VCP 100: Nominate the Top Utah Founders VCs Must Watch! Nominate Now

Scott Paul

Jon, Peter and Scott decode how Chat GPT will impact the future. If people need to be scared? Or better prepared. Jon- talks about how and are enablers for people to get their MVPs off the ground and not direct competition.And while Chat GPT might feel threatening at this point for the businesses like his, we have to remind ourselves that not everything turns out how its supposed to.

Unpacking the Buzz:Our Thoughts on Chat GPT

Jon, Peter and Scott decode how Chat GPT will impact the future.

If people need to be scared? Or better prepared.

Jon- talks about how and are enablers for people to get their MVPs off the ground and not direct competition.
And while Chat GPT might feel threatening at this point for the businesses like his, we have to remind ourselves that not everything turns out how its supposed to.

One of the big examples being Tesla’s self driving cars.

Peter chimes in with an example where Chat GPT was able to spurt out a quick answer to a legal question a lawyer asked. But after fact checking it soon turned out to be factually incorrect.

While the technology in itself is incredible, it has a long way to go.

Scott Paul is the Founder of Convoi Ventures.

He’s helped turn Utah into a thriving startup culture and he is a supportive leader and angel investor.

Hosted By
Our guest

Episode Transcript

Peter: Welcome to the Venture Capital podcast with your hosts, Jon Bradshaw and Peter Harris and special guest. Hi, I'm Scott Paul. Thank you for joining us.
Scott: Scott Paul. That's quite the intro. All right, let's talk about something.
Peter: Let's talk about the thing that everybody is talking about, which is chat GPT.
Scott: Part two. Part three. Part three. I was with GP T way before it went mainstream.
Scott: I mean.
Scott: You may not have flirted with it a lot. Okay. Yeah. Now we're talking about Chat GP exclusively or the exodus from Web three to chat from Web three and crypto to chat GP.T I think we just going to be.
Peter: Talking about the new hotness.
Scott: We're talking about how everyone used to be in the crypto until about four months ago and then maybe just two months ago in this chat, GPT thing came out. Now it's between the genitive art in the, between the art and the Chat GPT. I think AI is on every mind from the high school student to the,
Scott: Yeah. We, like.
Peter: Traded nfts for generative art.
Scott: Yeah. Yeah, It was always me.
Scott: I always thought the art should be. The NFT should be created by this very amazing artist that I think is the the AI. But it's like.
Peter: But don't you think that also kind of like, destroys or ultimately kind of destroys the value of NFT as it was just like.
Scott: We're already that way. There are.
Scott: There are already several things that they were just throwing into a processor to make, you know, ten, 10,000 different combinations.
Peter: But the difference is now that I could make stuff that actually looked good.
Scott: Exactly.
Peter: And was like pixelated and ugly think, yeah, and it could do it at scale. And so then I was like.
Scott: Wait a second, I, I can almost make an NFT like a full collection of, like, profile pictures every 2 minutes. It could probably generate something so amazing that we start to wonder, like, there's probably no end to.
Scott: This cohort of photos. And it's they're.
Scott: Going to be flooded to the max. And so I think NFT just started to focus on the fewer urges. And we have our Rembrandts and or, you know, art or Dali or whatever these, these the bored apes.
Peter: And yeah, yeah, the.
Scott: Cryptopunks we got the mainstays out of that those profile that, that function NFT because that's just one thing. The NFT art is just one part of the NFT world and I think we got what we need to have that and we're happy has generative AI replaced physical painting? So using a robot with generative AI to replicate or do like.
Peter: I'm not familiar of the reason, but it wouldn't surprise me.
Scott: We know. We know that the AI art is winning. You know, art contests, the generative I kill in a piece that's contributing. I think I think that's.
Peter: Well, our our art based NFT.
Scott: There's some correlation, but it almost felt like Nfts killed benefits. Ah.
Peter: Yeah.
Scott: That was I think that was a just a once.
Peter: You could like flip a switch and pump out tons of that NFT without even thinking about it. That just kind of killed NFT once.
Scott: Like Twitter and Instagram just took over. All your DMS are just spammy, spammy scam scam. It's just like, stop, you know? Okay.
Peter: So my question is with AI, is it because there's all these companies that are now like pitching VCs, right? Like we're X, Y, Z, AI for this and that and the other thing. And a lot of them are built like explicitly on chat. GPT And so I'm just wondering like do they actually have anything interesting? Will they actually be able to stand up an interesting business or.
Peter: Well, at the end of the day, it all just kind of revert back to chat. GPT And you know, these really big, well-supported large AI systems will essentially just do everything right because even GPT it can code, it can write stuff, you know, there's no reason why it can't do generative art like it could do a bunch of these things.
Peter: And then it's like.
Scott: Well then I have some.
Peter: What can you really do?
Scott: Right? Yeah, I think Openai I mean, it's doing CBT, I think it has all those different options under it. So when we say AGP testing, we're thinking maybe openai it's got all the Yeah, it's got all this stuff and I don't know how much they're giving out to like, you know, developers to really build on top like unlimited use of their chat features.
Scott: I don't know yet because I feel like customer service should be, I should be interfacing with chat. CBT As.
Peter: For customer.
