Exploring the Technological Advancements of ChatGPT with Puneet Gupta, Founder & CEO of Amberflo

(Intro Narrator at 00:00:00) Today, we're talking to Puneet from Amberflo about the capabilities, history, and workplace impact of ChatGPT. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:16) I actually used ChatGPT to write some interview questions. It was pretty fascinating. So I took what your history was, your bio on LinkedIn, and I said, "Hey, ChatGPT, I'm talking to him today. Here's his background. Here's his bio." And then I said, "This is the type of startup he's building." I just told it a bunch of stuff about you. So now they know about you. The AI does.

(Puneet at 00:00:39) I don't know. One is that's so cool, and I'm not surprised. You know what? We played around with it too. I've played around with it, and first, what it spit out was just insanely amazing. You know, I asked just stuff like, "Why would anybody use usage-based pricing?" or "Why is usage-based pricing better?" or "How would you get started?" I think something along those lines. And man, it literally just spit out line by line, pretty much a playbook. So anyway, we can talk about that. But my take is, as exciting and as powerful as it is, I have a feeling that they had to use Chrome and LinkedIn data as a huge source of input into this ChatGPT. There's no way an AI engine today can not just bring the right content together, but also formulate it in the way that it's spitting out content. Anyway, that's my thesis. They have a behind-the-scenes view that Google does not. And considering that, you know, ChatGPT is Microsoft and Reid Hoffman sponsored and all that good stuff, I think there's no way it's all because some of the stuff—what AI engine? I mean, AI engine is—we know what it is, but I think what they have built is a huge classification mechanism. You know, once you have the content, yes, in real time a question comes up which blocks to bring together and how to structure them, and that's magic right there itself. But there's no way that there's some AI engine out there that's articulating what is the value of usage-based pricing relative to subscription. It has to be curated by something that's already out there. If you look at LinkedIn profiles—I've written about this stuff, and I've written articles on LinkedIn, and then LinkedIn has specific groups where people talk about this stuff, right? So if you take that input into an AI/ML engine, yeah, now you've got a good source of input that you can do something better on top. So this is, I think, something that has been off limits to Google. So I'm just amazed nobody has really talked about this or even asked about this. Hey, you know, what's going on potentially behind the scenes? But what's your take? I think they have to have access to the LinkedIn data.

(Joel Beasley at 00:03:00) Yeah. I haven't asked it about LinkedIn stuff. I don't know where they're getting their information. And you're exactly right. One of the reasons why Copilot by GitHub is so great is because Microsoft bought GitHub. They used ChatGPT. They trained it on the entirety of the public repos for GitHub. It makes sense that I'm sure somewhere in LinkedIn there's a line in the privacy policy that's going to allow it.

(Puneet at 00:03:29) Yep. I for sure think, and I think this is—if you just kind of work your way backwards, I think just how the group came together. I mean, who's behind it? I mean, all the power to them. I think it's ultimately a good thing and a force for good, but I think they do have some competitive advantage that they have some behind-the-scenes view and can do things that others do not.

(Joel Beasley at 00:03:50) Oh, 100%. Yeah. And honestly, I kind of want them to because America as a country, which I love—I'm a huge fan of America. It's given me everything, right? But we are falling behind as a country as far as technological innovation and AI initiatives, publicly at least. I don't know what's going on behind the scenes governmentally, but publicly we're falling behind. You'll see spots like Canada and China advancing way farther in their programs and AI. So the idea that this is happening, you know, partly on American soil by a U.S. citizen—I think Musk is technically a U.S. citizen—that's great because then we get the open one. I think we can all trust Musk a little bit because I've been watching him for ten years, and I love the idea when I first heard about it. Let's make this—it's going to happen anyway. He was scared. He ran around. Have you heard the story about him running around to the government trying to tell them to stop this? And then when they said no, he ultimately just said, "All right, we're going to build it first."

