![](https://cdn.prod.website-files.com/63da75e1ec536b89a802545a/65c96acfba70c4104292045f_Noah%20Kravitz.png)
Noah Kravitz is the host of NVIDIA’s The AI Podcast, where he connects with leading experts in AI, deep learning, and machine learning. He is also the Founder and Chief Consultant at Resonant Digital, which works with VC-backed startups and Fortune 500 companies to deliver effective content strategies, development, and distribution services. As a seasoned journalist charting the rise of electronics and computing, Noah built one of YouTube’s most successful science and technology channels, served as a tech expert on various TV outlets, and wrote about society-changing technologies in media publications, including WIRED, Business Insider, and Entrepreneur.
Here’s a glimpse of what you’ll learn:
- What are LLMs (large language models)?
- The fundamental differences between various AI chatbots
- Noah Kravitz defines and provides examples of multimodal LLMs
- Practical use cases for AI agents — and how they’re transforming businesses
- AI agent optimization and personalization
- The long-term vision and future of AI
- Noah addresses AI’s role in technological advancements
- How new generations can navigate the explosion of AI
- Advice for leveraging AI responsibly
- Why Noah is regarded as infamous
In this episode…
With AI models taking the business realm by storm, individuals and corporations alike are strategizing methods for leveraging the tool. As buzzwords like multimodal large language models emerge, what is the current and future state of AI, and how can you prepare for what lies ahead?
With deep involvement in the AI space, journalist Noah Kravitz defines large language models (LLMs) as AI technology that uses transformers to process language. These models include ChatGPT, Google Bard, Jasper, Bing Chat, and other tools you can leverage for various content marketing campaigns. Multimodal LLMs elevate these capabilities by generating videos, images, and speech simultaneously. In the near future, AI will enhance productivity by automating repetitive tasks and workflows. Long-term uses of AI involve harnessing personalized agents that can replicate human behavior, creating a science-driven utopia. In the meantime, Noah recommends experimenting with public-use AI chatbots like ChatGPT and discovering how your skills integrate with this new technology.
In today’s Up Arrow Podcast episode, William Harris converses with Noah Kravitz, the host of NVIDIA’s The AI Podcast, about the potential of AI in the business landscape. Noah mentions AI’s role in technological advancements, how to leverage AI responsibly, and the main differences between AI chatbots.
Resources mentioned in this episode
- William Harris on LinkedIn
- Elumynt
- Noah Kravitz on LinkedIn
- NVIDIA
- The AI Podcast
- AI Superpowers: China, Silicon Valley, and the New World Order by Kai-Fu Lee
- Apt
- “The Future of Retail With Brittain Ladd: Hyperlocal Logistics, AI, Fast Fashion & More” on The Up Arrow Podcast
Sponsor for this episode
This episode is brought to you by Elumynt. Elumynt is a performance-driven e-commerce marketing agency focused on finding the best opportunities for you to grow and scale your business.
Our paid search, social, and programmatic services have proven to increase traffic and ROAS, allowing you to make more money efficiently.
To learn more, visit www.elumynt.com.
Episode Transcript
Intro 0:03
Welcome to the Up Arrow Podcast with William Harris, featuring top business leaders sharing strategies and resources to get to the next level. Now, let's get started with the Show.
William Harris 0:15
Hey everybody, William Harris here. I'm the founder and CEO of Elumynt. And the host of this podcast where I feature experts in the DTC industry sharing strategies on how to scale your business and achieve your goals. I am really excited about the guests that I have here today, Noah Kravitz for the past six years, Noah has hosted the NVIDIA AI podcast, interviewing nearly 200 guests and racking up more than 4 million listens. He's been working on technology for more than 25 years as a journalist, consultant and entrepreneur. He's newly a partner in AI ready a consultancy focused on helping businesses leverage artificial intelligence. And he is here to kind of walk us through what we need to know about AI and we'll try to tie this into E commerce a little bit. I think it's gonna be a lot of fun. Noah, welcome.
Noah Kravitz 0:57
Thanks so much for having me. I am I'm psyched to be here. We were chatting as we were setting up. I'm so excited to be in the guest chair. It's been a long time. I don't even know which appointment. Well, I have. We'll get to that later. But yeah, it's cool to sort of, you know, I'll share what I can. And I think it's definitely going to be fun. And thanks for having me.
William Harris 1:18
Yeah, thanks. And I do want to give a shout out to Andy Hetal, who runs digital marketing over at Winnebago industries. He is the one who put us in touch. He's an absolutely wonderful human being good friend of mine. And so thank you, Andy, for for putting us in touch
Noah Kravitz 1:31
shots to Andy. Good guy.
William Harris 1:33
Yeah. So there's a lot that I'm excited to cover this show with you, we plan on going through the immediate state of AI that we're gonna get into the near future of AI distant future of AI. And we'll transition a little bit into some more practical stuff like how do we prepare our kids and practical uses for E commerce. And we'll end with a little time where we can just kind of get to know you a bit more personally, because I understand that you are once infamous, which you hinted at just a little bit ago. So
Noah Kravitz 2:00
we'll go five days to five days to solve AI six days for the other to the other stuff to get to know you and then will rest on the sub sorry.
William Harris 2:10
Yes, something like that.
Noah Kravitz 2:12
Let's get into it. Now it's let's let's everything you want to know about AI.
William Harris 2:17
As I know, yeah. So I want to start with the immediate state of AI and the things that you and I were talking about before that I think are the the things that people are the most excited about. And that's getting the most chatter about right now are MLMs in multimodal MLMs. And so I want to talk a little bit about like, let's start with just what are MLMs and multimodal MLMs? And how do they differentiate? And then we'll get into maybe you know, why they matter and how they're evolving?
Noah Kravitz 2:43
Sure, yeah. So an LLM LM is an acronym for large language model. And it's funny because I I feel like, I started to choose my words carefully, because there are a million, you know, data scientists and engineers and like people who really technical people who know their stuff, who I don't know, and they don't know me, but I imagine they've listened to the video pod at some point, or at least once. Should I don't want to get it wrong. And I know I'm not getting it. 100%. Right, I'm imagining them. But essentially, an LLM is an AI model. Right? So it's like it's a it's a piece of technology. It's it's, it's not a piece of software, but kind of akin to that, right? It's a piece of technology that runs on a computer, you know, you can access from the computer. And it's built around some. And so it's really good at processing language as as the name implies, right? I was trying to keep myself from going into the weeds already. And so chat GPT is the best known application that runs on an LM championship chat. GPT is a chat bot, that open AI and apologies to the listeners to know this. Josh JpT is a chatbot that this company called Open AI released just about one year ago, right now, as we record this, and it became the fastest adopted consumer technology product every I think, if I've got that, right, right, but like crazy, you know, 100 million people, like in the blink of an eye, you know, we're using it faster than whatever else. And essentially, what it is, is it's a chatbot so if you've ever like you know, gone to a website to get customer service is the easiest example and you've got a chatbot that pops up and you talk to that instead. That's a chatbot and there are lots of chat bots that don't have anything to do with MLMs per se chat bots existed before MLMs. And, you know, so not all, not all chat bots are like this. But what open AI did is they release a chatbot that you just type to and it sends what you're typing to their LLM GPT GPT general purpose transformer, I believe I got that wrong. But essentially the AI thing, the technical thing that's at play here is a transformer. Right. And so an LLM is a model that that leverages transformer to process language. And so people started using, you know, chat GPT. And realize, like, it could do all kinds of crazy stuff with language, or appear to anyway, and we'll get into that. And I don't mean that as as a shot across the bow, I just mean, like, we'll get into what all that means. And so in the tech world, this calendar year that you know, we're in November now 2023, this whole year, and certainly in my world, has been the year of LM, and Chad GPT. And open a eyes, LLM are not the only game in town. And there are and, and you mentioned multi multimodal, we'll get into that there are other models that generate images and audio and things that are kind of in the same sphere that you put them all together and you get some crazy stuff. But to my LLM, there are others from like, meta, and Google some other companies, but open AI is the front runner by far right now. And GPT is the one everybody knows and you know, is used, or the people have used it has used it and, and so like all of these startups have been birthed on top of MLMs, and just all kinds of money being put into them. And it's both, like a crazy, you know, classic tech world hype cycle. And also, a lot of people, myself included for whatever that's worth, but a lot of people who I've, you know, been fortunate enough to interview or even just rub shoulders with, AR are really like, you know, AI is not new, get into that and but generative AI in this mainstream, you know, thing that's happening, you know, is sort of new, but you put it all together and like AI is a big, big, big deal. And generative AI and the stuff going on with MLMs is within that a very fast moving big, big, big deal. So there's a lot of big deal energy, you know, and a lot of it's real, and then there's a lot of hype. And so that's what an LLM is
William Harris 7:26
Yeah. So a couple of the LLM you know, obviously chat TPTB one of the biggest ones. I think metas is what llama llama, Google Bard and I've played with I've at least played with Bart. I got into that one the day of I haven't played with warm at all. So
Noah Kravitz 7:45
Bert is the if I've got it right Bert is the bot. And I think it runs on pom pom pom is a Google as a few MLMs I think POM is the one it's got LM like llama has LLM and yeah, so yeah, all these and then there's anthropic and cohere. And you know, now there's, there's like, I don't say there's new ones every day, but they're new ones every day. But those are those are we covered a lot of the big ones. Yeah.
William Harris 8:20
So like, what differentiates them? Like, in what ways are they different fundamentally, if you're aware of, like, at least in my own personal testing of some of these things? For me, chat GPT is significantly more creative. Whereas I felt Bard was more accurate, which was an interesting thing, at least in the questions that I would ask it. But like, how do these differentiate?
