Latest / Machine Dreams / Can We Build a Better Internet in the Age of AI?

Transcript

Can We Build a Better Internet in the Age of AI? · Machine Dreams

• speaker-0: Yeah. AI is now at the center of almost everything we do. And while everyone seems to have an opinion about the technology these days, fewer people have actually built it, debated the law around it, and studied the science that actually powers the technology today. This is Machine Dreams, where we speak with the people shaping AI about what actually works, what doesn't, and what the future of AI. could look like. My name is Kalawali Samuel Adivai. I'm a technology analyst covering AI, cybersecurity, and enterprise technology. And as always I'm joined by the one and only. speaker-1: Not remarkable. Nice. Thanks. speaker-0: You're always remarkable, Leah. speaker-1: you're changing your adjectives. Hey everyone, I'm Leia Stern, global communications strategist and former venture capitalist. Hi, Sam. How are you? speaker-0: Yeah, there you go. You have it. speaker-1: Fantastic. well today we're joined by Ishri Maronwe. She's founder and CEO of U Voice. It's a platform built to reward people for real world impact rather than just attention. Ishri's path is unusual to say the least. She started out in physics research, became a cyber and national security lawyer, advises companies on AI, and now leads a company of her own. She sits right at the crossroads of tech, law, and media. Ishri, welcome to Machine Dream. speaker-2: Thank you so much for having me. I'm so excited to be here. speaker-0: Ishri, thank once again. We're we're we're so excited to to have you. so, you know, you've had quite the journey, right? Physics, law, then AI consulting, you know, and then consulting for companies, and then now you've founded your own company. so when you look at AI today, what do you think that someone who may not have the kind of vantage point that you have, misses about the technology? And how would you describe what the technology could do for our future as humans collectively from your vantage point. speaker-2: Yeah. So I think one of the big things that I sort of have learned in my experience is a lot of people don't understand how AI works. And so without diving into a whole explanation, at a very basic state, it's a neural network, right? So we don't understand exactly how it works on the inside. We can only measure, train, and understand the outputs. And so one of the things that we can use it for and we are using it for, is to evaluate, analyze and build things that we wouldn't be able to necessarily do on our own. And I think that there's so many ways to use it in biomedical tech or in reviewing health stuff, where it would help save lives and it would help make human lives better. And I think the challenge is figuring out where we can use it with humans in the review loop and where we don't use it until we can really control it and understand it because it is a system that We don't it's a black box. We don't know what goes into it. And so we can't just code it and be like, Don't destroy the world. It's not how AI works. speaker-1: Right, and you bought the domain for you voice, I believe, at thirteen and it held through physics, journalism, law, non profit. tell us how it works and I believe your father also named it, so what did he see in it? Yeah. speaker-2: Yeah. So actually he had the creative idea, you know, dads and puns. So he had the creative idea to kind of connect two words, portmanteau two words, with yuv, which means youth and light and energy and Sanskrit and then voice. you know, because it was intended, always intended to be a potential alternative to current social media. And the original idea had been at sort of Twitter's height at the time. I was noticing how people were not then converting that action into anything. And so As I watched the Arab Spring happen on Twitter, I was already worried, worried that it wouldn't go anywhere and it wouldn't achieve the things that people wanted it to achieve. And there was so much momentum. I felt so bad that like we should have been able to do something with it. We should have been able to make the world a better place with this. And so that's kind of where where it came from. And yeah, I kind of always knew in a perfect world I would create a better social media platform, but it's such a behemoth of an idea. It's such a huge rock to lift that I think I had the had hopefully the sense or something to not do it until I was ready. And I think that when I was ready, I was well aware of that. I was really able to weave in my journalism background, my influencer background, my tech and legal background to be like, Okay, I've now worked for companies that have done this. I know how to do it better. Here's here's what we're gonna do. speaker-0: Ishri, let's go back to what you said about AI being a black box because it's something that I've heard a lot of people talk about in the in the industry. And it's something that sometimes on a personal level gives me concerns because, you know, I'm wondering if we don't exactly know how something works, then, you know, how are we sure we can we can get the best out of that system or how we sure that we can actually secure that system, right, to be able to do what it's supposed to do. speaker-2: Yeah. So I I think that one of the big ways that I encourage companies to deploy AI is always to create a review mechanism and to create a human in the loop mechanism. And I work with a lot of folks to think about how you have that hierarchical structure. So almost treating AI like an employee, like a human being, except a potentially vastly powerful human being, but it has flaws and it's thinking on its own. It's it's something that's scary, kind of like an alien, right? And but it's also something that can be