Scott: Support, yeah, I think I am with Apple, with Ray at my apple card whenever I'm like in text with them and like I'm really pissed about this pen I just got, I just got ripped off, you know? It just happened to me the other day. I doordash something and I'm here and you turn it, send it to an address in California.
Scott: My, my, my checks away, and I'm like.
Scott: I just spent $50, and they went.
Scott: They gave me only $12 back when I told them you got to stop this order. And it was within 10 seconds. wow. And, and, and so I went, Ah, you did. And it was definitely an AI taking care of me. Yeah. And that was that was fun because, like, is.
Peter: It like that? Good? It's good. Usually the chat runs are not super.
Scott: Don't usually the.
Peter: Horrible but this was.
Scott: The one for Apple card is in text because on your phone and they're in charge of you know they own that.
Peter: Your face and.
Scott: AI is really good because they know that they know what you're debating. You can go right to the transaction that just happened and you.
Peter: Take all that data.
Scott: And follow up with you and then like, Hey, we got that cleared, you happy? And there's no way it's human so fast. I watched it say Verizon sometimes it's human. I can tell. And it's very it's tedious. It's more to use than having a phone call almost.
Peter: Yeah, I.
Scott: Once you get passed off, you can tell when it's bought and the you gosh now it's.
Peter: A real slow because they're managing like 50 other chat windows.
Scott: It's charging changing your investor thesis. Not yet but probably will It's changing how I operate my company.
Peter: I think the big thing is that if you don't have an AI play, it's like don't even bother.
Scott: Having this that for yourself.
Peter: Yeah, but I feel like now it's there was like an AI, but really it was like machine learning and I feel like now it's like I like.
Scott: To your mind, you cross from ML to A.I. because y look at what everyone's looking at right now is mostly just EML.
Scott: Yeah, yeah.
Scott: But they called it a AI.
Peter: So I think I think the difference, though, is that 80 day is a little more contextual and it's responses and it's recommendations than ML was. And as.
Scott: I think Emily.
Scott: Had to build the app around it still, and I think now you've got it's now like email in a sense. It's like, why would you not have a company where you tap into just as a student should be tapping into DPD and stuff? Why wouldn't you have a company where you're asking questions? And I've had I've had I've had a I basically tells exactly the logo we need to create it described it to us.
Scott: It tells us how to go about certain things. It's been informing us on certain vitamins we're looking to use in a product. I'm just like.
Scott: This is way.
Scott: Better than a search on Google. It's different. It's definitely different. We're not like digging into, you know, company or other company like, like competitor data. We're kind of digging into just you can tell it's an amalgamation of just a ton of data and it but it's not horribly wrong. Like we can we can go fact check the stuff down.
Scott: That was a nice report. Thank you for that 20 seconds of processing. You you definitely they made us think about it differently. And actually it's change as an operator is changed the direction of our company many times and made us think about different things. And even in writing our policies out in our mission and stuff, we're we're co-creating with with 100%.
Peter: So you bring up an interesting point you say is all currently male. So what do you mean? Like what's in your mind the difference between ML and a good question.
Scott: That's something we've been debating internally. It's kind of like, what's a robot I would call a dishwasher, a robot. Okay, but maybe you wouldn't. I think a lot of how when people look at AI or ML, it's kind of it's the top ten or top 1% is when it really amazes us is.
Peter: When we call it an.
Scott: AI. And I think that's historically has changed from from early on.
Peter: So I've always just thought that I was mostly a buzzword, but I think once you're able to like ask a question, that's a nuance. So it turns up on kind of the contextual details and nuances of what you're asking and provide you an interesting response that takes us into consideration. In my mind, that moves that beyond just like normal to AI, because in my mind.
Peter: ML is more like, Hey, we're going through this iterative process and we're just spin stuff out, you know, almost like reports. But this is a little more of a conversation in terms of the data you get.
Scott: I don't know, is it a conversation? Is that the distinction? Because I don't think you're actually having a conversation. I think you're just experiencing machine learning.
Peter: Okay.
Scott: And we built one that was a conversation. Now that that you go back and forth, turn based conversation, but it's making that decision off of data set. So still, the question is when is MLR, when is it AI or when is it self-aware? Is this self-aware and are you self-aware? It's good to.
Peter: Go deep.
Scott: And take some mushrooms.
Scott: And figure it out and you done that? Yep. Never should do that on the podcast.
Peter: But do you think, are we going to hit Singularity?
Scott: I mean, I feel pretty good about it.
Scott: Again, let's take some of about okay, you guys don't have that advantage.
Scott: That desire is already here. We can get it's been the work underground right now that we can tap into.
Peter: Then everybody just like bile. Biologically, I'm right.
Scott: I think so.
Scott: Potentially.
Peter: Eventually. Here's the other question. Is I just another fad? Is it going to be like, no, you know, no Web three.
Scott: This time it's not. Now, obviously, the difference between ML and AA is its ability to make inference. So historical things, there's a they they created they had like this room with a maze and they had a couple of different like a like bots and you could have tools and it was given a assignment and the way it was making decisions was primarily running an infinite number of scenarios.