Noah Kravitz 8:41
Um, so I realized, as you were talking, that I can't be on a podcast and you know, 2023 and say things like, I think GPT stands for like, No, we're on the internet. I can look stuff up, so I don't, I don't get it wrong, some of the time. How do they you know? It's a great question. And like, they, so I'm gonna, I'm gonna answer two ways, right? That I'm gonna try to give the TLDR and then and then getting the lottery winning, cut me off your podcast host, you know, but they differentiate as you kind of as you started to speak to in the way that they actually behave and their actual, you know, capabilities and strengths and weaknesses, sort of relative to one another and that kind of thing. And so like, sure, I haven't used Bard very much, but I've been playing around with anthropic six, Claude. I think it's Claude two. I've been using the free version, which I think is too but I don't I think they're up to two though. And Claude I started using because I talked to somebody who, a developer who said that he'd been using cloth he'd been liking cloth Odd for kind of factually accurate sort of, you know, less creative, more reliable. So maybe along the lines of what you were saying, but it came up in the context of me looking for something I could use in my work doing like marketing writing, where it's a lot of customer, customer marketing type stuff. So it's not journalism. It's not journalism. But I and given my background, whenever I sort of liked to build around things that the customer actually said, my brain starts there versus starting with the marketing tentpoles either way is fine as long as you wind up, but point being it sounded clouds sounded like, oh, for the kind of work I do, it might be. And so I've been using Cloud more lately and liking it. I did actually a comparison of when I started do a comparison of like Claude GPT, and whatever Microsoft calls, which is Azure is Azure open AI. So it's, it's open AI is model it's GPT underneath, but, but it's Microsoft's bot bird. No, that's Google, what do they call Microsoft? Bing chat, actually, I think. But my comparison quickly broke down into like, which of it was for a legal tech company, and I was actually looking up. case summary. So I was like, you know, give me a case summary of spoiler alert, redacted, because we're going to talk about it later, probably. But this case, you know, and, like, three of the four or two of the three just flat out refuse to give me anything. For like, I spent too much time on Rhett. I enjoy Reddit, I'm gonna say too much. I enjoy Reddit. There's a lot of complaining in the GPT subreddit about GPT getting nourished over the past six months, or longer. I don't know enough to know. But it felt like I did add those fives. But anyway, so they did they differ though, right in these different these different capabilities. And you sort of separate out the chat bot from the LLM. And and I don't know how deep we want to go. But it's a difference of like, I'm using this as an end user. So I can just go to the web, or at least an open AI as case like, fire up. They say fire up open up smartphone app on iPhone app. I think they have an Android app too. I don't know if Sure. And I can use the chat bot, right. I can bang you can for sure you can you know you can access right? Versus do you need API access? Right? And are you building you know, even like a no code app or something like that, you know, you're building something where you have to get in and so they're slightly different? Or more levels of how they differ, right? Because I think that there is some level of the guardrails are taller and more impactful in the chat box, perhaps they are the API's. But yeah, so they differ in that, you know, some are a little more creative and not and you can tweak settings, like temperature and some other settings, to, you know, to get around that a little bit. And then a lot of it's like, you know, it's the, the short answer, by the way, the TLDR. The longer and yeah, so that's how I roll. The longer answer is that I mentioned before, you know, transformers, right? It's this technology that was invented, as I understand it, it was invented at Google. And it's a technology that sort of underpins how these MLMs work. And I'm going to stop there, because I don't want to get more stuff wrong. But the relevant point is that, and I learned this at an event I went to, I participated in, but I was sort of like the warm up act. Because the rest of the speakers were all like engineering leaders from all these companies who've been talking about, like the people who make this stuff. And they kept talking about it. And the one guy in particular had a slide in his talk, which is great. This paper, the seminal paper, which is called, all you need is attention or attention is all you need. I think it's a play on the Beatles song. And it was kind of like announcing the invention of the transformer. Right? And if you look at the authors of that paper, and they're all in this Google lab, one of them was actually like, a student intern from the University of Toronto, the guy who founded cohere. But if you look at the co founders of that paper are the co authors of that paper, and you trace out how much money companies are not even the money you trace them out. Right? And they're the ones building these MLM companies and you know, so But that's the root of the answer to your question is like, I'm sure there are technological differences, because this field is from research to product to societal business impact. It's advancing so quickly that like, I don't know, how much has changed on the tech side, you know, the core tech side of like, since that paper was published, actually, like, do we still do it that way? Or do we do it that way, but these things are different, or like, now, that's, you know, that's advice that now everybody's eating vegan, you know, like, I have no idea. But right, but but, um, you know, so they differ in the way that like, other technologies would differ, right, everything from functionality to how they're marketed to what level of DEV access, how much they cost, who their partnerships with, you know, all that stuff.
William Harris 15:53
So let's go. Let's go quickly, then into the difference between just like the regular LLM is we're talking about an multimodal ellos. The only one that I've really played around with at all that I that I'm aware of as being a multimodal, would be seamless M four T, that meta meta launched was really interesting for like a new translator. But what is the difference between the regular LLM we're talking about and multimodal LLM? So what does that mean?
Noah Kravitz 16:21
Right? So my understanding is that modal refers to the the mode of data, right? And again, the alarm bells going off. Like there's a technical term, I don't know it, but it's like, text, the modal communication, is it text? Is it still images? Is it you know, potentially other things? And and so it's not just like, there are models out there that are, you know, Chechi VT we've been talking about is text to text. dal e, which is open AIS image generator, and some of the stuff has been a little bit updated. But we'll get to that in a second. Mr. Dahle II is opening my eyes to image generator. Stable diffusion. What was a big one runway, there's one I'm missing that my journey is core. Thank you mid journey. Yeah. So those are all like text to image. Right? Then you have text to audio, music and speech in their, you know, companies doing that right? Multimodal is when you like, do more than one at the same time. And so my sort of like I interview technical people, you know, sort of understanding kind of tops out around there. And then I get fascinated by the practical and philosophical implications of what they describe right, the technical people. So it's not meant to be a plug, but it's a plug, I refer people to a recent episode of the NVIDIA pod if I may, with a guy named him fan, or even listen.
William Harris 18:01
Yes, I actually want to talk to you a little bit about that. Remind Chris is doing? Yeah.
Noah Kravitz 18:07
So Jim is great. And actually, it's not even the the I mean, that episode, but at the end, we got a little more into it. He's a great follow on I follow him on LinkedIn. I don't know what other platforms he's on. But he's always he's a scientist at NVDIA. He specializes in MLMs in a very broad, you know, and technical way. But he's always posting little snippets of things that he's he's seen, either firsthand that NVIDIA doing it himself or cars, or just, you know, other companies, other people on line are doing that relate to MLMs in the future about this stuff. So, so yeah. But kinda, you know, it's. So one example is GPT for Wii and the Wii is for vision. So open AI has this model that I actually I used to have a paid GPT subscription. And I stopped it over the summer, because I just wasn't using it much. And I keep you actually earlier today, I was like, oh, you know, what, before the pod like, I'm gonna, you know, I'll get plus again, and I'll play around with the customer GTs and the other stuff or whatever. And then of course, you know, that, but I will, what do they have? You can access it on Bing chat, though, is why remember the Bing Apple, the Bing app on mobile? I think you can, you can do it online as well. I don't. I don't know how. But on mobile and iPhone, I know it's pretty easy to do. You can access the GPT for the model, which is it's basically what we're talking about GPT the model by Jack GPT. So imagine Chechi VT, but you can also upload an image and you can talk to the bot about the image. And the bot can do some really cool stuff. And some pretty like, amazing if you start to think about the implications of mashing up, you know, image and audio and all these different things, right. And then also like being out in the world with a GPS enabled supercomputer in your pocket that can take a photo, and you can put the photo and then often like, cyberpunk sci fi, you know, territory, right? A little bit of that as well. But yeah, right. But so that's one example. So you can ask it, like, you know, the demo when they first launched the demo was, somebody's trying to like fix their bike, they're trying to raise the seat on an old bike, and they don't know how to do it. And so they take a picture, and they ask, you know, GPT, hey, look at this picture, how do I raise the seat and like, oh, you need a 14 millimeter Allen wrench, and, well, here's toolbox, which ones that, oh, it's that one on your upper right. And don't forget to grease your chain or whatever did right. But like, and then there are people who made more sophisticated like kind of trippy mind blowing, you know, examples on social media. But so that's kind of an early step towards multimodal another example. Just to kind of illustrate the, this part of the spectrum and sort of that, like, where it gets into the more mind blowing type stuff. And a lot of the stuff you have to like see it, you see it, we can talk about it, but seeing it on a video, even let alone doing it, or I'll show some of this stuff to, you know, some older family members who are, you know, they're tech savvy at this point, but they're not, they don't do it for a living. And they'll be like, you know, the first time they see an LLM bot. Right. So I did a bunch of work. This calendar year for a company called bubble, free float bubble IO. But the people I work with are super cool. So they deserve it. And they're a no code platform, right for like building web apps. And I was writing my deliverables is writing, in a lot of cases was with them. And, and as the year went on, I got to do more and more AI related stuff, which was cool. And they had, I don't want to get this all wrong, you should go look on their blog if you're interested. But they had an AI focused event. And as part of the event, they I helped, I helped ghost, right, sort of, you know, future of AI and no code type thought leadership thing. And one of the things that the exec I believe he's co founder CTO, again, you know, apologies is wrong. One of the things he talked about was training, trying to like train a generative AI model on completed bubble apps, to see what the model would respond with, right. And so as sort of like to understand, wrap your head around that right, like, bubble that you you know, with with, they call it no code. You know, in some cases, I think it is no code. But you know, with note with zero to minimal coding skills, you can kind of drag and drop and tweak, you know, tweak knobs and sliders, so to speak, and build a functional app that can you know, I don't know, escape a calendaring app like or, you know, business apps, people doing all kinds of stuff with it. And so they're thinking about how do we use Gen AI? And so one idea is, you know, I wrote a bunch of like, tutorial type articles for them and other clients, you know, and so one idea is like, Oh, well, for learning, you build like a learning bot. And you, you have it ingest all of the learning content you already have. And then people can ask it about, you know, ingested developer documentation and like, Yeah, cool. That's, but let's like, go further, right? What if in the same way that you train, you know, in an image producing Gen AI model and other images, right, hot dog, not hot dog, even a classification model, you know, whatever you like, you train it on on a type of data, data type image, right? What if data type was a bubble app? Completed apps, right? And so then the model could start to think of not thinking about that. But the model, the model, the model, if you wanted it to, could generate because the whole thing about generative AI is it's generating something, and really, it's really just generating like highly accurate, highly sophisticated predictions, right? It's all about like, you know, fill in the lake. It's like auto complete, gone, gone mad. Sure. And so like, what If you could get the AI model that was trained on all of these bubble apps people have made to spit out new bubble apps, what would that look like? Would it look like you know, and it's not just a bubble, right? It's like put in whatever it is that you know, Shopify? Is that my ecommerce? It's awesome to be on here. And like, I hope I should, I don't know, I don't know a lot about e commerce. So, you know, I'm fascinated by how it all ties together. But, but like, what if you could, I don't ask it to spit out I think I might have just had a tie in. So it out, like, you know, ideas for different types of apps, or you know, different industries. But even if you get to, like a more abstracted kind of like UI or a UI layer, right? Or UI UX, or like, what does the flow look like? Or like, does it in the quest to spit out an app? Because it's been trained on so many apps built for businesses built for commerce built for data generation of some sort? Or, you know, revenue? Generations? I'm sorry? Does it actually inadvertently spit out like, a million dollar or a unicorn business idea? You know, like, right? What's it gonna, you know, the early days, I don't know, I didn't, I wasn't super duper into it. But I paid some attention. And again, like, I rubbed enough shoulders with people that in the early days of like, mid journey, right, like and images, you know, image generators online. Like the term nightmare fuel became part of my lexicon, right? And you would generate something and you'd get back something you wish you hadn't seen. I'm talking about like, what are you what are your fears? Are that's a fear, but I'm susceptible to like the image burnin in my brain. But you know, like, what if the if the if you trained a Gen AI model, on whatever the closest data equivalent there there might be of successful Shopify. This is right. Like, what might it spit out?