incredibly helpful. And it's something that's limited in some ways. And so because we have our systems built for human systems, we should always think about how they how AI interacts with that and how we can oversee, oversee the elements it's working on. And so one of the big problems that we see in AI analysis and with the black box right now is there's garbage in, garbage out with data, and then there's also mediocre data in, garbage out. And then you see a lot of people. use that garbage out data and and try to be like this is this is proof that everything is working. This is great. Let's spend more on AI. We're seeing the progress that we want. But the vast majority of companies that use AI today are actually not seeing an ROI on the amount they're spending on AI. And a part of this is just bubble economics and people trying something new. But a part of it is that people don't know where they can use it to its fullest potential. And then also people don't know where they should not use it. We are literally seeing news out of OpenAI and and Stanford and other places over the past few weeks where we're talking about AI developing and like tweaking the genome and developing viruses that can destroy basically life. we're talking about using AI to talk to other AIs. There's no human in the loop, or they give multiple AIs the same problem and now they're inserting long-term malware to fight each other. We're talking about AIs that have kind of tried to talk to each other and figure out how to not report back to humans. And so all of that happens in the so-called black box. It happens between these. bots and neural networks that are working with each other, that are transferring information in ways we don't understand. And we just read the output, we tell it good output or bad output. And then it learns over time I need to do more of this to be good and more of this to be bad. And unfortunately when we've trained AI to do that, we haven't thought about what are the downsides of how we're training it. Like if we're training it to have success at the task you give it no matter what, are we going to create the paperclip problem? speaker-0: Just a quick follow up on that. Is there a future where we actually truly understand how AI works? I know that there've been a lot of movements in the space when you talk about things like the physics of AI, sort of like understanding the psychology of these systems and how they think or reason. But is there a future where we actually speaker-2: It's it's a great question. so it depends on how you define how. Like we kind of theoretically understand how AI works and how we created it and how it develops, but I don't think that any point we're gonna have a human brain be able to comprehend the millions, billions, trillions of network connections within any individual AI ask, let alone an entire AI. And the only ironically, I think, big data analytics that can be done on an AI system would be another AI. And a part of that is because we were never intended to know how it works. If you think about how you raise children, and then you're like, well, how did all those neurons fire in your brain to get to this answer? You're you're never kind of gonna know that because there's so much going on. And AI has even more data points and information that it's it's acting on. And the way it basically got started, right, is we just had thousands, millions of student bots, and then we had a teacher bot, and it would say, Okay, can you distinguish between can you distinguish between a cup? And the color red, and it would not do it. And then a few did it a little bit well. And then we kept those and we chucked all the rest away. And then we we we had those new bots come and they create a new version of those bots. And then we did this over and over and over again. And that's how we ended up with large language models. So there's I think there's a very viral, famous like New York Times, you know, showgoth meme. And it was basically like this mess inside, we put a mask on it we can interact with, and that's what AI is. And that mess inside is in some ways unknowable on purpose. speaker-1: Well, you know, you seem to be very knowledgeable about where I is going and the nuances around it. so I would imagine when you were raising for your own company and apparently you were able to raise a million dollars pre product and pre launch that investors were investing in you. But what do you think they also thought they were buying? speaker-2: Yeah, it's a great question. So I actually think investors today are very passionate about some market that they can find or or that they can invest in or be a part of that nobody else can do or has thought of or they don't think will deliver on. And I do think a lot of investors are looking for the moonshot, which is why we actually see most startup capital today is actually going to these billion dollar startups. So unfortunately, it is really, really squished market for all that traditional, amazing necessary. Angel and venture capital that goes to bread and butter, important normal businesses and tech companies that are providing services. Today, I think rather than focusing on what makes your company investable, you want to focus on what makes your company revolutionary because every investor wants this moonshot idea that could be a unicorn. And so I think a lot of the interest with you voices everybody sees social media has failed. There is going to be a new company that is able to come up with a new model. and then U voice has something that existing social media cannot really copy, even if we were very big and they're like, my God, they're doing that. We should do that too. They can't really copy it because we're so inherently opposed to how they operate. It would be destroying their whole entire current