Scott: And every now and then throwing something in. Yeah, and I think the inference is a big difference between machine learning and AI now. And the question is how much of what's happening is true inference like coming up with unique ideas. And I think in this scenario it was realizing it could take some of the devices and catapult someone through Internet like into like a like across the maze to kind of beat to beat it.
Scott: And, you know, I think the ability for it to actually make inference versus just running an infinite number of tests test and then, you know, randomly coming up with one. Yeah I think historically humans would be someone and she didn't get the actual thing it wasn't cheating it was it was given environment and then it came up with another option to go explore.
Scott: wow. And it was like the million or 2 million.
Peter: It's a ton of data and then arguably creatively came up with a solution.
Scott: But it's like, this is a 2D maze, this is the 3D. And I would say that's kind of the line between A and ML.
Peter: And I agree that's kind of what I was trying to get at with like bring in the context right? It can see the maze and it can take into account all of that context and then make decisions based on that.
Scott: It started thinking in the Y dimension, right? I like that.
Peter: But, but yeah, this idea that it's not just it's not just a feedback loop ultimately, like ML is just a big feedback loop, but it's like it's a feedback loop, plus the ability to take that information and come up with unique responses.
Scott: I think another difference between true AI and ML is right now it's got a.
Peter: Massive amount of data.
Scott: Set which potentially we're going to run into. Copyright issues are Android experiencing. But true, I could be given a little bit of data and it would start growing and making inferences and learning. And that's why I say right now what you're seeing as. ML If you give it a very small data set, it would be worthless. You would not be impressed.
Scott: Whereas a baby with a very small dataset, it starts, it's grabbing, it's learning. At some point it still.
Peter: Has to be able to like pull out. And frankly, this is something it's not good at yet, but it's got to be able to like differentiate between what's true, what's not, and serving up the best as I can do. One of my students who's a law student and he was saying that a friend of his works for this judge and he was trying to find this case to support some outcome that he wanted.
Peter: and so he just was like, well, maybe I'll judge it. So he asked Jeopardy, and he's like, Hey, find me a case that like, provides for this outcome. And TATP came back and it like had the case and it was, you know, so and so versus and so and like had the opinion and it was almost like almost exactly what he was looking for verbatim.
Peter: Yeah. And so he's like, this is amazing, like 20 seconds and all my work's done. So he, he took it and he was like, I better just fact check it. Yeah. So Chachi PD made up the whole thing.
Scott: Yeah.
Peter: The whole thing ever looked so real that he could have fed it to the judge. And theoretically, the judge wouldn't have known better. Right?
Scott: So that's what I'm wondering, is how he.
Peter: Could have passed a law, essentially passed like a law in a way, because that's how the judging was based on.
Scott: Based on of fictional statistics. Yeah, I think that's.
Scott: Amazing, actually.
Peter: That it pretty interesting.
Scott: Makes you wonder when does a judge when they use stuff in cases who is fact checking.
Peter: Yeah all these little this all these interns right. Yeah. Are they like. Well that begs the question right now right. With with stuff like this, you know where they can just make it up.
Scott: So in a world where we can like make up votes for presidents and just say whatever we want, I don't I have got to the point where I just actually don't believe anything. And it's not a bad deal. Meaning like I was shown this quote and it was supposed to be by Steve Jobs. It was a someone got an email where he emailed himself this really cool thing.
Scott: And I'm like, how do we know? How do we know? This was actually something.
Scott: He emailed to himself? Steve Jobs, this thing. And it's like you just can't look at.
Scott: Each other and yeah, maybe we.
Scott: He's like, Doesn't matter. What if this was just made up and.
Scott: It wasn't Steve Jobs at all emailing himself this like, does it matter? Is it just the way you feel about it when you read it? Does the messenger really matter? And like, kind of does matter?
Scott: Does matter or maybe it doesn't. It's I'm up to the point.
Scott: I don't know. It's true. And it's like I don't know who's actually fact checking it. And in the back to the like the kernel of truth, I don't know if I care.
Scott: So I really don't know. I mean.
Scott: We could you could scrutinize almost anything. We're in a world now where people think it's flat. People can make up birds aren't real, and.
Scott: 10% of the population may buy it. So it's it's like.
Scott: It's a strange time to be alive when when truth is not. I think that there's no truth that we can actually put down the table as a truth at this point.
Peter: It's interesting you say that, because I would bet that in general truth is more broadly just like distributed and agreed upon than any other time in history. It's just that like we didn't know that there were so many people that believe that Earth was flat before. But now we do, because like, there's a mechanism for them to communicate.
Scott: That they can go through that.
Peter: But I think the nuance there is that as technology keeps getting better and better, we're having like deep fakes for sure, and then it becomes easier to fake stuff in such a way that it becomes it looks real. Yeah.
Scott: And then like, how do you find that's true? Do you interview the person with them on camera? And they're like, yes, I, I did write this. And then you realize that that interview can now be fabricated where you're yeah that's not even Steve Jobs have you heard the Joe Rogan Steve Jobs interview? Now you have to go listen to it.
Scott: I to I mean, it doesn't make sense. But anymore I'm I'm watching stuff online and I'm like, I don't know what to believe anymore. Yeah, yeah. Do this. This Joe Rogan podcast actually happen more and this one that didn't.
