Jon: Are you in? Are you in the zone? I'm in.
Jon: All right. So let's talk about one of the next waves in Vichy, or at least one of the next waves, I think is coming, which is called generative tech.
Jon: For those who don't know how to do this pace, the best example perhaps of generative, that person will give an a quick definition. And it's when you have A.I. models teaming up with collaborators. So Openai has a product called Dolly detailed that E word allows anyone to create very robust pictures that like look like it was painted by Picasso and you can enter in a few words and it's using an AI data set and model that makes everyone artists and perhaps makes a historical artist feel a little relevant.
Jon: And they can create thousands and thousands and thousands of images without ever actually working with a designer. You can say, I want to create something that looks like a Picasso that has a photo of Steve Jobs, and it would put it together for you in a few seconds. So that's kind of the base, the broad base example of generative tech.
Jon: Maybe another example of generative tech is with movies right now as we talk about like phases in the Web right now, you know, companies like Netflix are producing, you know, videos for the masses, but generative tactic A.I. models. And it could create a movie specifically for Peter Harris or for you, the listener or for me. And the movie that you might see might may look very, very different from me.
Jon: And the ending that you have might be very different, and it might get to the point where it is these generative models, if I understand correctly, maybe you could correct me if I could, either the model would create it of its own or I would team up with it to create kind of a really good movie for me that I'd like.
Jon: Or maybe another example is right now an example of generative tech is right now. I could create a playlist on Spotify and share it with you, but in the future, generative tech may say, Hey, Spotify might come through and say, Hey, this is Peter Harris, and it would create a unique playlist of unique songs that have never been heard before that are specifically for you with a that be a good example.
Peter: Yeah, I mean, to an extent. So the only thing I'd add is it's the idea with generative tech is that it is this collaboration between a AI and a human right. So the human is giving it guidance and then what's interesting is generative tech is going back to these large databases and sources of data like the Internet and pulling on that, leveraging kind of the guidance of that human to create something that's useful to that human or to the broader population.
Peter: So in your Spotify example, it's yeah, they may create like a customized list of new songs that have never been heard before that I personally love. But it's going to be based on inputs and guidance that I'm giving it, and that could be by showing it the other music that I've listened to, right? It could be me providing other inputs from my life as well as just, Hey, you know, I'm in the mood for something like relaxing right now, right?
Peter: Give me something relaxing and jazzy, right? And then it would come up with some some type of music that would satisfy that need. Yeah, it's a super, super fascinating new development and, you know, I think it makes a ton of sense. You know, I can't remember where I read this, but a number of years ago they were talking about how these A.I. models were beating all these chess masters.
Peter: And, you know, everybody is kind of freaking out like, hey, you know, humans are becoming irrelevant, blah, blah, blah. Not only did that, not really truly happen, but the other thing that that this article pointed out is that, well, A, I could beat a human and AI plus a human could trounce any A.I.. And it's the combination of these two, right.
Peter: That kind of opens up these huge new opportunities. Because if you just go to an AI and say, Hey, show me good art, right? That's really hard for them to do and figure out. But if you go to the area and you say, okay, show me a futuristic neo punk world where, you know, this good looking man is riding on a futuristic vehicle and all of a sudden it can start pulling from all these different cultural references and other things and create that that picture for you.
Peter: And, and so it's like this collaborative thing because most people wouldn't be able to create that on their own, but they do have an idea of what they want, right? And so it's this ability to like, think creatively and then use AI to actually bring what it is is in your mind, hopefully into fruition.
Jon: Yeah. So there's a lot of heat in the space, obviously. I mean, Openai has raised hundreds of millions. Anthropic has just raised 700 million over a over 700 million. Like maybe they need to correct that. Yeah, they've raised over 700 million. I'm just double checking. And this space is getting hot from a VC perspective. How long will the space be hot for?
Jon: What are the industries that and what are the industries you think it will attack or, you know, completely reinvent?
Peter: Yeah, it's a good question. I think I think it's going to touch everything. I think this is where you see kind of AI 2.0 in a lot of ways, like start to really flourish. You know, I remember so our friend Diego over at the album, he funded a company called Latitude. Latitude makes a game called Dungeon. I is a deal he did a few years ago and it was kind of a crazy deal at the time.
Peter: But what what Dungeon AI does is it's this like Dungeon tax based dungeon crawler. And when you play it, it creates a world just for you based on your inputs into the model. Right? So, you know, do you want to walk down to the to the lake and you say, Yeah, I'm going to walk down to the lake.