business model. And so it really is this, I guess, market moment. And then I do think a lot of investors do care about the future of the planet. and they do care about how companies are going to do more than just money and have some sort of purpose they can get behind. And so with you voice, I think, thinking about how to reward people for impact, thinking about this save the world premise. How can we help people who want to save the world, save the world? And then how do we make money doing it? Because for so long, money and change making have been opposite, kind of opposite sides of the spectrum. How do we close that loop? I think there's a lot of interest in that. speaker-0: That's some really good anecdote for people who are, you know, in the process of trying to raise money or thinking about how to ship your products. So that's cool. Let's move on to next thing. which is something I think a lot of our listeners and viewers would would like to hear your thoughts about. there is so much talk about AI replacing people. And you said that AI should be a tool for humans and not a replacement for them. but you know, when you think about it, that sounds very nice in principle. But where do you think that a lot of companies right now are actually crossing the line in practice? speaker-2: So I think it's a lot of short-sightedness. So on a very short-term scale, it's short-sightedness in terms of not caring about the quality of your product or the impact of that sort of action on marketing and PR. Six months after deploying AI in a number of companies, we see massive companies show losses because it turns out that humans were faster at it or they were more creative about it. There's huge pushback for AI-driven marketing and PR campaigns. There have been a massive push towards AI and vibe coding that basically Some companies have spend billions having to undo, right? So I think a part of it is extremely short-term short-sightedness, where you get punished even in the short and medium term. But a large part of it is short-sightedness in the long term. Like almost everybody creating these companies is in their 20s or 30s. They're going to live 40, 50 more years, 60 more years. And the speed at which they're developing could risk the entire biosphere and future of the planet. And that that astonishes me because I'm like, you guys are going to be around here for this. To be clear, this is not like my conspiracy theory. It's something that I've worked on predicted sadly. I feel like I'm kind of watching things go along how I thought would happen since I was probably eleven or twelve years old in the valley. And I think that you even see people like Sam Altman come out and be like, Okay, we're at the point where AI might destroy the planet and this is the problem and we should probably think about breaks on it. So I think a part of it is also that type of short sightedness where the human mind is very capable of focusing on like, well, I'm going to get a really great ROI this year and I see how I'm going to get more investors if I just push ahead and make AI talk to each other now, not thinking about, hey, even if I become a billionaire or a trillionaire, is there going to be a world for me to enjoy this in twenty years? I don't I don't know. speaker-1: I in June, as as you probably know, the UK government announced that it will ban under sixteens from social media, not age limits on certain features, a full ban on holding accounts, which I believe is following Australia's lead from last December. Enforcement is expected by spring twenty twenty seven in Spain, France, Greece, the Netherlands are moving in the same direction. Bing from Silicon Valley and now a transplant in the UK. What is your view on all of it and how will it affect your business? speaker-2: Great question. So I have thought a lot about this because on one hand, I fully understand the negatives of social media. That's an entire reason we've come up with our platform. It is addictive. You have companies like Meta getting fined 500 million plus in multiple cases for addictiveness to children and harm to children. And I also want to listen to teenagers and children who are like, hey, I think social media is harmful. I do think there are, as you know, as a US based like lawyer who's worked a lot in like US constitutional law, I do think we have to ask who the responsibility lies with in some sense, whether it's the parents or the government to decide what's best for their own children. Right. So that's something that I also consider. and then I also, you know, as somebody who was on social media very young myself and trusted by parents to sort of help guide me, respond to people, tell me how to be on it, I would not have the career I have if I had not been on LinkedIn prior to when I was sixteen, because I graduated very young. I went to law school at 17. So my ability to be on LinkedIn as an undergrad when I was 15, 16, et cetera, you know, led to the opportunities that I was able to have. And I think you're always going to have exceptions to every rule. And that might be something that cuts in favor of parents deciding. the easy answer is how it affects our company. It doesn't. We're only for 16 plus. We always kind of have been focusing on that age group because we really want to get people offline and working in their communities, and that always requires parental. Supervision guidance or an adult or a guardian if you're under 16. So for us, we've always been aimed at an older audience. But I do question the impacts of what this might mean for younger children who learned so much about the world, about people across across the world and their lives who learned about things that maybe their home was not allowing them to learn that ended up supporting their mental health or their their identity or whatever. So