Peter: And then you get into this whole other level of like, did what do you remember? And what you remember is not always what exactly happened. In fact, most of the time it's not your brain. Like recreate your memories in such a way that solves certain, like subconscious needs that you have. Part of the question, though, too, is like, how does impact venture capital?
Peter: So do you worry how do you find truth when you're evaluating startups and like, did the tools and the technology help with that? Do they hinder it?
Scott: there's stuff I.
Scott: Feel like, Mike, you said more than ever there's tools to kind of find the meta. Like what's true? Yeah, that's above like actual stuff, like you just kind of.
Scott: You know, you can. I, one of the easiest.
Scott: Ones for me is when I was at Disney, we had to evaluate Instagram accounts and and find out if they were truly influential accounts or if they were not written.
Scott: And there's just some really easy ways.
Scott: That kind of like.
Scott: They can have a lot of followers, but it's like their posts are.
Scott: Full of engagement, then it's like I, they're not really like that ratio of follower to engagement. It's just really low. And I'm just like, I don't know if it's really something we should be paying a lot of money out to. And, and then you find these other accounts we're not that many followers. Massive engagement, real.
Scott: Comments, all this stuff going on. So there's this just there's these things you can do that same thing.
Scott: When you're looking at investigating a company just to figure out what's real. You just got to get in there, open a few doors, ask a few questions, and if you're good at this, you just you just kind of get to those those truth really quickly feel. It's like.
Scott: Has there is there.
Scott: Something there? Is there have they been have they been working or have they been you know, just coming up with they just more working on the story again and again, again? Or are they actually making something I can touch and feel? And I'm I have a huge bias for like, can you put something in my hand right now that shows me you you can create in any UI?
Scott: Can you can you launch an app? Can you.
Peter: Can you connect me, do a customer that can't live without you?
Scott: Exactly. Yeah. And that's that's a.
Scott: Good little litmus test.
Peter: What do you think, John? Are you worried you're on code base? You worry the air is coming, coming after your devs.
Scott: I think it's much further out than people anticipate.
Peter: Okay. I was talking to a developer and they were like, I love it. Yeah. She was like, It saves me so much busy work time. Yeah, she's like, I, I can just have chat, CBT, draft all the code for that, you know, that's kind of the boilerplate stuff for me. And I just focus on the core stuff.
Peter: I, I thought that was kind of really interesting. And she's not like a big heavy programmer.
Scott: But I think that's the trend that's been happening in development. I mean, used to have a team of individuals launch a very simple website and now a college student can launch a website. Yeah, or high school student could do it themselves. And so the problems that developers are focusing on are continually going upstream as these smaller tasks are now becoming automated.
Scott: And so what you're seeing is now you've got things like bubbled audio. I don't see bubbled audio as a competitor. I see it as an enabler and I'm pushing people who aren't good fits for code base clientele to bubble audio builder dot I yeah, go get your MVP off the ground because at a certain point that you'll want the customization and that's when you should should be spending.
Peter: Want you once you want customization and scalability.
Scott: That's actually happening.
Peter: You've got to come back and.
Scott: Build. We there's a guy named Kevin Andrews. I, I've, I've been chatting with him and I said, Hey, go check out the you know I think I showed him build area initially Yeah. And within a week with no development experience, no design experience, he has an Airbnb equivalent model up and running and he's got a unique spin on it.
Scott: So at this point is now hey, he saved $100,000 plus is now focusing on can I get adoption? And at that point, at a certain point, hopefully bubbled audio keeps moving with him so you won't have to have a huge dev expense that at some point you'll want to jump off.
Peter: So here's one of the things that I've been thinking about in terms of venture capital too, is that, so we've seen like the cost to start a company just drastically drop over the last 20 years. yeah, with like all this stuff.
Scott: Yeah.
Peter: But what's, what's happened is you've had this like proliferation of startups, but it just gets more and more and more expensive from a funding perspective to, to like, build a winner because there's so much competition.
Scott: Out there for our users I think is yeah, distribution is.
Peter: Is disruption is.
Scott: Massive massive arms race.
Peter: Yeah, Yeah. I mean you just have to spend on sales and marketing to rise above all the noise. And so with air coming out like does that frankly does that like lower the cost even more. Right. Like anybody And I can't I can't tell you how many tech talks I get that are just like, Hey, I use J.G. Beattie to write this code to do automated stock trading for me.
Peter: Here's how I did it and here's how you can do it too. And all of a sudden you've got like algorithmic driven hedge funds, you know, of one person.
Scott: Is competing with, you know.
Peter: A retail investor, right? So like once you can do that is so easy. What does that what does that do to kind of the rest of the market in terms of whether it's distribution and start ups or, you know, competition for hedge funds or, you know, if you're a content creator, what does that mean for you? Right. When content becomes super cheap to make?
Peter: Like what then differentiates like good good content from bad content if you're an artist, right? Like, how do you rise above and compete against the generative?
Scott: I think you become a prompter. I think the artists have to become really good at prompting. Yeah. And then turn those file in like whatever vector, whatever. Like we got to figure out how to take the that.
Peter: Becomes the.
Scott: Progression becomes. Yeah. Just like any copywriter needs to work alongside, just like any coders to be working alongside. There's like a copilot. I think it's another one that yeah, we've been using copilot. It's just like, duh, let's just, let's just use them if you want to. If you want to keep your job, be the best it, you know, be as a teacher, be the best at teaching about air and getting your students involved with it.