(Puneet at 00:04:50) Yeah. Yeah. Yeah. I'd heard him talk about that, you know, if it's not under some guidelines, it could kind of get out of control. I remember him talking about it in that context. And I think this is true. I mean, if you look at ChatGPT, it shocked a lot of people, what it is capable of doing. I think it's a huge leap forward, you know, because prior to ChatGPT, it has a little bit of this, a little bit of that, a little bit of enhancement here, there, whatever you might have in eCommerce or, you know—but I think ChatGPT is just—it was sort of this big leap, right, that we look for in tech, sort of an exponential growth, a step function growth. So, yeah, I think I don't know where he stands on this, but is he part of this group, is he? Elon Musk?

(Joel Beasley at 00:05:40) He either was asked to be a part of it and didn't do it, or he's actively involved in it. I'm fairly certain he was one of the lead investors at the beginning.

(Puneet at 00:05:48) Interesting. Interesting. Yeah. Well, okay. So then that's good. That's good. Yeah. And by the way, just—you touched on something, you know, as—America. I mean, I'll give you my perspective as an immigrant. Just to be on the record, I mean, there's no country on this planet, despite all the problems that we have, that is still the most immigrant-friendly and a platform for immigrants like me to come in and achieve a dream and really think outside the box. I mean, bar none. Nothing comes close when you put it in those terms. So there's still, you know, a bunch of things that are in flux right now with all the things that are happening with the divide and all that, but that's still a little disturbing.

(Joel Beasley at 00:06:33) Yeah. And I don't have a ton of global experience with immigration, but I do have several friends that have immigrated. And some just have their green card, some have obtained citizenship, and they are honestly far more vocal than any other Americans I know about immigration issues. Right? They get frustrated that they went through all the paperwork and the time and the process to get in and that other people are getting in in different ways. And so for me, I just let them run with that because they have more credibility than I do.

(Puneet at 00:07:09) Fair enough. Yeah. No, there's that. That is for sure, right? You know, the processes, I think, definitely can be improved. I think it's actually fallen behind over the years. But, you know, it is what it is. Yeah, I think that for sure. And I also hear stories about folks who are kind of caught up on the short end of it, and it can be, you know, pretty brutal. It can really have an impact. So I think, yeah, there's definitely a need for improvement there. But if you've come out of it—and I'm a living, you know, case study as there are several, many hundreds, thousands, tens of thousands—yeah, there's—I mean, the ecosystem that exists here, there's just nowhere else. And I've lived around the world. You know, I'm originally from India, but my parents' family moved to Europe when I was 13 years old. So I finished up my middle school and high school in Holland, in The Netherlands. So I lived there for a few years. And then I came to the U.S. when I was the ripe age of 18, started my undergrad at Ohio State. And it's never been—never looked back, you know?

(Joel Beasley at 00:08:06) When you were doing your undergrad, was it in the technology field?

(Puneet at 00:08:10) It was. Yeah. I went to Ohio State, did my computer science and engineering, and that was great. I, you know, I was four years undergrad at Ohio State, and then I stuck around for another couple of years, worked in the greater Columbus area. And then back in the day, there was a company called SGI, Silicon Graphics. That was a big hot ticket in town in the Bay Area. They're the ones who recruited me out of Ohio, brought me to the San Francisco Bay Area, and that's when sort of my journey started.

(Joel Beasley at 00:08:39) And now you're in usage-based pricing. And so when I mean, AI for OpenAI, that's going to be usage-based. Correct? That seems the most logical.

(Puneet at 00:08:49) Seems to be the most logical. I'd be surprised if it isn't. Let's put it that way. Yeah. Particularly for something like AI/ML.

(Joel Beasley at 00:08:57) Do you think that because it's—well, first of all, do you think it's going to become commoditized, like a race to the bottom for pennies for the ChatGPT technology?

(Puneet at 00:09:08) I don't think so. I think, to be honest with you, I think there's not enough information for one of the aspects that we just discussed. I mean, if they do have a disproportionate competitive advantage—if they have exclusive access to something like LinkedIn data—then I don't think it's a race to the bottom. Then they just have a competitive edge. I mean, you know, because I think the wealth of information that today sits in a platform like LinkedIn, I think that's leaps and bounds, orders of magnitude different than anything else. Nothing comes close to it. You can go crawl on anything else. You just—it's minuscule compared to, I think, what LinkedIn has. I think just because it comes pre-formatted, pre-structured in a way that gives you a huge, huge leg up for what ChatGPT and other platforms like ChatGPT would want to do. So if that checks out, then I think it's not a race to the bottom because I think these folks who are behind ChatGPT and, you know, folks like Reid Hoffman, I think they can hold it up and everybody else is going to try to play catch-up.

