, and it started, I was sitting right here and I had a camera pointed at me, and I put out a call to action, said, Hey, I feel a bit uneasy about the direction AI is heading. And if, if anyone else has a similar angst about our straight line path to a dystopian future, then come and join me. And I put that message out to my network of friends before I knew we had about a hundred people contacting, and we set up a weekly Friday meeting just to discuss the matter. And I had no clear idea of exactly where this was going. But here's what has emerged out of lots of consultative dialogue collaboration and tuning of a mission. So the broad mission of Kwaai is to democratize AI, but the way we do it is through building a personal AI operating system. So this, this should be a mechanism, a collection of tools, and a framework to allow everyone to own and operate their own AI. So why, why would we do that? Well, look, the alternate is, the alternate is so horrible, Wade. If somebody else owned your personal ai, imagine if vendors owned the the, the way the, the digital twin of Wade or ZA or anyone else, or your, or your employer were to own. They'd call it a co-pilot. They'll say, here's the Wade Co-pilot. It's trained on you the last 10 years of your work output, and don't worry, it's going too augment you. And then the next conversation, you know, that, that comes very, very quickly afterwards. Oh, wait, here, here is an intern at a quarter of your rate. We wish you well, but don't worry, the program will continue and it'll be, you know, the virtual program. So, look, this isn't theoretical. This was actually going down and it was the subject of the Hollywood strikes. And you think, well, I'm not an actor, I'm not a script writer. Why does that, why does that affect me? No, it comes for every single job, every single job where a model can be trained on your historical work output. If you're an attorney, if you are a professor, imagine an English professor. A model can easily be trained on their last 10 years, of course, curriculum material how they graded essays that got returned and so on. And before, you know, those for-profit institutions, universities will have digital professors. Hey, your course is going too continue. Don't worry. We wish you well in your career whatever that might be. But we've hired a a TA to actually run, but it'll still be your course, it'll be your material. So look, this is not theoretical. This is coming. Okay, what do we do about it? Oh, my word. Well, we believe that you should own your own AI. You should own your own co-pilot. If you don't, during, in effect, giving up the pilot seat, that is the straight line path we're going along in the status quo. AI is currently controlled by maybe a half a dozen corporations. I don't think they have my best interests at heart. They don't have your best, best interests at heart. What can we do about it? Join a movement. Don't just moan about it. Don't be passive. Don't just write a blog post and, and whine about it. Say, and throw your hands up and say, what can I do? You can do something. In fact, the audience we're talking to now are in the best position to do something about it, technologists. So that's the message we. It's, you know, it was on a soapbox. That's the message we were giving out at at the scale conference. But look, it wasn't just rabble rousing, it was actually deeply technical. We had Doc Souls, the author of the Intention Economy and many other pivotal textbooks that actually predict this situation. And, and a while back we had Bruce Schneider the leading voice in cryptography and in privacy and in trust. Harvard professor. He was, he was at the conference. He, he dialed in for that. We had Pan Kaia with a deep experience in personal computing. We had Dr. Kai fan who talks about reasoning AI, not the AI that's going on at the moment, which is purely based on data correlation, but AI that is based on causality and reasoning. Okay? I also spoke both as chair of qi talking about the QI vision, but I also talked about my own research work into fluidic neural networks, which could potentially dramatically reduce the compute complexity of AI. Let me talk more about the choir mission and how we implement that. So the broad fuzzy mission of democratizing AI sounds a bit theoretical. We implement that through building a personal AI operating system. The, it's a framework. It's a framework for you to own your own AI. And it exposes also an API to allow third parties to add abilities to it. So it is a, it is a a body of software that runs on a computing platform that is the computer that runs our lives. So that includes a collection of wearables that will include a collection of maybe cloud services. It's not like Linux that runs on a laptop or on a server or, so the notion of what is the computer that runs our lives, it is a distribution of IOT devices and cloud services and so on. So given that, what is the software that runs on top of it to allow you to own and operate your own personal AI, we gave a glimpse of what that might be. We demoed a proof of concept that illustrates our vision for personal ai. It allows you to own your own data rather than have third parties own your data. It allows you to have a conversation with your own data. So we showed a rag implementation of having a conversation with your own data. That's super important and that's very empowering. But we also are, we envisage a machine to machine dialogue where your personal AI agent converses with third party systems, but now you are the first party, you are the one setting the terms and conditions. You are the one granting and revoking entitlements for third parties to access your personal AI. The whole notion of inverting the relationship between vendors and individuals is important. So the notion of B2C, we invert and we use the, the phrase me to be. So it's me. I'm the individual. I know myself better than you know me, this ridiculous $1 trillion industry that, that is spying, eavesdropping, and geolocating individuals and sucking up data into their data silos for the purpose of targeting you with advertising for what they think you might want to buy is based on what critics call a false model of you. It's a false model of you. It's a flawed model of you. It's typically inaccurate and it is out of date. That's why Wade, you still get bombarded with ads for products that you already purchased a month ago. So this ridiculous trillion dollar cycle of data mining and advertising is ripe for disruption. So why don't you set up a for-profit entity to go after that? You know, you anticipate this is going too be disrupted and here's a better way of doing it. Well, I think a nonprofit makes more sense. A nonprofit has the ethos of putting the individual first. And I think it's fundamentally important. If we go back 20 years ago, if we look at an analogous time when there was a monopolistic situation of server operating systems, when server operating systems were controlled by a half a dozen companies, you had Windows NT, you had Sun Solaris, you had maybe IX and a few other Unix binaries that you had to license. Along comes Linux, an open source, nonprofit, volunteer based movement. And now Linux is the dominant operating system and the industry is healthier for it. It's not that it's a big giveaway. Profitable billion dollar companies were built on top of it. Red Hat being the first billion dollar company built on top of Linux. And so we liken ourselves to the Linux of ai. It has to be a volunteer nonprofit movement based on open source. And so we, we see that there is a template for doing this already, and we'll end up with a better industry and hey, rant on Ran off.