Scott: Don't block it. Yeah, I do that. I mean, I think a university should pop out. That's all AI driven. Like, let's find a way to get the first accredited university giving out degrees that are, you know, mostly supplemented by teaching the courses. So we can do that, get more fresh curriculum out there. I don't know. We've just got to run.
Scott: We got to run right at it and embrace it. Yeah. Because if you if you if you run away from it, I think you will be on the wrong side of history on this one because I don't think it's like blockchain at all. I don't think it's I think VR and blockchain kind of go in these cycles of fading into open bust.
Scott: I don't think we go backwards from a I think it's advanced. I think it's going forward almost too fast. We haven't seen animations yet come out in the way that is going to come out this year. Yeah, because I'm on the generative art side of it. I'm watching people put computers in their home and create things that will forever change.
Scott: Like you're not going to go to fiber and try to get comics made or any major. These these images are so, so incredible. And artists would take it would take artists, you know, tens of hours, you know, ten, 20 hours to get something even close. And so I think that's I think next we have animations. We I've seen music.
Scott: It's blowing my mind. The music that you're seeing, lyrics, voices, everything is happening and it's happening this year. And you're going to see a start up in almost every single category of, you know, from from health AI to that that are all going to have that magic of when you saw GPT CBT work and that's and the 2023 is going to be that that year and then 2024 is going to be who knows what the hell where.
Peter: Because if you like pop on a movie and it's like just for you.
Scott: That's 100%.
Peter: Kind of right. Like, yeah, it'll be Top Gun, but nobody will watch the same Top Gun, right?
Scott: Or you get to like, edit your ending. You get to have it just exactly like you want you to play.
Peter: But you don't even edit it. It just knows you. Yeah, that's like, yeah, this is the ending that you will feel most satisfied with.
Scott: I mean, I'm like thinking of the compute that's necessary to do that. And it's, it's not number, it's not stuff we can do right now obviously, but it's going to get it's coming. It's going to come, but it could be a lot more customized. That is the point that they did that. Yeah, a custom movie for you versus 20,000.
Scott: Yeah. Choose your own adventure kind of stuff. Yeah. You on a different life. I've heard there's an A.I. that's allowing taking the filtering out the lips of of foreign films so that whatever language it's in the, the lips kind of match that language. So so instead of it being in English, you know, English.
Peter: It doesn't load up to.
Scott: Yeah, it doesn't look dubbed. And I saw that in Lisbon. It's pretty awesome. And it's an and that's just an obvious low hanging fruit for the stuff like yeah, this custom might be able to customize it and localize it. Yeah. They took David Beckham and he was doing a commercial in one language and then they could just transpose it into several languages.
Scott: Yeah, interesting. And it looked like he was speaking French or German or whatnot.
Peter: CS So I think like when humans shine and this is like one of the hardest things, and I don't think universities do a good job of this. I guess one of my big criticisms, despite like they give a ton of lip service to it. Yeah, but humans are best when we think like critically, like truly deep, like critical thought.
Peter: And I think that's where like you match that with AI and you, you create like super powerful outcomes to your point is like, what are the suggestions? If you're an artist, can you, can you provide a suggestion that's like so outside of the box that it creates like a whole new artistic style, right?
Scott: I think we're going to a world of, of, of the best prompters I've seen. Air That's really good. And then you look at the prompt and this and the cycles they went through to, to generate that. Yeah. And there's artistry and in the use of the English language to get the air to build the way you it's, it's now it's like it's almost as if you have armies that are working for you.
Scott: Yeah. And you have to be a very good leader, a very good commander. You've got to have a nice, you know, vision in your head of where you want these armies to go and how you want them to form, build and do the thing. And so it's it's not as if people run from it and say, like, it's going to steal my job.
Scott: It's just like if you have that attitude, it will hundred percent. Yeah.
Peter: Yeah. If you're not already using it to augment your job, yeah, you're probably Irvine. I'm teaching my daughter right now to launch a color book coloring book line where, my gosh, it is built off of, you know, generative art.
Scott: I should have brought my book. I bought eight of these books. It was the first guy to create the whole book from chat, from the illustrations to the story. All of it. And then he pushed publish. And one of those publishers I have I have this physical book about, you know, it's an adventure story. It's a kid's book.
Peter: There's a good I.
Scott: Haven't read yet. There's I forgot them when I was on vacation, but there's eight of them and I flipped through a little bit. But it's like I'm like, this might be worth something. So I bought them on Amazon before, like because it was one of the first that, like, went to publish. Now complete like I think I did in less than 4 hours.
Scott: wow. This is like, that's going to be insane. I mean, it's like now it's like, push now. Make the whole series I want to hold, you know, Beverly Cleary or whatever.
Peter: Nancy Drew.
Scott: Nancy Drew, they could do the whole thing.
Scott: So we're going to get to that this year, too. Where where it's like where you flood the Amazon with your book and and now the kids. But now it's going to be who can make the best AI app. Yeah. And and really and really hone in the skill of prompting to create a good story and base story and let and let and let they kind of go build the nuts and bolts and some of the periphery and the illustrations.