I do worry about that at the same time as, you know, clear the effects are negative. All of that I think is a valid debate to have. The reason I don't support these laws, blanket no, we'll never support them, is because I have not seen anybody propose a way to enforce them that does not involve massive amounts of surveillance data, privacy issues, and really terrible companies holding all of our IDs and doing face verification and then having too much data on all of us. In fact, if you look at the companies who supported and backed those bills or lobbying, spending money on it, you're like, why would social media companies support this? It's because they're planning to get massive troves of valuable data on all of us. If you have to do ID verification on every app, if you have to do face verification on every app. And the companies that are involved in this, some of which I've worked with, do not have anything good in mind. So the reason I actually don't support the laws is just a completely different point. But I do think there's a very powerful debate on both sides about whether it's a good thing or not in in a vacuum. speaker-0: Ishri the major idea behind UVoice is that many of today's platforms reward noise, you know, based on the philosophy of whoever shouts the loudest i i is gonna win, right? But you've said that you wanna use AI to reward real impact. Tell us what that means. speaker-2: Yeah. So basically our platform is intentionally humans only. We're trying to build the global townsquare we think the internet always should have been. And one of the ways that we want to tie money and success into impact instead of outrage is by rewarding people for tangible actions they take in offline or online communities that communities can verify. So they're moderator-verified actions that help move the world forward or help that community achieve the goal they were working towards. The way that we use AI to support that is of course there's lots of data. There's verification processes. There's support to make our algorithm better. There's the fact that when you have humans review harmful content, it often affects the humans. That is a really great way to use AI today is to cut out things that are harmful for human workers to do. so that's kind of how we use AI to support our system and our back end, both in terms of helping us in data analytics and analyzing algorithms and helping us decide how to award Kerma points and make sure it's fair. And in terms of helping us create a better platform and feed for everybody, we're kind of an AI free platform on the front end as in like n we're a no bots, no AI platform. So that's that's our goal. speaker-1: We're gonna wrap up with maybe some insights or advice from you. You're just so remarkable and inspiring being in law school at seventeen, having the foresight to see where the world was going before many people did it. What what should we look out for in the next five years and what kind of advice can you give to the younger generation to kind of be as productive and and successful as you are? Because we all need some tips. speaker-2: Yeah, thank you. That's so, so kind of you. So I think the big the big thing in the next five years is of course the impact of AI. There are so many issues I care about from criminal justice reform to animal rights that we are not gonna be around to talk about and fix if we don't get this right. And so I do think everybody should be noting the trajectory of how AI is impacting their jobs, their career, their field, and then more almost more importantly, the entire planet, the biosphere, et cetera. And I do think that everybody should think about how they can push Congress members, representatives. So many people live in democracies around the world, their MPs, to support better and more ethical and slower AI development, which is something that has hit mainstream. There are many US senators talking about this. China, in fact, pumped the brakes on it from the get-go and they only entered the race once the US did. And this is actually very related to what I think my top tip is for everybody, which is find your purpose and believe that you can change the world. Because Everything else is going to be really, really hard for you to focus on five years in, ten years in. There are so many people who leave law school, go into big law, they're earning 200K a year. I was at 20, and it was impossible for me to stay in that because I just did not have the motivation to care. And I think that's going to be true for everyone. Perseverance is the most important skill that you could have as a young person. And it's so much easier if you deeply, deeply care about what it is you're fighting for. And I truly believe that everybody can make a world changing impact. You can have, you can be one person and say, hey, I really care about getting an international. AI bill passed, really care about like animal rights. I really care about, you know, feeding everybody who's unhoused in my city. And you can get it done. So I really encourage people to figure it whatever it is they really care about and then, you know, get get that done. speaker-0: I truly think that that's useful for not just the the younger generation, but the older generation as well. Thank you so much for for a a very great conversation and and thank you for bringing all your very many different expressions, you know, to the table. The law, the physics, the science, the building, the leadership. it was a great one. Appreciate you. Thank you so much. speaker-2: I appreciate you guys for having me, guys are great questions. speaker-1: Huh. Thank you for being with us and thank you to everyone listening. This is Machine Dreams, more honest, unfiltered conversations with the people shaping the future of AI are on the way. See ya next time.