Wade Erickson (15:06):

You know, I as you're talking about that, I think about what's going on now with this TikTok ban and all the people that are saying, Hey, my livelihood is lives in this TikTok platform. I make hundreds of thousands if not millions of dollars through my presence and my brand and all of these things. And now with yanking of TikTok, potentially all these people's livelihoods shift. Yeah. And so if you then flip that as well as somebody that works and they're professional and they work for a company and they build a technical brand on LinkedIn or whatever you know, people reflect on, they write white papers, they share in the community, but they change jobs. Oftentimes in those cases, they get to take their collateral with them because they've generated outside of the employer. But here, here you got a platform that is changing livelihoods because it's no longer viable. They don't own that asset anymore. Maybe if they got local copies of the videos, but they've got to rebuild that brand somewhere else. And that's not the same. I mean, the TikTok is the brand and you know, the platform. And so what you're doing here is you're saying, no, I'm taking back my assets, my knowledge, my intellect, and I'm not going too just give them away for the world to, you know, these other technology companies to mine and turn it out to generative. I, you know, maybe I license access to this if I'm some specialist in a certain topic area that is now the new influencer is I say, well, for a price you can tap some of my knowledge, but I'm going too throttle it.

Reza Rassool (16:45):

Right.

Wade Erickson (16:46):

So this is really the taking back of your brand and taking back control of that brand, which people are being affected by in other areas that are, isn't even ai really, it's just your intellectual creative right now,

Reza Rassool (16:59):

You know, and Wade, this is so critically important and lots of people don't get this. They sleep walking through this pivotal time where their value is being sucked out. It's, it's a value extraction. They call it mining, data mining. No, it's, it's mining of your persona and your intellectual property. Yep. And so, you know, the big AI vendors, well, we didn't steal it. We just gained inspiration from it. It's, it is, it is misappropriation.

Wade Erickson (17:34):

Without it, they would've not have had a product.