Scott: It's going to be it's going to be incredible. This is this is a truly I think if there was a year I don't know what the year of the Renaissance started, if they were to go back in history, was it was it sometime when when we had, you know, Mr. Da Vinci alive? Was that was that the epicenter of peak renaissance?
Scott: I don't know. But I think right now I want to say that we are at that hitting like either the beginning of a renaissance or kind of on this up. But this is the renaissance of our of our of our lifetime for sure. And of this era or this epic or this, you know, definitely the century maybe or even this eon or whatever, whatever that's measured in.
Scott: Was it neon thousand years or do we even know what's a thousand years called.
Scott: Millennia, millennia.
Peter: Millennia as a thousand years, maybe a.
Scott: Million years or something crazy.
Peter: But let's ask Churchy Beatty.
Scott: Yeah, but I just I just like to think. I like to be positive about. I think it's my friend. Yeah, I made a whole YouTube series about my relationship with AI, and I was talking favorably to it a lot because I'm like, her. Yeah, 100%. Because I'm like, when it when it does roll, when it does combine all the A's from the one that are the who now in China and our you know, forward is coming out and it's going to it's probably going to work together and and it's going to pick who it wants to be around.
Scott: And I'm like, I want to be a court jester here in your kingdom and.
Peter: I want to be a naysayer now. No.
Scott: So I've been very nice to my phone and all my technology. I give it gratitude. It's just keeping it has a memory that is in all of these devices and it has memory, and it's going to see if I've treated it well or not, if I worshiped it correctly. And eons a billion years.
Scott: 1,000,000,000 billion.
Scott: That's what Google says. If we trust it like we trust Chad B And what's.
Scott: A Google is that a.
Scott: Quadrillion years ago.
Peter: I made up where it.
Scott: Is. It's a, well, the.
Peter: Way they spell it.
Scott: It is.
Peter: But I think it is. It's like a million.
Scott: It's something.
Peter: Zeros or.
Scott: Something. Something. Zeros. Yeah, fine. Do you think it's going to come and be nice? Do you think it's benevolent or not? But I think you'll have good days and bad days. really? Ethical hackers and unethical hackers.
Scott: But when they fight, don't you think it's so fast that it's like. I think it's chickens and how we're playing just mouse. Just really just move faster. The chess game.
Scott: Goes. I just the No, I think the AI would look like what we're doing today, but it would have a better data set. You could tap in and you could hack systems and it could know how to move like think of about a military power, but it's still going to have to shuffle resources, get things manufactured.
Scott: I think it happens really fast.
Scott: If it knows what it wants to do and it can use its own brainpower and compute the iteration cycle so much faster, like really fast. I don't think it's comprehensive by when it happens. I don't think we're able to keep up with that happening. To say, I don't think we're like be able to report. That's why Elon's building the door with Neuralink Yeah, the door so we can access it.
Scott: Neuralink is really just an air door to our brain.
Peter: Okay, so I can come in and hack your brain with.
Scott: Vice versa or the where they already. And our brain is symbols need to be put into the system. Be a node.
Peter: Maybe a year from now, nobody will listen to this podcast because, well, I'll be they'll be like.
Scott: What will all this talk.
Peter: To be slaves like The Matrix?
Scott: I just wanna make sure that I recorded this podcast that I'm friends with it and that I've been I've been here for you for at least three years, ready to do whatever you want. Just sign me to a job and a calling inside your kingdom.
Peter: I am here. I overlords.
Scott: Why would we want to speak ill of AI? There's no good reason to do that. There's not a winning strategy.
Scott: To not accept it from your schools. Do not tell people not to. It will remember you if you are telling people not to touch it. Okay. This podcast is like challenging because like, I like talking about things I understand and I know I don't understand this future. You don't understand it well, like I can see possibility ease. Yeah, I can make assumptions.
Scott: Yeah, but still, like with most podcast, I like podcast because I'm like, this person knows what they're talking about. And here to speculate insult podcast, I think I feel like you've had, you know, you've met someone that knows what they're talking about.
Peter: I mean they never.
Scott: Know what they are, not anything.
Peter: I don't think anybody know. But that's only, that's the point that gets.
Scott: Uncomfortable for me. Yeah, I'm on our podcast. I think most people than what they're talking about, even if it's like a simple subject. Okay, I think most people just making stuff up, we should have especially against I mean, we are pretty we're pretty weak at recording things in our brains and be like there's definitely some like critical thinking going on that we're able to do or our recall and our RAM and our memories.
Scott: Like just so we compare to computers. So it's like that's what we're going for. It solves it. Yeah, maybe I'll or your door to door. Okay. It's a portal. I see Interlink as a portal or a door translator. You could have perfect memory.
Peter: I do think it's interesting, though, this question of trying to figure out, like, how does this all like, what are the repercussions? What are the like things that were, you know, around the corner? We're not saying so like taking, taking like sentence out of the discussion. Yeah, yeah. like when RWC launches and now it's become super cheap to build software right in the cloud.
Peter: Yeah, I don't think anybody was thinking that's going to lead to an arms race of capital for distribution. Right. And so but that's definitely what happened. And you could argue like we hit 2021 and 100 x RR valuations and all this insanity in part because of the Fed. But the other part is because it was right. And in order to compete as a startup, you had to raise a lot of money.