(Joel Beasley at 00:10:13) Do you think Google's going to mine the data it has? Because one quick story about them is I've registered on TikTok because I was doing an interview with the CSO there, and it was showing me stuff it thought that men wanted to see. And so I got introduced—even though I didn't like the content—I got introduced to this concept of these short videos, right? I hadn't really seen that before. And then I went over into YouTube Shorts, and it knew me so well. So TikTok, even after I tried to tell it, it just didn't know me at all. It kept showing me irrelevant stuff. But I go over to YouTube, and Google's known me for fifteen years. And they just were showing me things left and right, exactly what I wanted. So now, you know, I watch—I look at YouTube, you know, once or twice a week at the Shorts, and it's amazing. Yeah. Now they have all that data, right? Do you think that they're able to tap into that data and make a competitor to GPT-3?

(Puneet at 00:11:11) See, here's my take. I think things are going to get interesting. I'll answer that in just a second, but, you know, I want to throw out what my lens is, and I'd love to kind of get your feedback, see what you think. See, even before getting into ChatGPT, what has happened over the last five, ten years, and what collectively we've been calling the FAANG set of companies, right? Okay. Which is, you know, it's Facebook, Apple, Netflix, Google/Alphabet, and Amazon. I think we could reasonably say the FAANG has, you know, collectively—everything that a human does, you and I, you know, 80% of that interaction, whatever can be captured is being captured by FAANG. Okay? As you said. Okay. You know, because what is our daily thing? You know, it's Gmail, email, YouTube, TV content, right? And then professional networks like LinkedIn and then social networking space, whatever you have it. So this information is already out there. Now what I'm seeing is I sort of view within FAANG sort of two nexuses. You have the camp of Facebook and LinkedIn. I think I put them together because of their Microsoft connection. And then you have Google and their properties. And I think maybe the holdout, sort of the nonaligned, if you would, would be Amazon. Okay? But they have sort of their own things within each one of those two, right? But I can tell you, we can unpack the Amazon aspect of it too because I spent a few years there. So I can kind of share with you sort of what their thinking is, how they look at the world. It's very different from everybody else in the FAANG. Okay? So this has been going on and Google clearly had been leading for the five, ten years, right? It's just unstoppable. I mean, as you look at their revenue and their margin and what they're—they're just printing money hand over fist on the backs of advertising. So one part of me says, you know, I think ChatGPT is good because I think somebody can finally come up with that could potentially offer something to an average consumer out there as a matter of tools or value that could be just as profound and useful and in the flow of things, of daily things, as search is. Okay? And I think we're just starting to see some examples of this, but it'll be interesting to see where it goes if it sticks, right? So ChatGPT, like you said, I just—you know, they've got my profile. They probably told you, "Here are the five questions you're asking me." That's huge, right? Because otherwise, you're going to have to hire some folks to kind of go do research and, you know, spend some money and do all of that. So there's tremendous value there. And I said, it's still within FAANG. So somebody else could not have done it. You know, I could not have raised $100 million from a VC and said, "I'll go build ChatGPT" unless I had a Reid Hoffman backing me and I had a connection to Microsoft, and I could crawl LinkedIn and I could crawl Facebook, right? So I think what we're going to find is that if this is what it plays out, I think what ChatGPT will be able to give you will be, on a value metric, perhaps over time more than what Google can give you. I think, to me, Google is still more in the realm of entertainment. Certainly, they do their properties. You know, they give you free tools, but then on the back of it is all advertisement. But I think ChatGPT is perhaps opening that canvas a little bit where work can get done at a productive level that we have not seen before. Okay. So I think there's a new vector coming into this whole equation. Google has always been doing—what you are seeing is, you know, that's sort of their bread and butter. They've been able to connect these vectors, you know, where you spend time on the backs of the cookies, right? Cookie tracking, all that. That's really what's going on, what you just outlined. That's basically it. But that is sort of age-old technology. They've been at it for the last five, ten years. I feel there's a little bit of pushback coming into that. If you've been reading about, you know, particularly the European Union pushing back on cookies on the browser—what would we call it—third-party cookies, right? So blocking bills. So I think there's a little bit of push there. And I say that because that aspect, Joel, I think has—is nearing its maturity curve from a—right. I think people are ultimately sort of getting tired. Governments are discovering the odds. It's kind of gotten a little too much, right. And I think we need to sort of temper it down or need to put some guardrails around it.