Peter: I'm going to pull out my sword, right? And then it creates a world around you. So that's one example. I think there's a ton of examples. I think you're going to see instances where, you know, maybe you're a sales rep and you're going to go into a meeting with a client. The ability to say, okay, tell me what's the best way to interact with this client?
Peter: And they're going to go and pull all of this data from all these different data sources, and they're going to feed that back to you and say, okay, here's your cheat sheet. I like.
Jon: That idea. There's a good business idea. And I feel like the first places where we've seen generative tech in an actual business model because I don't think right now is a true business model. It's more of it's kind of like investing in the future. Sure, it's things like Jasper A.I. and copycat A.I. where it's it could completely replace a team of like ten copywriters.
Jon: Now, if you just have an ad, you know, you can take your your marketing copy team from ten people to a team of like two people that that work with this. Yep. My favorite ad that I see from them is, hey, this ad is showing you I, you know, I works.
Peter: Yeah this ad was built by. Yeah, there you go.
Jon: That's the but that's that's exactly what it was.
Peter: Yeah. Well, and I subscribe to this newsletter and, the one of the authors of this newsletter, because it's like a group of them over the weekend kind of built a text generative AI tool to help him write more of these articles. And what was interesting is he showed like, okay, I'm going to write this basic idea of what I, what I want, but then, hey, I give me like ten supportive arguments for what I'm, I'm trying to say.
Peter: And then also like, give me ten reasons why this doesn't work. And, you know, he goes through and he's like, Hey, look, half these are not very good, but the other half are pretty good. Or their direction really pointing me in ways that I wouldn't have thought about before, Right? Yeah, They're kind of sparking additional creativity and helping me accelerate my writing process a lot faster.
Jon: What do I think about generative tech in the medical space? So, for example, we were at the doctor's office just today.
Jon: And, you know, it still amazes me that between the doctors we've been seeing, they're not sharing all the imaging. Yeah, they don't all have that data. And for probably the last ten years, to my mind, you know, the guy in the 20 question games, a little robot. Yeah like that is an incredibly smart game and I'm like, why has the medical space almost not interacted to that level where a doctor could be working with a 20 questions game?
Jon: But they could grab all the data, They could perhaps be asking the patient question in advance and have a much more guided process that would probably save them a lot because, you know, our visit today, the doctor had literally just 20 questions for for us as part of the visit. And some generative AI models have already had that done before we even up the doctor.
Peter: Yeah, well, look, I mean, I think one of the questions that you always have to ask yourself as a venture investor when you're looking at a company is why now, right? Why does this opportunity exist now? And it didn't exist before and you know, tomorrow will be too late. And I think if you look at generative AI, the reason it exists now, like why is now the time?
Peter: I would argue, you know, a lot of people would say, well, it's because like the open source finally got released and was, you know, barriers came down, etc., etc.. But from my viewpoint, it's like processing power reached finally reached kind of these points access to data reached these points of the internet so big there's so much content out there right Five years ago there may not have been enough content to actually feed these engines to do things that were super useful yet.
Peter: And so kind of coupling those two things and then also making it just like the cost to produce these things, the cost of processing power and everything is like driven down so far that like anybody can use it. And so once anybody can use it, right, like this guy in this newsletter, right. Like he just went and tapped in to kind of these AI engines, all of a sudden you see this proliferation of different ideas or people like, we could use it for this.
Peter: Coming back to your question around the medical field, you know, why hasn't this been done before? I think, you know, to a certain extent, there's a lot of the pieces that you could argue should have like we should at least have been on this road sooner. But if you think about, at least from my viewpoint, part of the reason why now is the time for generative AI or generative lab or generative tech to really take off is that access to data.
Peter: And in the health care arena, people are not incentivized, frankly, to share data. And so you have all these data silos, and so it becomes really difficult to start pulling in that data to run true AI because it's so dependent on, you know, ingesting data, analyzing data, identifying trends, all of those things. And if the data quality is not very good or it's just not available or it's the sample sizes are too small, there's not much you can do with it.
Peter: Right. And so I think that's that's a big part of it. And I think the other part is you have these doctors that are highly protected, right. To be a doctor to practice medicine, You know, you have to go to school, you have to receive certain licenses. You have to be malpractice. Right? There's all of these barriers that exist that make it hard for just anybody to start creating tools that could solve these health problems.
Peter: And doctors, again, have no incentive to really support and grow those types of things because it's a threat to their their livelihood. Right. I think what's interesting about this art and the writing and these other things is that there aren't barriers to being artists, there aren't barriers to be an author. And so it's a lot more, it's a lot easier for these types of technologies to spring up and, and, you know, create really interesting solutions.