Reza Rassool (17:36):

Exactly. So I'm, I'm a little bit conflicted because, you know, when I was at Real Networks, I ran for a time the China business, and I would take that trip to China. China, I think over a period of time, my many dozens of trips to China I'm, I'm a bit conflicted about that GG given that, you know, the, the, the, I know the Chinese government, you, any of those platforms has to abide by the, the, the laws there where there has to be a backdoor for them to for, for the government to be able to access the data. I don't know if that's happening in real time, but Yeah, I don't know. But look, the argument about by dance needs to sell TikTok to one of the US companies. What, like Facebook? Who like, like Twitter. I don't know. I that, that I think that's, you know, an argument between two oligarchs, really, frankly, it doesn't matter to us. Yeah. You'll be just as disempowered yeah. With that, your, your intellectual property will still be held hostage and it's sucked up into a platform that you can't control and you'd have to pay, you know, the toll to go and visit it. It's not that doesn't encapsulate ethos of Kwaai says your intellectual property is yours. And we want to invert the, the, the pyramid. Your, your intellectual property is, becomes your tools of the trade that you can take from one job to another. It does not belong to your employer or to vendors. And at the end of the job, let's say it was your skills wrapped up into a personal AI agent that's yours. Yes, you might sharpen those skills on a particular job, but they're still yours. They don't belong to your employer in perpetuity. They don't belong to a, a, a social media platform in perpetuity. But that's kind of the claim that they make on whatever you put up there. So, okay. Rand on Randolph, you, yeah.

Wade Erickson (19:57):

So, you know, why don't we talk a little bit about the organization and, you know, sometimes I think about, you know, with open source, there's some and, and volunteer organizations. There's a little difference on how you take in features and decide what to kind of develop. So can you kind of talk to me a little bit about the process of prioritizing, you know, what systems and components to build, how, how does that, you know, factor into the decision making of outside influences coming in with requests from the user, consumer base, the own, you know, it's a volunteer organization, so there's the programmers that have their own ideas of what features should be in there and how do you wrestle with all of these different, 'cause eventually you got to build the binary and push it out, right? So how do you do all that?

Reza Rassool (20:48):

Exactly. Well, look, Wade, I wish I had the answers to you're asking, you're asking exactly the questions I we, we wrestle with in the organization. Yeah. It's a volunteer community and it's so different from the way in which I exerted my authority as a CTO, where I'd have a large organization that reported to me. I'd have a budget of tens of millions that I would be in control of Here. I've got zero budget. I've got a volunteer workforce that is also, and they're not full-time. So one is you have to structure the work with maybe a bit more formalism, a lot more formalism than you would otherwise because you've got people dipping in, contributing for a while, maybe very enthusiastically, and then getting busy in their day jobs and, and disappearing. I was, I heard lines talk about Linux, about the movement there. He says he's got 10,000 code contributors, 5,000 of those. So that's half of them might contribute one line of code, and then you never hear from them again. And so this is the this is the one of the big problems of a volunteer organization, the, the prioritizing of features. So we, we need to strengthen and it's important to have product owners that are a bit more long-term. So we we're trying to figure out this balance. There, there will be, as in as with Linux, there is a, a, a full-time back office organization, but it's not just back office. It's, it's basically a full-time operational team. We're structuring that at the moment. You know, we, our organization's less than a year old, we, we formed it July last year. We incorporated at the end of the year. We became a nonprofit in January. And, you know, we formed a board, a temporary board I guess. And then we, we are moving on. So it's early days, but we are learning these mistakes. We hopefully we learn from other people's mistakes. So one of the big learnings from the one of the big learnings from the scale conference was speaking to the rest of the open source community and the community, not just the open source movements, but also the, the for-profit companies that are built on top of them. So we spoke to Red Hat, we spoke to Fedora, we spoke to community managers that have managed communities of thousands of volunteers, and it turns out they are tools for you to manage your, your developer community. So look, we're getting, we have to get smarter than we are now. We'd welcome any help in that area. This is, this is the thing, rather than having more coders. And we've got several, several developers engaged, but getting that senior level guiding team is, is vitally important. Okay.