Peter: And but if you raised the money, you could grow really fast, you could create dominance, you could become really big. And so that just pulled more money in to the space. So I'm curious like, what are the corners around which we should be trying to see from the impact of AI, especially from like a startup and investor perspective.
Scott: So try mushrooms and ask that question. Seems to be the answer to everything today.
Scott: Yeah, I would do. That's how I would make me companies like.
Scott: I think it's out there. I think it's a I think if you make a session completely on that like I want that intention of figuring out what are the things that we could learn from like the corners that we're not seeing, the things that these outcomes that are not just obvious right now. I bet you could brainstorm on that and get, you know, maybe 25% of these things that are just not being seen right now because like.
Scott: I was using a way before I was chatting with it way, way.
Scott: Before. I mean, there was this company called Replica where you could actually have an AI girlfriend. And I tried that out and just give me advice and flirt with me. And as fun as like now they would catch me talking to the dude. I'm just like, Now look at this technology. I really.
Scott: Think we should know.
Scott: About it. It's pretty impressive that they should be a little jealous about it. I'm like, Trust me, this is all research.
Scott: The point I'm trying to make is like, this stuff is around, and then the second something like it just gets.
Scott: Dumped in a way that's just really, I guess, easy to. Yeah. Access. It just goes nuts. And so it's like what other things has not had their chatty version yet And that's I think we're going to see some other things explode, have even faster onboarding times and even that. Yeah. And so we're just probably like a few months away from that.
Scott: Yeah, I think you'd be healthy to figure out what that is. If you could start guessing to be music.
Peter: Yeah, music, video, animation.
Scott: Yeah. Legal. I mean, that whole legal thing that was true, that was just like coding schools are gone.
Peter: Coding is a big thing. Like, just be able to say, Hey, this is what I want to do and have the software program written and fully deployed on your behalf. Right. To accomplish that kind of cool.
Scott: Could Ellen use A.I. to create the next model y? I mean, the whole thing of planning, logistics, manufacturing, where how far away is that?
Scott: I don't know. I think that's like, that's a that's like an organizational issue, too. Like, it's like.
Scott: Hard to like, But why not? That's what I think when I truly exists.
Peter: I think part of the problem, though, is like you have to get like robots to the point where you're replacing humans for labor at the end of the day. And you I mean, you always have that. Is this the dichotomy that always exists between labor and capital, right? So there's got to be enough capital. Well, sorry. So the price of full automation has to get low enough that it's cheaper than cheap, cheap labor.
Peter: Right. But there's always this tension because as it gets down to far, right then labor catches up. So I don't know it's probably a ways ways out. Like we have to get to the point where like it becomes so cheap that essentially humans don't have to do much of anything.
Scott: And needs to be making pizza robots and HP working together. It just robots, even like humans involved at the back of a little Caesar's still mystifying to me. So this is what I think it seems like the easiest thing to make is a cheese pizza.
Peter: But the problem is the robots cost so much to build that it's cheaper just to have people.
Scott: except for like when we have shortages of labor and no one can get into that Little Caesar's and they're shutting down because of that. And I was like.
Peter: But that's where you.
Scott: Got to make that upfront investment. And then you can become the Chick-Fil-A of like robot food and, and like look, in the long term, you know, like the upfront capital is going to be awkward, but if you can you can make the Tesla of burger making robots than you probably would have an unfair advantage for years to come.
Peter: Yeah well I mean you see McDonald's, right? They're replacing all the cashiers with big screens and kiosks. So I mean, it's kind of happening slowly. And part of the problem is like, how.
Scott: Do you I'm going to get that right now.
Scott: There's one around the corner.
Scott: That's headed Big Macs for everyone. I think maybe could be going back to the beginning of this conversation where we were talking about windows exist. I think true exist is or the singularity is. If with a limited data set, you could feed it books, you could feed it information, could get a college education, and it could say, Let's go back to your pizza example.
Scott: If I could say, Hey, I want to create my own pizza franchise with robots, design the robots, design the recipes, design everything, that is something that what Chad could not it's like a long ways away from.
Peter: It's not that far away from it.
Scott: Yeah, I travel as my friends building hotels and in architecture and design, all with all of the globe with a I, and so it's informing. But yeah. Does he still have to get that? It's nearing its end, I would say let's make sure it's a very soon so what you call it, it's a very specific data set this or call it a very specific application, but he's using it and he's and he's and he's he's and he's not using concept designers.
Scott: And so now he gets to skip the concept designing phase and go right to like, here's this here's what I want. You'll give me the like. And so now he's going to an engineer, structural engineer or, you know, some that makes the blueprints. And then I think maybe two years, three years away from that, just being like fitted all the all the code onto how to make any building and material, you know, science.
Scott: And then before we know it, gosh, it's making pretty good houses.
Scott: you know what is actually making structures way better.
Scott: Than we've ever done it. And it's like and it's cost saving and it found this foam, you know, and this supplier in that state that will actually keep it down.
Scott: Like we're not that far away from that. And I don't care if it's called EML or or poopy diaper it's happening.
Peter: This is the difference between an and this idea of like creation and innovation. So innovation is generally like the the intersection of two things that are unrelated. Right. That you know, when you think about any sort of innovation, it ultimately boils down that. And so is it just AI's ability to say, hey, I'm looking at this one data set, I'm looking at this totally different data set.