Jon: Can with, with things like generative tech can founders and VCs be competitive against some of these really large behemoths like an IBM, you know, and moving forward is a still kind of an innovators game startup. And, you know, VCs or as the amount of cash to build these models and collect this data more of an enterprise model.
Peter: I think that's a good question. You know, my impression is, is that open source is going to provide a tremendous number of opportunities. It's that's the way it's been for the last 20 years or so that more and more corporates are realizing that there's tremendous amount of power within open source communities and so forth. And you could even argue that the greatest gains in generative text and the areas where generative text has the potential that to offer the most gains is within this open source community where open source developers can leverage generative tech to expand and grow and develop additional algorithms and models and access to data and so on and so forth for for
Peter: more and more and more applications. That would be hard for anyone corporation to keep up with. The flipside, though, is, you know, will will it be hard to create barriers to entry? Right. If you're a startup, if everyone has access to these similar types of products and tools? My guess is probably yes, because let's be honest, there's there's nothing that special about most SAS applications that are out there, and yet many are able to build out very, very interesting businesses and interesting moats over time that they give them an advantage.
Peter: And I don't see any reason why I wouldn't be the same for generative text or socket generative tech.
Jon: If you were to pick two or three industries to either focus on as a founder, as a VC yourself, what would you focus on?
Peter: So, you know, it's a good question because, you know, we're heading we're likely heading into this recession, which means people are going to be thinking a lot about, you know, how do I cut costs, those types of things.
Jon: Isn't that where generative tech wins? Like, I mean.
Jon: We will space. It's one of the models I've thought about in the past historically, and the data for the most part is becoming much, much broader, much more available. And then plugging in engine on something like a legal zoom or.
Peter: Whatever could be really, really compelling, right? Both for contract creation as well as contract review, contract negotiation. Right. A lot of those things. Yeah. But I think you're know, you're always going to still need somebody that can kind of guide it, that knows what they're doing for sure.
Jon: But even like you said.
Peter: But they can be way more efficient.
Jon: You know, you know, creating a service maybe just for contract review. It doesn't have to look anything historical, just it look through it and say, hey, how do we think this document may be challenged or misinterpreted? Yeah, and that data with a skilled attorney, instead of taking, you know, you know, a $10,000 bill. Yeah, it could just be a thousand or $2,000 bill.
Peter: Yeah, I think opportunities. I mean, at a fundamental level, I would look for opportunities where if you think about generative tech, it's this idea that it amplifies human talent and ability. And so I would look at like, what are the things that are the most costly for humans to do, especially from like a knowledge worker perspective and focus on companies that are building tech for that.
Peter: Because theoretically, all of a sudden you're able to get a ton more leverage off of that individual than you could get before. So like if you think about like software developers, really expensive on a per hour basis, right? Well, what happens if you provide, you know, your rock star developer with a generative tech tool that really amplifies what they're doing and makes them ten X more efficient?
Peter: Like all of a sudden they're ten X better and you're you're working with a developer that's creating ten, ten X or 100 X, the value just on their own. Then you know somebody that's not. And so I think I think that's an interesting area to focus on.
Jon: But that's interesting because that's something we're actually looking at, at Code Base right now. So there's an application by GitHub called Copilot. And the idea is that it's it's paired programing, but typical paired programing is like you have two Ruby devs or two full stack devs. In this case you have an AI engine powered by GitHub and then you have one of our develop a developer.
Jon: And so we actually have some of our developers running through a project right now just to see how, how do we like it, how you know, how does it compare and yesterday asked, you know, others on LinkedIn, I feel like a lot of people don't have experience with this, but I think they said short term they probably don't see a lot of value right now, but they definitely think long term.
Peter: They could have a ton of value. Right? You sit down and you're like, Hey, we're going to build this type of app. I need you to go find these different libraries and just dump them in right? And have the AI just automatically find it, know exactly what you're doing, dump it in, you can look at it, review it, say yes, yes, Now let's fix this.
Jon: I think that's also like what Builder Air does as they come through. It's what do you want to build? Yeah. In the user age to scan the web and to say, here's the cheapest way we can build X, right?
Peter: Yes, I think that's interesting. I do think, like there's this idea around, like I mentioned earlier, around just like being more emotionally intelligent that in our conversations with each other. Yeah. And pulling in data and leveraging that data to have richer, richer and deeper conversations with people and kind of remove a lot of those barriers. And so if you apply that to sales, right, like how much more interesting it would be if you talked to a sales rep and they, they already know enough about you that you can have like an interesting conversation and they know how to engage with you at a more personal level that really matches with your personality.