(Puneet at 00:15:41) But here comes ChatGPT, totally revolutionary in its early innings. I think there's a long way to go before people will want to stop it because, quite frankly, consumers want it. They're going to consume it. They want to see. They're excited by it.

(Puneet at 00:15:57) So, anyways, that's sort of my take. So what I'm saying at the end of it is I think the tools for me are a little bit orthogonal. You know, Google will continue to do what it is, but I think it's out of tricks when it comes to something like ChatGPT.

(Joel Beasley at 00:16:10) Do you think ChatGPT can create new information from existing datasets and make inferences?

(Puneet at 00:16:19) I think eventually. I'm a little skeptical that it's already there today. I don't think it is. I mean, I'd be just shocked if it is because, you know, this is like achieving—this is like Ray Kurzweil achieving singularity. Right?

(Puneet at 00:16:38) I think I'd be just shocked that it just happened. And, like, when did these guys start? Three years back, four years back? No longer. Thereabouts.

(Joel Beasley at 00:16:45) Right? So, yeah. At least, yeah. So definitely I heard about it after I started the show, and I started it six years ago.

(Joel Beasley at 00:16:52) So—

(Puneet at 00:16:53) Yeah. So, no, man. I just, I think, you know, like, if that's what's happening right now, I'd be just shocked. I don't think they're there yet. But I do believe that they can get there because they figured out that they have such a huge step in—you know, they launched on this great launchpad on the backs of LinkedIn and Facebook data. Yes.

(Puneet at 00:17:13) I think on the backs of that, as long as the input continues to come in—you know what I mean? So as long as, you know, you and I continue to pay our way into LinkedIn and make sure that, you know, we are updating our profiles, and, you know, if I'm launching a company to do usage-based pricing, I'm writing my blogs and articles and all of that, I think they could eventually get there.

(Joel Beasley at 00:17:35) I see a whole app store emerging. Right? Like, let's say we've got a company that is, you know, collecting like a ZoomInfo or—Josh can remind me of what the other company is that I talked to some really bright people from.

(Intro Narrator at 00:17:51) Dun & Bradstreet.

(Joel Beasley at 00:17:53) Dun & Bradstreet.

(Puneet at 00:17:54) I didn't meet you.

(Joel Beasley at 00:17:55) Yeah. They have these massive datasets. Right? And they're constantly scraping and updating and processing APIs and moving insanely amounts—like, massive amounts of data around the world every day. And I could imagine that they would let—they could let an instance of their ChatGPT consume all of that, and then I can activate that as a source inside of ChatGPT for a fee.

(Joel Beasley at 00:18:20) And then now my GPT knows all of this new stuff. Yeah. Do you think that's going to emerge?

(Puneet at 00:18:26) That's a fascinating thought, idea. Now the way you've outlined it, 100%, because I think if that is their goal—and I see why not, which sort of lends credibility to what I'm saying—is that you kind of have to bring some semblance of accurate data. Right? So as we, you know, as we say, you know, in God we trust. Everyone else brings data. So I think, you know, for you to leverage, at least today, what I see from ChatGPT.

(Puneet at 00:18:56) Yes. So if you're a ZoomInfo, you already have a directory, sort of white pages of folks, their titles, where they work, and whatever that source is. I'm sure it's traditional sources. They have people chroming things. Whatever is accessible, you know, Wall Street, you know, whatever we have it. So they have that repository.

(Puneet at 00:19:16) Yes. Now if I can layer the ChatGPT engine on top of this repository, yes, I do believe that they could really streamline, first, their own operations, but further enhance how quickly and the quality of what they can output to their users. I think that is pretty much a slam dunk. So I think what you just said, I hadn't thought of it that way, but yes, a marketplace of sorts, right, an app store. Yeah.