William Harris 27:17
You know, and I think that's a great segue into kind of where I want to go with this, which is basically, why this is exciting right now, why people got so enamored with MLMs. And like you said, AI has been around for a while. But for the general public, this is when it just skyrocketed. And one of those I think, is when we're talking about like the new, it's because of the near term applications of this. And one such that I know, that I'm really excited about is AI agents. And I think kind of like, this is where you're going where it's like, one of the things my brother and I were talking about, he's he's a developer is, you know, it's almost gonna be to a point a race between whose AI agent has been better trained for it's like, what do I train my agent on? Versus what do you train your AI agent on? And How capable are those AI agents, and so they can be trained to, you know, pick stocks, the way that you like to pick stocks, they can be trained to do all sorts of things. But I think one of the ones from an E commerce perspective that has me really saying, Oh, that's interesting, is something that Brittain Ladd, who was a previous guest on this show, said, he said this idea of in, even right now, but in the very near future, going physically to a website is the Stone Age, there's no need for you, as a human being to actually go to this website anymore, your AI agent will go there. And so if you say, hey, I need a pair of new shoes, your AI agent gets, you know, potentially just say like, great, here's three great options, which one do you want and it will have already known enough of what what you want or just goes and picks and buys the one that you want. It doesn't even ask you eventually because it's so good at predicting what you want. But from the concept of these AI agents, what gets you the most excited about AI agents and let's just say this near term application of what we're seeing, right,
Noah Kravitz 29:12
right, so I feel like I need to physically spit the sentence out so I could focus on near term longer range like things could get crazy right? And I don't I don't say that in like a conspiratorial way and I don't say that I was trying I was talking to somebody yesterday, you or I met but hadn't met before. And we're you know as a work call, and I said like look I'm not I'm not like extremist in my like, you know, technology or nothing you know tech utopianism kind of thing. It's our center but like, like aI gives us a lot of opportunity, you know, It's that kind of it's an internet. It's an electricity like, at the NVIDIA pod, you know, the tagline, I think was Andrew ings, quote, actually originally, but like, you know, that AI is going to be like electricity, or it's the new electricity, right? Because it's not such a thing. It's like, what was the man? We don't? Was it three? I'm really dating myself with this reference. But there was some company in there add, I think it was three M that used to say, like, we know, one of them was we bring it No, we, we don't make the blah, blah, blah, we make the bla bla bla better. And that was kind of the, you know, yeah. So. But near term, you know, and with agents and stuff. The idea of agents is super cool. And I learned about the idea and went sort of hands on, I was at the beginning of this calendar year, right. And I keep saying calendar year, because I like, between, like, kids and family stuff, and education world. And then like the work world with the fiscal year, which I you know, no. But sis clear. This year, I went down the rabbit hole. I mean, like a lot of people because it was right after GPT, and the holidays and everything, I went deep down the rabbit hole of trying all of these, like, new things are being built a top GPT three, or whatever it was, that was three, five, whatever was new then. And agents were like, the thing I spent the most time messing around with, and none of the things that I tried, did anything, you know, sort of did and he right, the end product was anything particularly useful. But watching them work was sort of fascinating. And then I dropped out. And so this is a long way of saying that. I don't know what's been going on in the space more recently. But I can kind of, you know, my perspective. So the an agent and tell me if you have a different sort of working idea of it. My understanding or my when I think of an agent, I think of like, you know, a bot that's out there doing stuff on my behalf. And it's trained, built, developed programs where I think trained, well designed and trained to focus in on a specific thing. And so, I've heard people talk about them as tasks specific. So like a travel agent ran haha, but you know, that's a good way to think about right travel agent, like, I have an agent, that when I call upon it, it's to book me travel. And that agent already knows all the stuff it needs to know about me to effectively book me travel, like, you know, and I might need to tell it, you know, am I traveling? What, you know, I just thought like, where am I going? Am I going alone? Or I go with the family. And it just knows, like, right? When I'm alone, I like to travel, you know, if I can get an upgrade, I'll take it right? And if it's the family, it's like, Well, only if there's four upgrades is never for upgrade. So you know, how do we get a little extra leg row without you know, paying $35,000 A ticket, whatever knows all that stuff. What I like to do when we get there, blah, blah. And so then the other type of the other way I heard people talking about agents is kind of a role based agent. So in the the world of say personal productivity, and, and sort of supercharged assistance and that kind of entrepreneurial stuff, right. I've seen things where, you know, you can have an agent that's like a super assistant. But you can also if you're not an entrepreneur, have a group of agents that's like your, you know, your co founders, right, like your CTO, you know, if I'm the CEO, they're my technical co founder or you know, that kind of thing. Or you can have like a virtual board or like whatever, there's one company who I tracked a little bit earlier in the year who has as a demo of their of their tech of their sort of, I don't even know framework, I think it's a talk about the infrastructure of an agent, Agent system, but as a demo, they had like, you know, great figures from history sitting around the table, you know, disgusting things, right? And so they're each trained on different persona and written works of those those people. And so this idea of agents going out and like taking care of stuff for me, for us is really popular, and then along with it. And this is kind of where I've had a little bit more hands on experience and I'm more in a realm of, you know, again, like communicating and marketing and thinking about it and actually building things myself. But on the other side of it is personalization. And so I worked earlier this year with a company startup called you AI, y o u AI. And they're founded by an ex Googler, who's got he's got like a track record of a guy named Dimitri really nice guy, really nice people shouts to them. A product guy right from Google and other ventures. And I, he co founded or founded this company called co G, which is like a creator economy platform of some sort. And you AI is somehow birthed out of that, or birth adjacent to it or whatever. And so it's a platform for people to build. AI is is how they call it to build build AI eyes on it, right. And so and we can get into this or not, but the recent, you know, everything's moving so fast and open AI just recently about announced a bunch of new stuff, some of which undercuts the unique value proposition of companies like UI, and UI is case does not their sole value proposition. And they actually let you use different you can use GBT, you can use Claude, you can use whatever, but just illustrative of the larger wave of like, this is always a fast businesses are built and fall down, whatever, you aim for that. But anyway, um, so they've their thing has been like creator tools, right. So sort of like low code, no code, you know, type things, where, if you've been playing around with GPT, and you can get it to do some stuff, and you sort of have a feel for prompting. You can apply those skills and create things that like, you can reuse, like an app, you know, like, you can reuse, or you can share with people or you can sell, they have, you know, create a platform, a new AI, right. But one of the things that they offered sort of, and I work with them for a few months on a project. And they were like, I don't know if this is product, loneliness, product lead growth, I don't know my business terms. But they were like very, you know, a startup and building out their platform, but doing it very publicly. So they went beta pretty early. And so like, you know, they've come a long way publicly, right? In a short amount of time, which is just thinking back to when you know, 20 years ago, like you didn't do that. But you were saying hardware or a DVD ROM, you couldn't really do it. But so one of the things that they had early on, and they still have it, but I don't know how big of a thing it is, right? Currently, how much of a used features in the platform was this ability to personalize things agent style from the beginning. And they had this, this very simple user interface is kind of like give you like a game I played with one of my kids sometimes at bedtime, you know, it's like this or that, right? Like chocolate or vanilla, you know, Linux or DOS, you know, whatever. Right? And so
William Harris 38:16
an extra class with her kids.
Noah Kravitz 38:18
Just say that out loud. Honestly, I wasn't gonna talk about that. No, I don't. I was trying to bring it back to I try to bring it back to the Yeah, right. I was like, wait, no, this is a tech podcast, don't say ecommerce podcast, wait, no, right, like very simple. And they had a nice little UI to make it very easy to create, like a stack of cards, right? Where you pick one or the other, or some of them are multiple choice or like, you know, rate from zero to five, where that kind of stuff. And it was using that data to train, essentially train your agent on your preferences with this vision down the road. That you know. And irrespective of where you AI's business model is at at the moment. And then beyond you AI and just to the thing of agents and personalization, you know, this vision that down the road, we're not going to be going to web, we're not going to be going to we're not we're not going to look at websites anymore. We're going to get the data in a different way. And I think when chat GPT first came out there was this huge like, it's the new Google, right. And I think more because I'm an old dog with old habits or you know, and not because I thought I was smarter than anybody else. I never really used GPT that way. But a lot of people did and still do. And there's definitely potential for Gen AI in search. But we're super early innings and all of this right but I lost my train of thought.
William Harris 40:02
Okay, I think where I get excited about that with these agents is, you know, for awhile we've had like, let's just say, SEO. And we've we've tried to optimize websites for what we think the search engine wants. And Google said, don't do that optimize for what the people want. And but what's interesting is I do think that there's going to be like a new push for No, actually, you need to optimize for what the AI wants, because the AI is going to be the agent that actually has agency over this. And so I would say for who's going to coined a term, the thing that's going to emerge from this in the marketing world is AI EO, like your AI? Agent, optimization, right. So how are you optimizing this for the the AI agents?
Noah Kravitz 40:44
Sure, yeah. I mean, in marketing. I have, you know, I've always I've worked a lot in marketing, both in kind of larger orders. And then, you know, on my own, I've done entrepreneurial stuff, and, you know, very clumsily tried to market, things I'm involved with, and you know, whatever. But I always function at the end of the branch of the tree, you know, I'm a writer, right? And so you give me the brief or whatever. But I've done some projects with SEO, like people who were like, SEO wizards. So I have great respect for the dark arts, as I think of it, right. And I've heard people this year, people who know talking about, you know, SEO is going to break, and et cetera, et cetera. And like, is that necessarily a bad thing? And exactly, to your point, like, No, it's not a bad thing. But you know, it's just gonna be replaced with another, you know, carrot to chase or whatever. But I think that the, you know, so the train of thought that I lost, which really, it's good that I did, because the actual TLDR version is that the notion of the way we interact with dynamic data, right? So like, whether it's, you know, the internet now, right, like live data of all forms coming in from around us, or it's some sort of application via AI or not AI or whatever, you know, video games, something that's canned, so to speak, right? The way that we interact with all of that will almost certainly yes, be mediated by some kind of agent in the future, and we won't see, you know, you and I won't see the same version of, you know, shop frenzy dot whatever. Right. Right. And even though we don't now, because of targeted marketing, and cookies, and ads, and you all that stuff, like it will be so different. And so the ideal is that, yeah, we will have these agents that are personalized. And they're, you know, it's kind of funny, because I think about it, and I see sort of starting with me, and then branching out by tasks and by roles, right? And there's the like, travel agent, and the one who can get me like concert tickets, and the one who can do my taxes and you know, right, and then I see the other hand, the leg, well, there's the super assistant, well, let's break that down. It's, there's, you know, the CTO and the CFO, and then there's the the parenting guide to help me with be apparent, all those kinds of things, right. And then they all sort of eventually just kind of morph back into one super agent that just knows me. Right? And is that the singularity is
William Harris 43:37
yeah, like, I actually wanted to bring that up a little bit, like, like, how far are we from that? But yeah, so you know,
Noah Kravitz 43:46
to your, to your point, originally, your friend's point? Yeah, like websites like everything else, you know, like human beings ourselves are eventually just a passing fad and and AI and what's happening with MLMs and a natural voice processing and and, you know, screen technology and and AR VR XR, just all these different technologies are like, I don't know that it's all headed towards a particular singular new way of doing things. But there's a paradigm shift happening right now. Yeah.
William Harris 44:23
Yeah. Well, in some of these agents that are around and you hinted at a couple of these one are like the GPS that you can use with open AI, right, where you can start building a lot of these. You mentioned AI and what have you built in it? I have not one No, it's
Noah Kravitz 44:41
good ask. Yeah, I know. Be AI guy but I just haven't gotten my hands. You know, time to do it. Like I'm saying, Yeah,
William Harris 44:47
this is yeah, this is q3 q4, Black Friday, Cyber Monday run up here. If I could build an AI to sleep for me, that's what I would do.
Noah Kravitz 44:56
Is it too late to build a Black Friday fry? Now I want to make a bad pun. Yeah. Ai, you know, like fry
William Harris 45:05
AI or Black Friday? I
Noah Kravitz 45:08
mean, they just keep me away from spending money. Yeah. Oh
William Harris 45:11
my gosh, right. So then the other thing though, too, is meta has some that they released, I know that the universe of characters at your fingertips that they had, right where it's like Chris Paul pro golfer helping you perfect your stroke, like, like they built some of their own versions of these, like miniature ones. But I still get mostly hung up on that idea of, let's say, as we transition into, like the longer term vision and future of AI, where it goes towards like this super agent of sorts, where it's like, it's your agent, optimizing, like, what you want and what you desire in all of these facets, and it doesn't necessarily have to be broken down. And so but I'd say like moving beyond the agent side, like, what about the future of AI gets you the most excited when you think about like where this is headed, and this could be 50 years from now or longer.