Peter: All of a sudden you're going to be more likely to to want to do business with them.
Jon: Could you go through someone's LinkedIn posts and figure out what is in sales of the four quadrants? Are you an analytical or an emotional person? Yeah, and it could like quantify them beforehand and to say, Hey, here's the approach that we might.
Peter: Yeah, sure, why not? And then tap into data sources like, you know Claire but Zoom, etc., where it's like, okay, you know, you've already scraped all this data around the individual was tap into that and leverage that as well. yeah. So I think, I think a lot of those represent opportunities. There's probably some opportunities in the creative space.
Peter: I think my concern there is that, you know, going back to what I said before, like if there aren't many barriers, you basically move to like a commoditized environment and it's really hard to generate much money. There. So like if I am if I'm one of those website, like a like an iStockphoto, I'm probably terrified because now instead of going to iStockphoto and like hunting and searching and hoping that I find that, you know, some photographer somewhere took the picture that I'm looking for, I can just go on to one of these AI engines and say, Hey, give me this, and boom, I get exactly what I want.
Jon: And move the the picture, the palm tree to the right. Yeah. And add in a picture of a couple that. Yes. Mixed race. Yeah. And give them three children. One that's an.
Peter: Amputee. Yeah. One of the things that like personally I'm super excited about is you know there there are some of these new tools that are coming on board for like video editing and you know, you could imagine like plug ins into word and PowerPoint and other things that are you know, it's just like, hey, this background, get rid of it.
Peter: And I just automatically finds it and gets rid of it, right. Without you having to like, go through and manually like, select every single little bit that needs to go, go or like pick this person and follow them along. Right. so I think there's a lot of tools like that that are going to be like, you know, really cool and they're going to democratize access to, to more people to be more creative.
Peter: I think there is a threat here that that's going to put a lot of knowledge. Workers theoretically could put a lot of knowledge workers out of work because they get replaced by the AI. But the flipside is I think it's going to open up so many more opportunities for more people to be creative and create in partnership with AI that that those gains will will offset.
Jon: Yeah, I like to I hope that the that AI ends up being like what the cotton gin was for us. Yeah. Were took just to get a shirt. There were so many people who had a touch it manual labor repetitive that was very, very painful. And then when you know the cotton gin, which was one of the first one of the first industrial inventions that replaced a lot of labor, so put people out of work.
Jon: We can focus on other things. Greater advances in medicine. Yeah.
Peter: Well, it's like a lot of people don't realize. But, you know, when this country was founded in the US, 90, what, 95 plus percent of Americans were farmers. And today it's, you know, it's a few percentage points. Are farmers, maybe even less than that. And, you know, what are we doing? Are we all starving now? We're doing things like building airplanes and writing code.
Jon: Recording pottery, recording.
Peter: Podcasts, arguably maybe less valuable than growing food. But the yeah, just opens up these new opportunities. I will say, though, that like one thing that I guess I worry a little bit about is that like as we move to this like hyper personalized environment, do we lose community to a certain degree, right? Where it's like you talk about that Netflix show, right?
Peter: This this idea of like, okay, it's going to create Netflix shows that are perfect for Peter, but then do I lose something where it's like, I want to talk about that show with you, but we no longer have this, like shared experience.
Jon: Where you just be living in the metaverse with all your your metaverse friends that probably never existed.
Peter: That don't exist. They're just figments of my imagination. Yeah, there's a bunch of ones and zeros. Yeah, potentially that. But I don't know that that's the dystopia I want.
Jon: To live in. I don't know. It'll happen, but we'll see what happens.
Jon: All right. Well, is there anything else you think we should know about generative tech before we close this podcast?
Peter: No, I just, you know, we as humans, you know, I think we've talked about this before, but we as humans aren't very good about thinking in terms of exponential growth. We think more in terms of linear growth. And I think this this generative tech movement is one of those examples where I anticipate that it's going to grow incredibly fast at an exponential rate and it's going to blow people's minds in the next probably couple of years.
Peter: The things that are possible with generative tech that, you know, nobody even thought about today.
Jon: Maybe we should try a generative tech website building tool for University Growth Fund or for this podcast. Yeah. Venture capital firm. And see what see what it comes up with.
Peter: Yeah, that's not a bad idea. Maybe we'll just end up getting replaced by generative tech podcasting.
Jon: There we go. Well, awesome. Well, thanks everyone. Thanks, Peter, for coming down. Good. A venture capital law firm. And you can find us on Spotify, Apple, YouTube, all of our links are right there. Venture capital, a lot of them. And we will see you on the next episode. Thanks, guys.