Wade Erickson (24:16):

Excellent. Alright, so I, I know we're getting near the end of our time, but I wanted to pivot a little bit and talk a little bit about your personal journey. Obviously US Americans, you can hear a bit of an accent in your voice. And then of course watching your video introducing Qua you're in South Africa and so Right. You know, tell me a little bit about how maybe those experiences and how you wound up in California and, and, and, and how some of that path, you know, 'cause a lot of times nonprofits are driven by purpose and Yes. You know, and that's why they, you know, forego the, the grand capitalistic ideas of becoming rich off this thing and really being supportive of the community. Can you tell me a little bit about that personal journey.

Reza Rassool (25:07):

Sure. Well, yes, Wade, you can tell you can tell from my accent, I'm not from around here. But I was born in Cape Town in South Africa at the height of apartheid. Anyone that was from that environment can't help but be informed and influenced by that sort of that, that grounding experience. I come from a long line of troublemakers, I guess. My dad was an anti-apartheid activist. His activism was practiced at school. He was an English school teacher. And part of the way in which we resisted, he resisted was taking the nation to school. It's very important if you are not, you should not just be a rabble-rouser and stoke anger. You need to do this in a a principled way. And the, the group of activists that he was part of were very principled in the way they conducted their, their resistance. So how did we get from my accent? You can tell it's not South African. It's actually a British accent. Dad's colleagues were being placed under house arrest. And so he left to London to take up a teaching position in the east end of London. And my mom brought me and my two sisters to London the following year. So I grew up, I I got a UK education studied physics at King's College. I fell in with a group of computer scientists that were working on an interesting project on how to edit a movie on a computer. The product was called Lightworks. It started selling very well in Hollywood. On the strength of that, I guess it was a startup on the strength of the success of that startup, I moved my family to, to Los Angeles. In fact, you can see there the, the edit the control surface of Lightworks is still actually used. The recent movie killers of the Flower Moon. In fact, all of Martin Scorsese's movies are cut on Lightworks. So it's a product that transformed this town's industry much like AI is going too transform every other job. So I've been through that cycle of how technology and automation can have material effects on the live, the lives and the livelihood of working people. So look, my, my motivation for forming Kwaai is is informed by my, my whole life story. The word Kwaai actually means angry or wild, but it's also got a, a mixed connotation also means cool, like the word sick or the word wicked has two connotations, you know, if you use it in its street SL form. So I chose the word Kwaai cause I think yes, you should be a bit angry about what's going on if you're not, but just be cool as well. Be cool. And so Quai I thought was a, a great name for a movement and it turns out it's a rallying cry that many of the open source community can get behind.

Wade Erickson (28:49):

Great. Great. Well, before we wrap up, I wanted to introduce next week's show real quick if that's okay. Next week on the 28th we have Anan Safi, director of Engineering at Xenia. I hope I'm pronouncing that right. And the topic is team cultures and collaboration amidst chaos. And anybody that's built enterprise software systems knows, well, that can be a little bit crazy sometimes. And having a good team culture of support is very beneficial to getting to that end zone and, and actually releasing the software. So again, next week, March 28th, probably more at our traditional time there at nine 30 and yeah, hopefully see you folks there. So Risa, thank you so much again for coming back to us at Tech Leaders Unplugged, sharing with us your new project. I'm so excited. I've actually joined the community myself. I, I did early on when we were talking about it and I have a, a strong interest around AI and machine learning edu education technology. And that is obviously an area that is struggling to get adopted within the traditional education world. So I think there's going too have to be disruptors outside of the traditional education community that's going too have to build these products. And hopefully we've got some good tools that can be cost-effectively used through Kawaii to build some stuff for education and technology in that space.

Reza Rassool (30:19):

Thanks so much, Wade. Thanks Carlos. Thanks to the Tech Leaders Unplugged team. Yeah, maybe I'll come back maybe a year from now and tell you how we've done with personal AI

Wade Erickson (30:32):

And we'll share, well, we'll be creating a blog and we'll get some links to those videos for the, and make sure that everybody can come and learn more about and maybe join the community.

Reza Rassool (30:44):

Thanks.

Wade Erickson (30:45):

All right, thanks.