Noah Kravitz 46:05
So, several years ago, I actually used to look this up the other day for something and it's longer ago than I realized. And I actually had twice, I had the good fortune of interviewing a gentleman named Kai Fuli for the NVIDIA pod. The first time was probably, you know, 2019 ish. And then a couple years later, he came back. And so he, he's been in the field for a long, long time at very high levels. I think this is correct. So I'm going back to my disclaimers. I believe he led Google China for a long time. And then he started his own venture capital fund. But he's been in AI for a long time. And he's an optimist about this stuff. And so the first time he came on, he was promoting a book called AI superpowers. And then the second time was another book that was interesting collaboration with a fiction writer. But the first time as superpowers kind of about, like, just things related to AI. And this was still several years before, you know, we were doing the pod, right, and Vidya you know, but like, most of the world wasn't talking about AI, except for the occasional movie or whatever. Right. And so he's been in this for years, and, you know, in significant ways. And one of the things he talked about that I hadn't really heard before was this idea that, I mean, I just haven't heard it, but it's been around this idea that AI has, like, legit has the capability to help humans enable humans to usher in an age of abundance for humanity. So that like, like, the technology itself, and the applications in, like, the world that we've built, right? Like, yes, they're sort of the like, what if AI can figure out how to build a nano replicating machine? You know, there's that, right? And that's not necessarily impossible, right? But even like, without going there, right, like, how, you know, without, like, starting all over on Mars, or whatever, like how, given where we're at right now, could AI like, legit, you know, and sort of laying it out, right? And that kind of like, like, opened the door for me, because, you know, I've always been in the technology. But my deliverable, so to speak has always been either content, like writing video, I used to do a lot of videos back in the day podcasting now. And then I've been involved kind of more in my own entrepreneurial ventures, but a couple of times in, you know, real jobs. Now, that's not in jobs working for other companies, where I was kind of more on like a producer, you know, type role. And in those roles, my tech, I've always had tech skills that top out or on prototyping, right. And so, so, working with Nvidia for several years before the podcast opportunity came along, because they started the podcast with somebody else. And then that person laughed. This was during the first year of the pod. And so they then asked me if I was interested, and I kind of took over from that and they've had some other people host some episodes since then they had somebody do a lot of the auto related stuff in house, whatever but but anyway, my work is in video prior to that had been writing about Like, pretty, non mainstream things that the mainstream doesn't think and talk about, like, like surveillance, surveillance cameras, security cameras, traffic, camera light, traffic, light cameras, that kind of stuff, which, like, was already using machine learning to do things right? Early, you know, delivery robots, and like, we're not drones, you know, people have drones now. But like, warehouse drones that were using computer vision cameras, again, an early form of machine learning, or not early, but I just mean, predates what we're talking about now. You know, so. And then NVIDIA also had its hands, obviously, in gaming, but also a lot in sciences, because of medical imaging, and all the data required for him in videos, prowers with graphics and all that. And so I got to kind of like, occasionally work on, or at least just hear about, even as a contract when I was in full time that I went to some events and you know, whatever, and see some of the wild stuff that you know, people were using in video tech to do in life sciences and neuro, you know, brain research and neuroscience, which is my dad's field, and, you know, space exploration, we've had some, some astronomers come on the pod and talk about the James Webb telescope. So all these different things, right. And so like, I think that listening to Kai Fuli talks about this idea, and then going back to his book, and then reading other things, you know, having that background and sort of understanding that like, yeah, like, things are already happening, right. And like, you know, I think COVID came, right. And AI was such a big part of, and again, I don't mean to, AI has become the sort of I'm trying to think of nice way to say it, umbrella term for lots of things, right. But you know, that like that, that it was such a big part of developing, I mean, COVID vaccines and treatments, but it's been such a big part of drug discovery, all these things that affect lives in material ways, right. And so then, you know, carefully kind of, like, that experience, sort of making me think about like, yeah, like, if we diverted, you know, more resources towards, and then this gets into, like, politics and all that kind of stuff, and just, you know, human nature, et cetera, et cetera. I don't mean to brush that off as insignificant. But you know, like, actually, yeah, like, if you think about it, there is sort of a science based path towards, like, abundance, for lack of a better term, meaningful, you know, for for all people, right. So the idea that, like, big, positive, world shaping things are more possible now, because of AI, and not just because of AI, but are more possible now than they were, you know, five years ago, 10 years ago, certainly 20, probably two, you know, whatever. I like that, that gets me really excited. And then I think kind of near term. It's ironic, because I thought, this would be like, sort of a banner year sort of professionally, for me because of AI. And it's been really interesting and educational, which are always code words for you. The bottom line is not quite my thought was gonna be sure, I think there are a lot of opportunities for you know, I it's marketing speak, but it's true, like smaller teams to do more. And whatever that means to you and your priorities. And if that means, like, cranking out more quality blog posts, if that means like, you know, the numbers are running in the background, you don't have to hire an accountant, or whatever it means, right? Like, like, the way I like to think about it is that if it can be expressed as data, AI, theoretically, can do something positive with it. Right? And so like that, I mean, that just as big as it sounds, I can't Yeah, right. I mean, it just as big as it sounds, and then it's also not quite that easy. And so or I should say it's not as easy as flipping a switch. There's work to be done to get there and so be anything of this year I thought, I'm going to learn how to use. I'm gonna do all this research and I'm going to pick the AI writing tool, my favorite AI writing tool. So I tried like Jasper and writers write sidekick, I think some others. And then this is going to help me scale my because I earning my living mainly as a freelance writer for the past, you know, 10 years, whatever, doing other things on the side podcasts, obviously, but but that's kind of my bread and butter. I'm going to use this to scale my business. And I'm just gonna, like get because it's getting harder. I've been doing it for so long and experience level. And, and I you know, where I live, like expenses are high, but obviously like, you know, salaries compensate whatever, but there's not a lot of room to charge more per hour, kind of. And it's more about scale. Right? You know, you're, I'm sure you and a lot of your guests are way more expert on that than I am right. But one man service business, you can't, you know, but oh AI, and I can just set up, you know, set up an AI, so to speak that rights like me, are actually set up multiple ones for multiple clients and train it do all this stuff. And it wasn't, it wasn't that easy. It's never going to be that easy. But it's closer yesterday. I actually yesterday when I got the feedback, and it was positive feedback on an article that I wrote for a client recently, that took me way less time than it would have before. And I was just using free GPT three, five, I wasn't using the paid GPT. And it didn't write the article for me, right. It's like, it's been enough time. And this is what it's good at. I know, some I know, my workflow, I know the parts of my workflow that take time and other parts of my workflow that I hate. And so it takes time because I avoid doing it. Right. I know, like, and I know, the parts of my workflow that like can be automated and can. And I never said it in those words before. But these are the words to use when you're thinking about AI and AI, right. And so for me, it's like I have an assigned a writing assignment. Sometimes there are one or more transcripts to work with, sometimes there are one or more other primary source documents to work with, you know, websites or PDFs or whatever. Sometimes there's nothing and I just have to, like, do the research, or maybe I know stuff about I'm the SME, whatever. So I sort of know, in my workflow, right, and so I've been figuring out this year, like different tools kind of patched together to, you know, summarize transcripts, or generate, you know, five, blah, blah, blah, on blah, blah, blah, for those kind of listicle articles or whatever. And yeah, so I didn't, I didn't at the beginning, it wasn't like AI, I'm scaling, you know, stacks of money everywhere. They work either way. But now I am at the point where actually like, day to day, and I've built built, but I did you know, some stuff using other people's tools, and really just kind of like, whether it was sort of behind the scenes prompting and customer instruction, you know, talking about the TPTs, or some other, you know, tools built on top of MLMs that offer customization and I kind of tweaked them, tweak them to help with workflows that I'm familiar with. And so So the interesting thing about this is that, and the thing that kind of sparked this train of thought was that you mentioned the consulting thing that I'm doing now called AI ready, which is, you know, working with businesses to help them get started using, you know, get ready to use AI. And so our initial sort of premise without getting into the whole backstory of it, but our initial premise was, it was sort of like, there's a gig that somebody needs done. And oh, that's the kind of gig that I've been looking for, you know, I think I said the person needs to know, it'll make sense and product market fit and all that good stuff, right. So we need some sort of offering to like, go to people with the biggest takeaway from this experience, you know, this year getting this consulting thing going. And it hasn't been all year and it hasn't been full time. But still, it's been enough time. I think the term would be pre sales, education. There's so much pre sales education that we've been doing with, like, you know, decision makers at like, you know, companies that have and there's nothing wrong with companies that have less than 20 to 50 people I've you know, I've been a couple have one or two most of my life, but companies that like there's money, and there's jobs, and there's other people's livelihoods on the line, and there's revenue, and there's all this. And they're also like, you know, like, most companies these days, like, pretty digitally sent centric, right? Like, either their, their product offering is some sort of web platform or whatever, right. But they're not doing anything with AI internally. They don't even really know they haven't spent time thinking about it. They sort of know, maybe they have, like, yeah, I've got an engineer who, you know, has a co pilot subscription that we reimburse her for, you know, like, we got a couple of power users, but we just let them do their thing. Or yet Nah, man, like, you know, corporate IT and legal, we're like, no, GPT in our house, you know, whatever. So there's a lot, I'm excited. I'm excited about the like, you know, maybe we have a chance at everybody having a good life, right, we said that big picture possibility. But nearer term, there's so much that so many people could be doing big and small, particularly in the workplace, to just to like make the grunt work better somehow. And then on the other end of the spectrum, I find, you know, AI stuff to be like a good sort of brainstorming partner. So kind of short term, there's a ton of possibility as well. I
William Harris 1:01:29
like where you went with that. Just like you said, the short term productivity possibilities in the long term, this idea of not necessarily utopia, but a much better version of existence for where we're at. Couple of just oddball ideas that I've I've heard, one that I think my dad said that I really appreciate. It's just the idea where it's like, I don't know how practical This is. But I think that the reality of it is right around the corner, where you could go and talk to grandpa, right? Grandpa's day, but you've got like, you've got aI version of grandpa, and it's almost indistinguishable from from real grandpa. And so you can actually just go walk into grandpa's house, talk to grandpa interact with him ask him for advice, like, what? What would you do in this situation? You know, similar to maybe almost like that Amazon Prime TV series called upload? I don't know if you've seen it, but it's, I haven't
Noah Kravitz 1:02:23
watched it better. Oh, no, we did watch it. Yeah. Yeah.
William Harris 1:02:26
Yeah. So it's just that same concept. But I think that's a very interesting idea of like, where we're going from pros and cons to all of this. Right. I think that's one thing that I want to call out that it's like, even the idea of the increased productivity from smaller teams. The con to that is that there's job loss, potentially, or at least people haven't figured out transitions and adding, you know, like you said, maybe we get too much into the politics side of it. But then it's like, well, then is there a bigger need for you know, ubi, universal basic income then in a situation like that. And so while a can of worms to discuss, like, but I think that we're
Noah Kravitz 1:03:00
here, Cannon, I don't know what the purview of the you know, I'll follow your lead on it. I didn't mean to say, we can't go political. I just, I just didn't want to get derailed there.