(Puneet at 00:19:44) Certainly makes sense to leverage the ChatGPT engine in that way.

(Joel Beasley at 00:19:48) I'm glad you're knowledgeable about this because I've been thinking about this quite a bit over the past two weeks. So I grew up being a software engineer, did that for seventeen years until the show got popular enough to where this became my full-time job. And then we started making shows for other companies, so that's the trajectory of it. But I still follow the technology and get to talk to great people like you doing interesting things. And I had never really been concerned about any technology.

(Joel Beasley at 00:20:17) I mean, I see technology come and they'll abolish an industry, and it's always really clear. Like, okay, this technology is emerging, maybe an accounting efficiency software. All right, that'll probably drop accounting team sizes at large corporations because they're automating mundane tasks.

(Joel Beasley at 00:20:33) But when you see something like ChatGPT come out, it's brilliant. Like, I was trying to struggle to say how it's not intelligent, and I'll give you an example. I've got kids under the age of five. Right? So I'm really into watching how people learn from a young age recently.

(Joel Beasley at 00:20:52) And so if you give it information and ask it for conclusions, like, it'll come to conclusions on the information, and then it'll remember it in the future. And then you can talk to it with a lack of context. You can use "this" and "he" and "that," and it'll figure out who you were talking about previously in the conversation. And so I'm struggling to figure out if you put the concept of soul aside—like, spirit and soul aside—if you put that over there, how is it not just intelligence?

(Puneet at 00:21:24) See, again, my take is—and maybe, you know, I'm just biased, and again, you know, both you and I come from technology backgrounds. So my immediate reaction was when I look at it, my brain starts to go and starts to unpack it immediately. Okay. Now, yes, I would say I'm not the de facto expert in AI/ML, but I've done a little bit of that. You know, we've all kind of tinkered around with it.

(Puneet at 00:21:47) So I'm generally aware of what the capabilities are, at least today, where that sits. And one of the things in AI/ML that before ChatGPT came out is, you know, garbage in, garbage out. I mean, if you're just feeding the model garbage data, it's not just suddenly going to give you some kind of insights that, you know—so if you look at all the popular AI models that are out there, they basically talk about that. Right? So they can do some interesting things and how they connect the dots within that data.

(Puneet at 00:22:20) But if the data is just garbage, everything else that's going to come out of it is not going to be any better. So that's my understanding of AI/ML, that it is not just, you know, some magical thing, you know, black box that is going to spit out some, you know, euphoria-type or, you know, some kind of a utopian aspect. So having said that, my view is this. I think back to your thing of, you know, is it intelligence? I think, Joel, what has happened is, like you said, in the last five to ten years, the base has been created. And what I mean by that is you and I have divulged everything about us somewhere already.

(Puneet at 00:23:08) Okay? And I think the missing link was, which I think ChatGPT has brilliantly done, has finally brought it together in a way and made it so that you can now interact with it in a meaningful slash real-time way. So there is some technology leap there. I certainly call that out. But the data has already been out there. So what you just said, right, when your kids type in or as you type in some information, it can actually output a conclusion.

(Puneet at 00:23:43) I don't know. My first reaction is that somebody probably has already spoken about that conclusion on some forum in LinkedIn or Facebook or something like that. Somebody has expressed their opinion on this topic. Now it wasn't exactly in the same words in perhaps what you asked. It wasn't exactly said in the same words that ChatGPT put it out there for you.

(Puneet at 00:24:05) But I think the general, you know, sense has been out there. That is what got fed into the model. I don't think the intelligence aspect has quite emerged yet. So that's, you see what I'm saying? I think it's still there. It's there in the repository. Yeah. And what it's doing is it's picking it up at the right time, at the right place, and maybe merging it with other narratives that have been indexed.

(Puneet at 00:24:39) So it's aware of the context, and it's, you know, it's brilliantly tagging how information is there. I'm losing it in general terms. I'm sure it's more than that, which is what AI/ML is. But I still believe that it's what it is today, just based on stuff that's already been said.

(Joel Beasley at 00:24:49) 100%. It's consuming the collective intelligence of humanity and then being able to query it. But it can also learn. Right? And so I was just trying to figure out, with spirit and soul aside, how do you differentiate organic human intelligence from computational intelligence?