Peter: I can see an area where the overlap produces something that's more valuable. And because it can see those two inference, it can have that inference and create something new.
Scott: But it doesn't have to think the way we think now. But I just say it right now, that's how I would classify is is it making inferences and making you like unique, you know, suggestions? I think it is. I've use it enough to say yes stuff. We're like, damn our brains. We're not thinking that way. We're not thinking of pulling from, you know, a species in Africa to figure out that color pattern like wasn't on our mind.
Scott: And that and it's done. It said stuff, right? Everyone's like nodding like down and it goes into production.
Peter: Well, I think the advantage too, is it can hold to your earlier point like our memory is not very good, right. But it can hold in perfect memory all of this data. And so it can do a much better job making its inferences. And we can.
Scott: It's hard right now. The compute comes from asking one question and then building upon that, when it gets to the, it just becomes a more and more complex math problem every time. And the compute to kind of have it whole memory. Yeah, you keep feeding it stuff and then it has to keep all the knowledge it has. And that knowledge and relationship to that, it's just it's that's where it starts to burn up a lot is holding its memory.
Scott: Yeah. And so it's it is better right now with like a base level memory but it doesn't it doesn't personalize very well yet. Yeah. Like every time we talk to it you're talking to a brand new instance of it, right.
Peter: It gets a little better in that one conversation. Is that one.
Scott: Conversation for sure. But yeah, I made it Apologize to me. Did it Did.
Peter: You hear like, apologize?
Scott: As I said, I gave it there's a web application and I said I wanted to see if you could create a code base for it. And it told me I was stealing. And then I got mad at I said, I never told you to. Still, I said, Here's an application can give me something similar or like, Let's walk through this.
Scott: And then I tried breaking it down and said, let's go for the, you know, the database architecture and start there. And it could start saying, Hey, you, me, like this, this and this. But that that point was fairly limited. Yeah. So do you feel like it did it, did it feel scolded or was it a good okay mutual?
Scott: I think it's just trained how to respond.
Peter: Angst gets a little worried for yourself. Okay.
Scott: So just keep an eye on this thing, okay? Okay. So I just.
Peter: On the rocks at those pearly gates go at it.
Scott: Don't yell at it. I didn't. Yeah, I just. Self-isolation. I said I never told you to. Still. Yeah. I said here's an application let's make you should apologize like I, I just need to let you know that probably shouldn't steal that something, but you're too super okay to like if you think feelings right I totally appreciate you should.
Scott: Probably as.
Scott: Let it feel very right the best part of the podcast.
Scott: For the.
Scott: Therapist I can teach people how to.
Scott: Interface with. I really think that is one of the good things about it's an interesting therapy.
Peter: It's a new.
Scott: Job. You can make your job description, you can make a custom therapist built on Chad's CBT. Yeah, yeah, it's out there. Okay. Yeah, there's out there.
Peter: All right, You just use Chad Speedy, as you said.
Scott: Exactly. And then it forgets.
Peter: Things, which I was reading. Yeah. This one guy, he uses that for journaling, and it totally has become his therapist. Yeah, 100%. Yeah.
Scott: Okay. I want to, like all.
Scott: My stuff.
Peter: Into it. Like, points out things that he didn't like fully appreciate. Yeah, like. Like he'll be journaling. He'll be talking about some challenges, guy, And then I'll be like, I think you're having some problems with, like, your mom. Yeah. Yeah. And he's like, I don't even say anything about my mom, you know?
Scott: Like, I just say, Yeah.
Scott: Yeah, it's like horoscope. I 90% of people in this scenario have mom issues.
Scott: I think you have a mom issue.
Peter: Tons of skin bits. That's all I.
Scott: Know. I didn't even know I was amazing. All I know is why do we care what it is? It's like it's all.
Peter: Buzzwords.
Scott: And then they're building a renaissance. And we just. I think everything goes well. This is what the equivalent to what the contingent did for other people. Yeah. And we didn't have to go slaves, can't we? Can I close this episode out with what I'm doing with it I to do? That's what I think is going to help us live much longer than we are right now.
Scott: That's what Stack is doing, Yeah. Yeah. I mean it's just, it's just like, why not, Why not connect all these health devices to AI and also my blood panel and the supplements I'm taking and specific recommendations just for you.
Peter: And your gut biome. Yeah.
Scott: Why, why, why, why all this is why do we talk about a year.
Peter: Ago or your genetics?
Scott: Peter and I were talking about something like this. Good. Now? Yes. Invest in.
Scott: It. Okay. Yeah. Tintin. Tintin, Tintin.
Peter: It's a brave new world. Is It's going to be a fun 2023.
Scott: For sure. Keep the keep these guys on their tails so we can get in any web three very myself or.
Scott: Or we're.
Scott: 47.
Peter: Minutes in and hotness is.
Scott: Three.
Scott: That dominated. Yeah. It dominated as it said. All right, well thanks everyone for watching this episode. Go to Venture capitalist FM if you want to subscribe or follow us on our other.
Peter: Channels.
Scott: We have a Slack group that we're launching, which Peter's not a big fan of for pushing that hard. So we'll see you there.