William Harris 1:03:11
But you know, yeah. And I mean, we absolutely could it I just mean but like, in general, I mean, like, as a society, like, these are the conversations that I think our 100% are imminent to figure out. And I know people have talked about it a little bit. One more, one more positive spin on this, or maybe I don't know, it's not, but then I want to get into like, what are the things you're afraid of with AI? But like, what do you think about even, like, neural link in like, are you even able to comment on like, but like, what Elon Musk is doing with neural link and how that integrates into this idea of AI.
Noah Kravitz 1:03:45
Um, I don't know, a ton about neuro link. Other than that, it's, you know, brain implants and AI. And, you know, who the founder is. So, no, so I just, I don't want to speak specifically to everything I can speak more broadly, to, I mean, I could speak about brain implants actually, but I can speak kind of more broadly, I sort of just the idea of like, you know cyborg Man Machine enhanced Johnny Mnemonic, Johnny Mnemonic, right? totally right. I mean, I think what was it you were just talking about before that though? Oh, about talking to grandpa right? So I think with all of this stuff, it's good to remember that none of these ideas are new. I don't think the idea of preserving grandpa's memory you know written words and and photographs and audio conversations and videotape. So you know, so AI AI. AI is different in ways both, I think, based in science and based in sort of humanities tendency to create stories slash smuggle things with magic, which I don't think is a bad 10 Good or bad. It's just what we do. Right? So I think like AI has been imbued with some magical qualities that maybe it doesn't have. And I'm the guy who just said, like, it can help lead us to a world of abundance. Right? So you know, I, but like I, for instance, don't think there's a ghost in the machine, at least not in, you know, I don't think chat GPT or Google Bard, or whatever it was. shows signs of sentience for Yeah,
William Harris 1:06:02
not yet. I don't think we're there yet. Yeah. You know, so
Noah Kravitz 1:06:05
AI has got this thing where like, the idea of preserving grandpa, as an AI you can talk to, is both technologically different from a videotape that you can watch, right? Even if it has like grandpa's likeness, you know. But the interactivity and just the the mystique around AI and everything, right. But at the same time, when you were saying talking about that I was thinking about, like, I know, I know, some folks working on companies using AI to do the kind of thing you were just talking about. And I also was talking to a friend the other day, who's like a diehard Apple fan, and he is really excited for the the vision Pro, the
William Harris 1:06:50
only for that, yes, yeah. Yeah.
Noah Kravitz 1:06:53
So he's, like, mostly excited about it for I mean, you know, just all the geeky reasons and everything, but like, the sort of spatial computing stuff he's super excited about, right. But he also, whatever phone you need to have to record spatial video. And I don't know, if it's just the newest one, I used to be up on all this. I used to cover cell phones for a living and now Yeah, I got like a 13 Mini or something. But whatever, whatever. He has a phone that's now capable of recording. I think they're calling it spatial video. It's Apple's new format that is compatible with and so they just recently showed off. I don't know if they just described it in words. Oh, no, some journalists got to go see it. And then the journalist wrote about their experiences. And they said that spatial videos like wild, it's like, somehow, you know, familiar and new all at once, and blah, blah, blah, but that my friend, so what the journalist imagine, though, was like, record everything in spatial video, because as memories, whether the people that are alive or not, this is the format I want. Because when I can view it on the vision pro in, you know, whatever they're calling their, their mixed reality environment, like, you know, 30 minutes with it was enough, like, this is what I want, right? And so my friend told me, we were talking about the other day, and he was like, Yeah, I started recording videos of my 15 year old cat, which I'm not a cat person, but that's like a really old cat, you know, and so, right so you know, so that tendency that wanting to preserve and and you know, all those things are part of us. But AI applying AI to it isn't new AI is the newest thing. So yeah, there are a lot of I mean, I have never I have never been a big fan of the idea of work as the dominant activity around which people's lives are organized. That sounds like a really highfalutin way of saying I'm lazy and there's some truth to be fair, I think what I mean is like not work but like Purdue working to create revenue right? Like whatever the right way to express that is right. I don't have I don't have a good answer for it. You know, I'm staring out my window at the trees think what's the answer? But part of the allure of AI and this allure of like the age of abundance thing, right is that if we can get if this view that like AI is best suited for a lot of this grunt work stuff, right. And earlier in the year when I was nice that earlier in the eyes haven't had this kind of conversation for a while, but I was like I was keen on these metaphors of like the orchestra conductor, for what creative type roles, like having worked a lot in marketing, right? Being a writer, and so you know, you have the like, you know, copywriter role or a creative directors and the visual on the writing side or project me or whatever, right? And thinking about like, if the AI can do we did a pod a long time ago, my partner and AI ready likes to shout out to Neil. Neil likes to say like you were talking about generative AI six years ago, a decade ago, 50 years ago, you know, I talked to a guy who talked about it five, six years ago, let me say that, but you know, he was a graphic artist and a VFX guy on video games, and talking about, or an art director, but talking about, like, at that point, he was using AI to do things like he would create a zombie or a couple of zombies, and generative AI would fill out the rest of the zombie army. Right, it was really good for landscapes and that kind of stuff, right. And so the sort of like the conversation we had, it's a long time ago, but I think it was a great episode, because it's making me think of these things now. And they didn't come from me, they came from him. The idea that like, as a graphic designer, or a creative director, like you can use AI kind of in the way that you would use a team of junior artists to do all of that work for you, right? So you've got this dichotomy that you you refer to before. On the one hand, one person can do so much more, right. And you can look at that in different ways you can look at that. It's just like, raw efficiency and productivity. But for me, I might look at it as I can write, I can't draw to save my life, let alone paint, you know, let alone like, I started using Photoshop, and I go into a rabbit hole, and 90 minutes later, that one tiny part of the image I was messing with looks horrible. Right? So maybe, you know, maybe generative AI could help me flesh out the visual parts of, you know, my stuff. So that's super cool. On the other hand, you know, you're talking about job loss, potentially, and you're thinking about, like, as somebody myself, who is both like, self employed and a cog in the machine, I guess we all are, to an extent, but, you know, like I haven't had the percentage of my working life that's been in full time jobs is comparatively quite small to a lot of people, I think, really people my age, I think it's more common these days. Excuse me. So, you know, I can see how I mean, as a consultant, right, I'm sort of like, what are the metrics that we could sell our service use this audit services, these businesses? Well, you know, you can save money? Well, how can you save money, while instead of paying a human, or, you know, instead of hiring a human, hire us for less, really, we're just going to help you use AI. So you know, I get it right. I also beginning of this year, hit a point this year, that will, you know, without getting into it. Contract and full time job availability in I don't know, the tip, people say the tech sector, I'll just say, in the places I've been working for the past 20 years, has been tight for the past year and a half, two years, whatever, right? That's not news to people. I just don't want to presume anything about anywhere else. I'm in a bubble. You know, when we're on our own bubble to some extent, I don't know mine is what it is, right? But works been work has been hard to come by. Right. And so I have myself gone through the cycles of letter saying like, I'm gonna get three times as many gigs writing and do them in a third of the time because AI is my buddy, you know, and then like AI is taking all the jobs, you know, and then it's kind of like, well, AI is not actually taking any jobs yet. But because people are convinced AI is going to take you know, who was the there was a tech leader who got a lot of press like a tech leader of a large company, I forget which company who got a lot of press for talking about like in the midst of all the layoffs and hiring freezes. We're not going to rehire until we assess, you know how much we can replace with AI. So we know this stuff is real. And even if it didn't come, even if AI didn't come for all of the knowledge worker jobs in 2023 Like some people, myself included, were At least, you know, reasonably worried about. It came for some. And I think a better way of talking about it than it came for is it has already started to significantly change the way that people work at organizations of you know, from one to a billion. Right. And not only is that not showing any signs of slowing down any soon, it's moving, or I should say, of stopping any soon, it's not showing any signs of slowing down, and it's moving at breakneck speed. So yeah, I am a fan in theory of I mean, I'm a fan of theory of UBI. But I think what I think UBI should be is a lot higher than what we're talking about. I don't know how realistic, realistic any of it is. I'm not an economist, you know. But yeah, I mean, I think I don't know, I, I fully understand that. There's new tech hype every day. And you know, all that kind of stuff. But there's, there's some pretty pretty intense stuff happening with, you know, really big implications. And yeah, there's theoretically a way of steering it. This isn't about the technology, this is about people. And this is about to help people, you know, come together on everything from what to prioritize, to how to, you know, should a phone have lightning or USB C on it, right. Like, we disagree about stuff. Sure time. And so it seems very, very far away. And then it gets back to things like well, you know, think globally, act locally, and et cetera, et cetera. Like, the impact that the smartphone has had on the world, right, like nobody really thought that I was listening to some. There's a pot, there's a guy named Ben Thompson, who writes a blog called street Tegrity. And he has, he has had different podcasts over the years, there was a strict equity podcast. Now, I think he's got a little like, mini, elegant network, little mini empire of different things. And one of them is called sharp tech, because his co host is a guy named Andrew sharp. And I used to be a big street equity, regular, it's business. And it's like strategy, business strategy and technology. And so for me, it was a great intersection of like, the tech stuff that I keep up on all the time, and the business angles that I, you know, aren't my wheelhouse. And so anyway, I listened to hadn't listened to him in a while. And I listened to a recent episode that was about the new open AI, the dev day announcements, and, you know, the future of of chat bots, and all lambs and all this kind of stuff. And in it, and it was based off of an article that he had written. So somewhere maybe in the article, actually, he quoted Bill Gates from a long time ago, talking about why Windows Phone or Windows Mobile, I think it was after the fact about why they lost. And so sort of like, I mean, that's a great if you're interested in tech, and the history of tech and business and all of this stuff. Great thing to go back to seminal moment. But the sort of broader thing I'm thinking about is how like, nobody necessarily, oh, right, Gates was talking about how he imagined a future where we will carry multiple devices, right? own laptop, or tablet or something. And then like, your main computer, you know, whether it's a desktop or a laptop, you're like, heavy duty sit down and computer. And he was like, the phone is not going to be the primary device. And as it's turned out, the phone is the primary device for so many people in the world. And some it's because of financial limitations. And some it's because of like, they're just mobile, you know, people just don't want to go and you just do it all, but whatever the reason is, right. And so like, I'm pretty convinced that and in the same way, the internet, you know, maybe we saw some things coming, right and maybe the.com, the first.com Boom, saw things coming in that we just kind of weren't the infrastructure wasn't there for yet or whatever. But I don't know I think about like, in the same way that the smartphone has given so many people access to it. We all have all of this knowledge that they couldn't have access to otherwise, like, throw AI on there, you know, and it's hard to imagine where we're going to go. But the upside is huge. There's a lot of downside related to AI. And otherwise, you know, and a lot it gets back to things that might prevent us from getting to that. Sure. You know, it's not utopia, but it's real nice standing place. Yeah, you know, there's a lot a lot of fun upside.
William Harris 1:20:33
So let's, let's run with this. upside and where we're going with this. And let's say, practically speaking, knowing what you know, about AI, as you talk to so many people and where it's going, how would you practically begin to help your kids navigate the very imminent future that they're going into? Because there are a lot of jobs that maybe just don't make sense to go into anymore? Yeah. Aside from telling them to start playing with it, and using it, which I've got three, three daughters, 1310, and eight. And I've had them all testing out, like mid journey in GBT. And they're they're writing their own stories with it, it's a lot of fun to see the kind of stuff that they come up with. But, you know, aside from that, like, how, how can we practically tell the next generation to begin navigating? What this world looks like?
Noah Kravitz 1:21:33
It's good question. I mean, I think so in some respects, it's similar to what I say to, you know, business clients, or professional people I talked to, which is what you just said, which is like, start, just go use it. Right, right. And the thing that made Chad GPT is such a such a success, and then sort of sparked this whole boom that we're in, is that it's so easy to use. Right? And does that mean, you're going to get exactly the results you want the first time? Like, you know, no, I was almost gonna say like Rosie, the robot on the Jetsons the old show? And then I thought, Well, no, actually you are. Because in order to get something from Rosa, you have to like, tell her what to do, or ask her what to do be nice and ask. It's the same thing with the Chatbot. Just like asking for something. And then you get into a new start using it and you're like, oh, my gosh, like, you know, my, my dad, who I referenced before, is a retired psychologist, and his lot of his career was spent researching the brain and working with patients with chronic pain from different stemming from different different places and events and such, and sort of trying to, you know, both learn about how the brain deals with pain, but then apply that to, you know, working with people and helping people. And so a lot of therapy type, you know, work. And the first time I showed him, Chet GPT, I showed it to him in the context of, you know, having instructed it to act like a therapist, and he was just like, can you send, like, send me this link? What is this, you know, like, this crazy, and people have that reaction. And I think that, and I'm not saying it's the end all be all or whatever, and but like, that's the place to start, right? I think the other place to start, which is much harder, is like it's one of these things that I feel like, it's easy for me to say, and I don't know how good of a job I've done it in my own life. But it's still worth saying, because it's the thing is to focus on, like, what do you want the world to be like, for you, and for everyone else, and everything else that you share it with? And I don't mean like, you know, go as chat GPT although you could, like I've asked for like what, you know, what would a world look like that was blah, blah, right? And really cool is to ask, you know, there's been all over socials, but ask one of the image creation things to generate what that word looks like, you know, and it's funny how the images tend towards a certain I don't know, I think of it as sort of like mid 60s mid century modern space age sort of, you know, sure. Maybe the way I prompted it, but like for outdoor want to say it being somebody who my own career path like has not been, I mean, there are threads there are strong threads through it, but it's not been like that. up jobs, I spend a lot on plays, right? And that's just how I am and how my brain is now, my personality and all that kind of stuff. I sometimes feel like it's hypocritical almost, of me to talk about, like, find your passion and think about how you can use it to help the world and go do that. Because I'm not an engineer, or a scientist or, you know, lawyer or fighting on behalf of the underrepresented, or you know, what I mean? Like, whoever right, like, I'm just sort of who I am. But in answering the question about I am, who I am sounded, that's not what I wanted to say, I am, who I am and who I am as somebody who like, primarily is a writer, and observer and analyzer and delivers, you know, content. So, right, I can't go out and build a better, you know, solar power geothermally, and just shared mousetrap. It's not gonna come home, I'm gonna snap my own finger off. But I do think if we're talking about AI, like pragmatically speaking, sort of, you know, what's the thing called Maslow's hierarchy of needs? I think, right. And so if you're talking about like, we're talking about like, our kids, right. And I have kids around similar age to yours ones, older ones, a younger college. I don't, I mean, I went to college, I went to graduate school. wasn't that long ago, but you know, 20 to 3020 25 years ago, 25 to 30, whatever it is. And it wasn't cheap back then. But it's like, you know, more now. So you can look at it sort of pragmatically, financially, right? Like, what's your return going to be? And that's so talking about Maslow's hierarchy, being able to pay the bills, unfortunately, is, you know, that base level need, right, and that gets back to my whole UBI. And when I actually think about all that, and like, just redistribute everything, so we all have enough, but I don't know human nature, and we wind up just finding our way back to, like, freshman philosophy in five minutes, right there. But, you know, you can use AI, like, like learning the skills, and everything will help you get jobs in the future, even if there's only six jobs, right. Being able to interact with AI will be a prerequisite for all six of them. You know what I mean? Like, I don't see it going, it's like the internet in that respect, right? Like, you know, beyond that, A, it's really hard to say, Be, it depends what kind of work you're doing. And so then that just gets back to the like, it's a tool, it's not really all that different. Yeah, it could make things super different. Before you and I are dead, you know, like, because things are moving that fast. And like I was talking about before, if you want to get into sort of the like nuanced stuff. I haven't seen a lot of these jobs myself yet, like content. And marketing, again, is kind of my area that I'm more familiar with. I was on a panel a couple of weeks ago, marketing and AI. So it was marketing specific. But there was somebody on the panel, who worked at Salesforce where I've done a lot of work over the years, but I'm not currently. So I don't know what their whole situation is. And her title. And it turned out she had just gotten or watched your session, and she just gotten this title, like in the past few months. Her title was like, marking it like, like, marketing AI ops manager or something along those lines, right. And I was mentioning, before that I built some tools that I could use in my workflows or had used in my previous workflows. One of them was like, take this raw piece of video and turn it into a piece of content, you know. And then, you know, get it all the parameters of the sort that I would use in this bullet. All right. And I showed it to one of my former colleagues who's still at the company, and you know, and they were like, we had this conversation that net results that we had this conversation around how like there's so much potential and there are people who are aware of the potential and there are people who are not, and it's also about economy, and it's also like, internal priorities already, and it's also like, huge companies, you don't just like turn the ship suddenly, you know, and all that stuff. kind of goes into light All these jobs have the sort of know the integration of these tools in the workplace. A lot has been happening on the download for years, as we've been talking about before, like, you know, especially with data and number crunching and things like that. There have been algorithms in the backroom, random closing things, autocomplete, autocomplete, turning into auto completing your phrase, your sentence, you know, right, those are sort of step going, same progression. So you learn how to use those tools. And the jobs that like specifically are, you are there to figure out something with these tools, those jobs are starting to come, they're not there yet. And so like, the idea of, you know, the writer becoming more of a creative, or the marketing person becoming, I showed one of the demos I built to a friend at another company. And they wrote back, and they're kind of like this, but a grain of salt. But they wrote back and said, You know, I see this eliminating the role of writer at our company. The downside is that none of the marketing people I work with, like could tell a story to save, save their lives, you know, but that's the thing, right? Do I need to be a writer? Do I need to be an illustrator? Do I need to be a voiceover artist? Well, yeah, one of those would probably be like, expertise is always really good. But really, you need to be a storyteller. And if you can have a be a storyteller, who knows how to use the tool of the day. And if the tool today is a prompt based AI system that today can spit back text, tomorrow can be multimodal, like we were talking about, and then maybe next week is actually integrated into, you know, Adobe suite of tools, or, you know, whoever, Shopify suite of tools, right, all these things, right? Then that's where we're headed is that whole Asian thing of like, as the human, you're sort of the super agent, and you tell the agent what to do. And then it performs as a junior copywriter, or a senior interaction designer or, you know, 25 time CFO with a year, whatever you needed to do? Yeah, so you know, the advice isn't that different? Really,
William Harris 1:32:22
it's good, though, it is good advice. And I would say, you know, one other practical thing that I would give to people to have you heard of a tool called try aptx.ai. It's a TR y, a PT. It's essentially, like helps you figure out what is your ideal job, and actually, I test it out, but it's using AI, and it runs through all of the different main personality tests. You've got like the Myers Briggs. Okay, yeah. So so it goes through all of those. And then it asks you a bunch of questions also about like, the things that you enjoy, and it, it pulls us all together. And it suggests jobs, that would be ideal for you based on all of these different factors. And it's wildly accurate. I will say that I tried it for myself, and everything that it came up with, it's like yeah, that's that's where I'm at and where I'm aspiring to be. Try it with my wife. I had to laugh because the first result, it showed up, she's a very creative person. Well, the most creative people I've ever met, and would not be so I'm ADHD. She's ADHD where the different type of ADHD though where it's like she can Her phone is for four days. Yeah, she can, she can do what for four does not know where her phone is for four days, kind of, like like that, like my ADHD does not allow that to happen. So but I remember when it showed up, her results came up with like, project manager, software developer and I'm like, I think this is broken, babe. Like, I don't think that's it at all. There was just like a glitch, it was just taking a while to load her actual results. And then it gave her much more but it was wildly relevant. I've sent a bunch of friends to it and they're all feeling very good about like results I'm thinking this is a good idea in a way to use kind of like these different pieces to practically start figuring out like what is and they've actually done some work to eliminate from the job recommendations jobs that they feel are very likely to be more replaced or something along those lines and so it's like okay, how do we actually just start guiding people towards practical things?
Noah Kravitz 1:34:26
I mean, you know, so there's this I was looking at apt apt in the background and it's funny because it reminded me of like three different things from Sure. That really just reminded me of how old I am thinking about like, where do the questions on something like apt come from and creating content and AI's role and all that kind of stuff, right, but in the in the in this context, and we're like, Well, you know, How do I know to trust this thing? Like who wrote the questions and who's grading the questions and what how's the algorithm biased? Just made me think about, you know, in in AI, there's this expression called human in the loop. Which I don't know if it's, it comes from AIS originally, or more computing in general, or something else. I could look it up now, but that'd be rude. We're talking al listeners can look it up, I'll look it up later. But it's one of my favorite expressions, because what it means is, you know, refers to the notion of keeping if the, if the automated system is doing something in a loop, right, whatever steps are in there, they're being automated, one of them has to be human oversight of some sort. And, you know, you can extrapolate that out, and how often and what was a human do blah, blah, but it you know, sort of colloquially it, you know, kind of has come to mean, or I've come to use it as like, whenever we're talking about AI, whether we're talking about, you know, on the practitioner level, and you're talking about training and inference on a level and understand, or we're talking very, very broadly about like, are the robots coming to kill us or not? Like, the idea you have to keep in mind is that there's a human in the loop somewhere. And and so the bet it's a best practice right now is like having a human in the loop as a prospect, sending an AI off to just go do stuff on supervised is not a good idea. Maybe there's somebody very technical out there who's hearing me say that who's like no, actually like, okay, cool. It's not a good idea. Like, hey, write Microsoft chat bot, that they released the internet trying to be a Nazi, and three hours or whatever it was right now. Right? Now unsupervised? Yeah. It's like tweeting. It's like tweeting, don't tweet. And don't, don't, you know, sends a good hint
William Harris 1:37:00
towards something coming up here, too.
Noah Kravitz 1:37:04
Yeah, it's not tweet anymore, it's x or don't do access.
William Harris 1:37:09
It's also true.
Noah Kravitz 1:37:12
But so so all of the let you know, the people who I've the people actually know what they're talking about, eg the people I've interviewed, right, it comes up again and again, that, that it's not just that we're not at a place the AI is not sophisticated enough. But it's also just like, you know, what are we doing this for? Right? And, and, at the end of the day, whether you think of yourself as very self centered, or very unselfish, you know, unselfish, or if you think of, you know, the world is a human centric place, or we're just a speck of dust along with, you know, the deer in the canyon behind my house, and, you know, the whenever else, right? Like, we're humans. And if we're creating something, there's some sort of human interest in agency, right in it. And so keeping that, in, I feel like I lived in California for too long now. And I'm sounding like that, but I think I'm just trying to get it like keeping keeping humans in the loop. Whether it's like, you know, just checking in once in a while to make sure that like, that content marketing flywheel you set up is, you know, not only like not spitting toxic things out there, but also performing the way that you want it to. That's important, right? Like, like, having human interests broadly, infused into what we're doing is important. And so like, if you're just out here trying to make rent, and you don't literally don't have bandwidth, to go, you know, feed the hungry, because you're amongst the hungry. 100%. Right. But in the more collective thing, talking bigger picture about, you know, what, why are we doing, you know, AI and best practices and technology and all these kinds of things. I mean, that's the thing I keep coming back to is just sort of like, I always used to imagine that I always used to kind of want to hit it big, right? I mean, I still want to have bigger ones. Sure. But I used to sort of imagine because I've had this sort of career path where like, I've worked with some big companies. And I've lived in, you know, the Silicon Valley, San Francisco area for a long time, and I'm not complaining, I've done a lot of really cool stuff. It's not that, but I haven't had like, I haven't either had sort of a long term career at a company that took off, you know, didn't work at a startup that took off and then found my own thing that like took off big time. So I didn't like you know, maybe one time a long time ago, very randomly when I was younger and used to have like random nights out that would wind up in Yeah, now like sketchy places, so which is just random places, I wound up at this bar in San Francisco. And I mean, this is like 20 years, a long time ago. And I went to go, I was at a whole bunch of people. And it was like the last stop and kind of last night and I went to go get another beer. And I was like, out of money. Like I, you know, I didn't realize I had like, two bucks in my wallet or whatever. And this guy is sitting at the bar by himself bought my beer for me, you know, as like, a nice, thank, you know, blah, blah. And I was like, happy drunk at the time, you know, and this guy was not used by himself and not like, whatever. And so we chatted for a minute. And he told me that he was like a tech millionaire, who, you know, had made it big early, and sold his company or the company, he's involved, whatever. And now he's kind of like this young, lonely guy who lives in San Francisco, and he's, you know, whatever, however old he was in his late 20s, early 30s, and had like, made it financially and whatever, but like, so, you know, money is not the end all be all, but it is the base of the pyramid and everything. And I think that a, I'm repeating myself, I think, I think it's the portion of the podcast from no repeats himself. But I think maybe it doesn't happen, I maybe it doesn't happen. I used to have the fantasies around. If I did make it big, I've started a couple of companies back in the day. And if you know, you start it, and you're imagining, like what you know, and we can get to like, whatever valuation and my my shirt at the time is 51%. If we can get to like, you know, 10 million, and then I can sell and I have 5 million, can I retire at age 32? With 5 million? I don't know, but I think I can. And that's the goal, you know, whatever. That never happened for me, right? But I used to fantasize, well, if it did, what would I do? And I used to like to think that I would spend a lot of my time Well, I used to like to think I would start a band. And you know, we tore for the rest of my life. Because I wouldn't need to worry about getting pay, I could pay my friends to be in a band with me. But I used to think like to think that I would do something like, do glittery on a grand scale, you know what I mean? And I could just devote my life, you know, whatever. And it's kind of like, I might still get there in my lifetime, who knows, I might get far enough in my lifetime to be able to help my kids get there in their lifetimes, you know, Senator, so early now, AI is kind of like, I think about it in the same way that like, okay, it might help me, like, scale up my writing business instantly. And then I don't have to like work anymore, I just come downstairs, hit a button and drink my coffee. That didn't happen. But actually, over time, I did figure out how to use it. And also the technology advanced, you know, but figured out how to use it in a way that just this past week, helped me write a draft and an ad and do it for me, it was part of my it was a tool in my workflow helped me crank out a draft, not draft, I submitted the RMS first draft, but a final work product Sure, way faster than I could have without it. And the feedback I got was, you know, was great. So hey, that's a step on that path. Right. And so in the same way, it's like, short term, of course, people are figuring out ways to make money with AI to make more money for themselves and their small business, to have their team at work, you know, produce better, so they can get an extra headcount next quarter, maybe to like figure out how to replace the whole HR function at you know, whatever giant tech company, it was, so that shareholders profit more, right, of course, that's what we're all doing. I just hope that somehow given all of the things going on in the world, good, bad and otherwise, that we have, we can all sort of get along well enough for long enough and direct enough of our resources and priorities and attention towards, you know, sort of what's possible, right? And I think in that way, like, I might not hit it rich and retire early. But, you know, I'd still look pretty good for having a pretty good life. And then like seeing my kids go further than I did. You know, it sort of feels like that with AI like it. It may not solve all the world's problems immediately. But if you're an optimist generally, and you're optimistic about humanity, and you're optimistic about the fate of I don't know if it's the world or earth or Mars or the Milky Way You don't know We like if you're optimistic. I think there's a lot to be optimistic about, you know, that AI is and will enable. Yeah.
William Harris 1:45:11
I want to be respectful of your time. We are, we aren't at our
Noah Kravitz 1:45:16
time, or you were either the shadows are falling. Well, I was gonna
William Harris 1:45:20
call over that one in a second, too. But I, there's at least one more thing that I want to talk with you about, if you've got a little bit of time here. Sure. But I did want to Yeah, I did want to acknowledge that for anybody who's watching this. I don't think that I've done this during like, daylight savings times, or whatever we're in here now. And so I do have like sheer curtains up there. But it's usually not a problem. And I interesting light coming through here, right? In my eyes,
Noah Kravitz 1:45:45
I can hear the voices of essentially everyone who's ever known who's known me for any, like, meaningful period of time, like seeing this and being like, yeah, of course, no, a talk to you until like it got dark out and you fell asleep. And like, of course, shut up. So yeah,
William Harris 1:46:02
no, this is good. And so in the one more thing that I wanted to talk about, I like to get into the personal side of who people are, too. And there's so many more things that we could do there. But at least the one because we've hinted at this a number of times, you were infamous, like the word infamous came up. And I got to know the story of what happened that made Noah Kravitz infamous.
Noah Kravitz 1:46:23
Sure. I want to insert an arc joke in here to get us started. But you know, years of childhood trauma coming back to Yeah, so So I think, you know, we talked a little bit about how I was a writer, journalist, phone reviewer, YouTuber, whatever. I worked for a company foreign with company called fonedog. In it was like, we moved to California and I was a teacher for two years. And at the end of the second year, when I left that summer, I started freelancing. And they were one of the places I was freelancing. So whatever year that was 20 ish years ago, right? It was before the first iPhone came out. So it was right around 20 years ago, actually, we're rocking that Motorola Razor in the background. That's right, exactly. My little wallet things is the OG iPhone, nice razor, I have, I have a small number of phones that I kept from those days. If anybody cares, go on YouTube, and look me up from phoneDog, fonedog, Noah, and there's some videos, when I had office space, where you could just see behind me, you know, those shelves that just have the open cubes, and the phone boxes just stacked? Because it was just I just reviewed phones all the time. So anyway, so I was reviewing phones, and and it was great. And we were successful. And during that time, I started a Twitter account. Twitter may have this may have been when Twitter launched or tweeted, Twitter may have launched slightly before that, but it was early days of Twitter. And I created a Twitter account under the handle fonedog, Noah or phoneDog underscore no whatever it was, I actually don't even remember, I think, I don't remember the details. But I had an account called phoneDog. And you know, along the way, like we started doing YouTube videos, this was like a very small company. And I was freelancing. And but I was like doing the, there was one guy who's like the business guy, one guy who was the programmer who built the website and all that stuff. And then I was kind of there was an office manager person. And I was kind of like the lead content person, right. And so I had a lot of freedom. It was a great, great gig. And so like i i made a video one time for a contest that this is like early days of YouTube, literally, there's all this stuff. Sure. So so for a long time, we started doing videos, and the videos were hosted on my YouTube channel, because I created a YouTube channel under my will like an alias but as mine. So that's how people knew me on YouTube. And so point being that, like, there wasn't a whole lot of forethought or anticipation or, you know, sort of corporate guidance for us to follow, right. And so what wound up happening is I worked with them for five years, something like that. And during that time built, what at the time was a decent Twitter following at least in our space, it was like 17,000 followers was nothing by today's standards, but and nothing compared to a lot of people back then. But you know, whatever it was, it was a thing. And so when I left the company, excuse me I took I cleared, I talked to them about about all this stuff. And we like transferred. Like YouTube by that point we had already because of YouTube figuring out, you know, oh, we can sell ads. And we can sell more ads on an account called Phone dog than we can on an account called, you know, nose basement 5000 Deluxe or whatever, right? And so like, we transferred all the video stuff, you know, official channels, whatever, but the Twitter handle, like had phone dog in the name. And so as part of my leaving, I was like, hey, what do you want to do at the Twitter handle? You know, I can like, grab a new one. And then like, like, I can handle it, I'll just tell me what what address you want, what email address you want fonedog Noah to belong to because like, I'm going to keep the followers, right, I'll change my account handle. And then I'll immediately grab fonedog Noah, so nobody, because we'd by that point had something squatted on, you know, some other handle on some platform we wanted. So nobody gets burned, whatever. And they're like, Yeah, cool. And so I did it. And there was documentation. Sorry, I'm talking about it. And I'm like defending myself. But you know, there's documentation, it was all whatever. And I quit. And then like, and when I told them that I was going to leave, I was like, prepared for you know, like, it wasn't going to be all all you no hugs and goodbyes. Um, and the vibe was weird. And it happens, you know. And so it's easier to do the breaking up and to get broken up with. And I don't mean that about myself, I mean, generally. And so what wound up happening is I left, and they owed me some money. And I mean, just within the realm of like, like, agreed upon stuff. And I wound up God, this is so long ago that I had the opportunity to write a magazine article. And the novelty of writing for like a friend of a friend. And the novelty of writing a magazine article was like, I've been in the internet world for so long, and I came up on print and like, yeah, it'd be cool to do a piece for this magazine. And you never know when they might need somebody else. And they're local. And it's like a friend of a friend. And so you know, they needed a phone expert to cover something. And my old bosses got wind of it. And we're mafft Just because they were miffed, you know. And it started this whole thing. And they they stopped paying me like the what they owed me. And that turned in, um, tons of really long version, sorry, I'm like reliving this. But anyway, you can cut as much of this out as you want. It was basically a dispute over a contractual thing. They owed me money, and I tried to get my money back on my own, and they wouldn't do it. And I got a lawyer who wrote them a letter and do it. And he referred me to another lawyer who like took the case on. And, you know, the whole thing tried to do everything we could and they wouldn't negotiate, or they wouldn't move on the terms which are absurd. And, and they had stuck on this thing around like the social media, and I think they just missed having the reach of the account, you know, which makes sense. But it wasn't their account. And we got to the point where we were like, look like we've prepared a lawsuit and we're going to file this lawsuit and the lawsuit is over. Like, what in the scheme of things isn't all that much money, like it was enough money to fight for. And for me at the time relying on this money as part of my like, transitioning out of that job to figure out what I'm doing. It was a fortune, right. But in the scheme of like fighting a lawsuit, you know, we were like, look, last chance before this gets just dumb. Like, come on. They responded by, I'll never forget, I was watching TV was a Sunday night and we were watching like, it was either sopranos or the wire. It was back in that era of HBO Sunday night TV. In our living room of that house. The couch, there was a window behind the couch. And that was like the wall that the front door was on. And our dry. We lived on a hill and our little patch of gravel that served as a driveway. It was like right there. And it was summer. And so the window was open or in California who knows what time here. And I heard like tires on the gravel. And I looked and I saw this car and recognize pulling into our driveway, and I was like, Oh, they're serving me with papers, aren't they? And sure enough, this guy gets out of the car and I'll get this thick envelope and he gives it to me. And I opened it up. And they were suing me saying that I stole the Twitter account when I left amongst other things, but that was the main thing and they'd come up with this valuation that With a Twitter account at a value of like $300,000, us give or take. And so they're suing me for. And this is many times the amount that was in contractual disputes. Sure, they just went nuclear, you know, and we battled it out behind the scenes, the moral of the story, if anyone is ready to hang up or turn this off, because they don't want to hear the details of this anymore, is avoid lawsuits at all costs, if you can't, sure. You know, I once heard somebody, some some business person, say, you know, look like, the only two checks, I'm happy to write every month or to my therapist, and to my lawyer, because everybody's got issues. And if you do enough, you know, entrepreneurial type business, eventually, somebody's gonna sue you. And so get a good therapist, get a good lawyer, you know, sage advice, right. But I would add to that, if you can avoid the need for, you know, at least lawsuits, because sons, and so it dragged on for a long time. And there were all of these, you know, like, like jockeying behind the scenes between lawyers or whatever. And I was very fortunate, in my my legal situation, shout out to my lawyers who cut me a break, and we're awesome. But what wound up happening is that we had a hearing. And I had, you know, asked my lawyer, I was like, look like, I'm a journal blogger, and I have lots of friends in the press. And like, at any point, if it would help us to get some press, you know, let me know. And he was like, I'll let you know. But you know, right now, the best thing you can do is just stay quiet. And it was frustrating. And so what's happening is we eventually had a hearing. Because when you have a lawsuit, there's all these hearings that like, I didn't know, I'd never been throwing myself before. And so so there was a preliminary hearing to have, like their complaints thrown out. And I think one complaint was thrown out. But you know, the case was allowed to proceed. But the upshot is that the the result of the hearing became public record. And so because that's what happens with case law. And so a legal blogger picked up on it. And they're like, oh, wait, social media account. This is interesting, because there's no law and aren't even sufficient law. But there certainly wasn't any back then. And there was no precedent for this case, precedents. For a case, it was like, there was one case involving on TV anchor, I think, but that never went to court. So a lot of court. So you know, so it got picked up on and then that got picked up on and we started getting some press. And whatever the timeline was, we finally reached a point where my lawyer was like, you know, you can go ahead and like mentioned something to one of your press friends. And let's just kind of see, because I think I had gotten an inbound requests from some sure nothing, there's a little bit of, you know, and so it's in print, so I can name them by name, it's fine. So I reached out to this guy named John Biggs, who was journalist who was the editor in chief of TechCrunch for a while. And I figured, you know, maybe if I'm lucky, this will get picked up on TechCrunch. Because I had a combination of like, I'm very sure I'm in the right here. And, you know, the ego of somebody who'd been making his living off YouTube for a while, right. So attention shy, you know, it's, I guess, shine lights, let's go fighting for this. And so, um, so he was interested. And he calls me up, and we have like a, you know, a call, Skype call or whatever. And I'm thinking, you know, maybe let's get picked up in TechCrunch. He does. And however, long later, a couple of weeks later, whatever it is, I get an email from him. That's like, Hey, you're gonna get a call from blah, blah, blah, at the photo desk to set up the photo shoot, just giving you a heads up? And I'm like, What do you mean? I was like, No, I think the email actually said New York Times as if what do you mean photo shoot, and what do you mean New York Times? Like, what's going on?
William Harris 1:59:10
Wow. Yeah. So
Noah Kravitz 1:59:14
December 20, Christmas Day of that year, whatever your it was, was on a Sunday. And I remember because I had I was at my in laws house, we were having dinner with the whole family. And I had a cold and I had this like, horrible like, sinus headache situation. And I was able to get a phone appointment with a doctor because it was a holiday. And you know, and the doctor was like, you know, normally I'd want to see you but because it's Christmas Day, and whenever I'm, I'm 90% certain this is a sinus infection, and antibiotics are cleared up. If you're cool with that. I'll just write you a script, you know, and there's a pharmacy that's open and whatever. She's like, okay, you know, let's do that. And so I remember like, I drive to the pharmacy or the hospital that the pharmacies opened and whatever. And it's deserted because it's Christmas, you know, at 5pm. And, and my phone was in my pocket was because this is pre CarPlay. You know, so my phone was just so much. Yeah, yeah. And, and my phone had been buzzing right while I was driving. And so when I get there, I get out of the car, and I look at my phone, and I have a message. That's like, you know, have you seen the New York Times homepage or something of that effect. And so I click, and there's this picture of me, like, top center of the New York Times homepage. And it's the picture that the guy had taken. And there's the article. And it's funny, because I was sick, right? Like, I had the cold, I was sick when they took the picture. And I forgot to put shoes on. Right, because I like work from a home office. And it's California. And we in that house, we had carpet, you know, rugs perfectly, never said walk around barefoot. And I was like, and I wasn't even I got dressed just for the photoshoot because I was sick. And I totally forgot to put shoes on. So I'm sitting there, you know? Anyway, um, so so. So the story comes out in the print edition, Monday morning, December 26. It wasn't on the front page of the presentation was on the front pages or the business section above the file. Right? There's no news. We're in like, a politically an economic whatever else, like everything's chill. It's the week between Christmas and New Years. So like, businesses are shut down. There's no news. Oh, here's this blogger who got sued for $350,000 for his Twitter account what I did everything from CNN live to like, am radio shows to like, I had, I had like, I was getting tweets from people who were let you know, and I had a Twitter following, right. So it wasn't like nowhere. But I was getting tweets from people that were like, your picture was in the paper in my hometown and like, you know, Reykjavik, like like, just wild. And it went on for like, two weeks, you know, because there was, you know, that first week, and then a little bit of spillover. And then the world, you know, came back to business for the new year and it and then eventually we did settle out of court is the upshot. But yeah, that was my infamy. And it's absolutely wild. Yeah. Yeah.
William Harris 2:02:27
I can't even imagine, you know, everything that went into that, like you said, of thinking this is absolutely crazy. And then it turns into something that you just didn't even think would be, you know, in a blog, let alone CNN and everything else.
Noah Kravitz 2:02:42
Yeah, I mean, and I, it's funny, like, I haven't told that story in a while, I have tossed around million times, but I'm telling the while and telling it in this context, I realized that, you know, put all the detail into the like, what happened from my perspective beforehand, and I totally glossed over the like, and then I was on all these things for two weeks, and whatever, right, because I can go through them on that as well. And I don't, I don't mean to present that, like, working with fonedog was awesome. It was so fun. And it was such a like, changed my professional trajectory. And, you know, it was not I was on YouTube, and I was on CNN once it all went down or whatever. But it was not me alone, right. There was Yeah, existed before I came along. So I don't mean to like, it was a bad breakup, and it happened and whatever. But I don't mean to go. It was just telling. And of course, I'm thinking about, like, how just awful it is to get into that with somebody and it sucks and whatever. But yeah, that's the infamy and you know, yeah.
William Harris 2:03:46
Yeah, I do want to make sure that I like I said, I'd like I already took you pass the time. I want to be respectful of that. Yeah. Um, so thank you very much for coming out. If people did want to follow you on Twitter, or x, or wherever, where's the best place for them to reach out follow you and interact? Got,
Noah Kravitz 2:04:04
it's a sign of the times, both because like I said, it's been a transitional sort of professional year, but also I'm always freelancing, and also just my age, I guess in life now. But LinkedIn is where I like, I mean, I spend time on reddit but I don't post much and I'm not It's not under my name anyway. And I'm mostly just reading you know, it's where I go to get my my sports fix. Mainly. Sports, Drama, Sports, drama isn't real. So you know, we can all just like, yeah, all right. Good escape. But LinkedIn primarily because you're the NVIDIA pod. I mean, follow the podcast. Yeah, that's on wherever you listen to podcasts, you know. But I've been doing stuff on LinkedIn, and I'm doing these random. They're not random, but I mean, they're sort of not my own publishing or my own channel, but like this like other people's channels. And usually I will post about that stuff on LinkedIn, I haven't. I stopped using Twitter, weaning myself off of Twitter was a long process. And I'm not 100% I went on the other night, because I don't know how it is these days for things. I kept using it because it was so good for like hyper, hyper local breaking news. Sure. And living in a city. And in an area where there's wildfires see all different things, right. It's like we all have whatever that is. And I'm not on like next door, and I'm not on Facebook, that, but I wean myself off of Twitter, because it had become a, like a bad procrastination habit more than anything else. And this was pre it turning into x. And then I just haven't gone back for it much, because the reviews aren't great about what's happened since then. But LinkedIn, and then you know, yeah, yeah. I was gonna try to think of like, why publish here, there. But now I, you know, I did a newsletter earlier in the year, but even that LinkedIn was where to find out about it. So sure, yeah.
William Harris 2:06:05
Cool. Yeah, I highly recommend the podcast as well, like you said, it's lots of lots of what you've got going on there. And there's interesting stuff there even just like, the rules that we're trying to establish around governance, and, you know, there's, it gets very deep and very interesting. So I appreciate that. And definitely appreciate
Noah Kravitz 2:06:26
that. And I think the cool thing about the pod, which is, again, just the same for me the same theme with what's cool about AI, you know, is, um, we've had guests from doing so many different things. Yeah. Right. And, and before the Gen AI thing blew up. And we still do, but before that, Jenny, I think blew up, we had, you know, guess who are like the pioneers in the field, right, like Peter Abele and Kai Fuli was I work for and, you know, these types of lots of people from NVDIA. It's ironic, I didn't, I didn't do this interview. And it wasn't on purpose. Because they knew about it, it was just because at that point, they were using somebody in the house for in video people, but a high school classmate of mine, the, you know, back in the day, I don't know if kids still do but back in the day, we stuck about the smartest kid in the class and the fastest kid in the class and that kind of stuff. But legit by all quantifiable and anecdotal measures. The smartest kid in my high school class, is a guy named John out or was a guy named John Alban, who is now the Senior VP of GPU engineering, I think it is, but basically known as kind of like, get Jensen started the company. But Jonah's like one of the key people, he's been there for, like, 20 odd years at Nvidia. So you know, it's kind of wild. But one of the cool things about the pod and I get to interview is that, like, NVDIA, deep scientists and icons, this field, and then we have like, this kid who built the dank meme machine. This is before Gen AI, right? And so I mean, Gen AI, but before mainstream, and so he had built a system that could produce dank memes and it could identify with reasonable accuracy. Is this a dank meme or not? And we spent half the show me trying to get him to define what a dank meme is. And laughing at me for for not just not knowing, but for being the kind of guy who's like, Oh, come on, please tell me you know, so like, yeah, yeah, there's something for everybody. And like, that's good. You know, science, creativity, music, blah, blah. It's yeah, it's a cool show, because of the guests. I'm super fortunate. It's, you know, I'm not full time. It's a small piece of my larger puzzle. But it's been through the pandemic, man, that was my lifeline, right? Yeah, like stuck in the room, who knows what's going on. But I am interviewing this guy who's building a supercomputer across the world that's, you know, helping research like vaccines or whatever. It was just amazing.
William Harris 2:09:06
Well Noah Kravitz, again. I cannot thank you enough for being on the podcast, sharing your time, your wisdom with us, and a lot of really fun, wonderful stories.
Noah Kravitz 2:09:15
It's been my pleasure. You know, I hope whatever makes the final edit because I went on for a while. I hope it's useful to folks out there. But thank you for having me. This has been a blast and I appreciate the you know, the line of talk, give me a chance to kind of open up so thank you.
Outro 2:09:31
Thanks for listening to the Up Arrow Podcast with William Harris. We'll see you again next time. And be sure to click Subscribe to get future episodes.