(Joel Beasley at 00:25:07) What is the test or, you know, in law, you have to meet certain requirements for conviction and things like that? Like, what is the attribute that defines the difference between computational intelligence and human intelligence?

(Puneet at 00:25:25) I think if you think about what AI/ML is—and as you just said, it can learn—I think the lines are going to get blurry. I think we're already seeing some of that. Like, you know, I come across these videos where somebody's talking, you know, these snippets, and they tell you, well, this seems—well, this was computer generated. Right? Yeah.

(Puneet at 00:25:46) Unless somebody had told me that, there was no way I was figuring that out. So, yes, because somebody may have already said it, so the technology is able to superimpose, map it, put it into context, where a little bit of our own psychology—right?—because, I'm sure there's a school of thought on this where, you know, the mind will immediately categorize something, right, whether it's real or not, based on our experiences.

(Puneet at 00:26:15) And because now we are seeing things that look real, our mind immediately says, yeah, that is real. So I think the distinction between intelligence and the human aspect of, you know, soul and that innate ability to decide right from wrong, just as one of the examples—I think the lines are going to get blurry because it's going to play with our mind, and we just have not encountered that. Our mind has not been developed to pause even when we see something that our mind tells us is real. We just don't have, I think, the DNA to, even when the mind is telling us that it's real, pause and then question the mind.

(Puneet at 00:26:55) So I think, and that could be a little bit of, certainly, uncharted territory, and I think we'll have to tread that carefully.

(Joel Beasley at 00:27:04) I agree. This is why my mind is so unsettled. Typically, I'll see a technology, I'll do the instant categorization, I'll see the clear path, and that's it. My mind understands it, and it wraps it up in a nice bow. Trying to figure out how GPT is going to impact the world is crazy.

(Joel Beasley at 00:27:23) So one of the things I thought about was, let's say, in general, it can replace a professional, like, in the next five years. Let's give it some time. In the next five years, it can replace a professional in 80% of professions that are a knowledge worker up to three years of experience. Right? Well, what does that do to the job market, and what does that do to us as, you know, business owners?

(Joel Beasley at 00:27:51) Right? Because it's our job to constantly be watching the market because it's competitive and it's a jungle out here, and you can just get sideswiped out of nowhere. So it's important to be aware of the new technologies coming. Right? You don't want to be the Blockbuster.

(Puneet at 00:28:05) Yeah.

(Joel Beasley at 00:28:06) But how do you think about that?

(Puneet at 00:28:10) Yeah. You know, sitting here today, I'm still holding on to my optimistic lens. Okay? And I think maybe we're just right at that inflection point where I think my own views will change in the next year or two, but I'm not there yet, and here's the reason why. So as you said, you know, look, I've also seen it. You know, I've been around twenty-five years plus in tech.

(Puneet at 00:28:33) I've seen waves come and go. I've seen, you know, the bubbles burst and all that. But I think all said and done, if I compare it to what life was in the nineties to where we are today, I think by and large, generally people have benefited from technology. I think it has uplifted all of us. And I think we've also kind of maintained that. Now, okay, so here we are seeing perhaps a little bit of a step function leap, right?

(Puneet at 00:28:56) There's a compounded effect. Maybe we are seeing with this ChatGPT. But at least the historical data tells us that an ecosystem typically follows that tends to—yes, we set the equilibrium, but generally moves things forward positively in driving overall economics and the standard of living. So I wouldn't be able to tell you exactly what that would look like, ChatGPT post. I think if I had that crystal ball, I think that'd be worth something.

(Puneet at 00:29:32) But I think you may have already alluded to something like that. So you just mentioned that, okay, I foresee an app store emerging on the backs of that. Well, there you go. Okay.

(Puneet at 00:29:44) If there's an app store emerging on the backs of that, then shouldn't we please for a second pause and say, well, you know, app store emerged on the backs of Apple iPhone, and that jewel basically created a whole economy in of itself. Gave a lot of people new jobs and things like that. So that is where I hold on to my optimistic lens. Okay? Now if it happens to be such a compounding effect as we're seeing that there could be, right, where it skips a few steps on the ladder, then there may be a bigger turmoil for that interim period before we settle into that equilibrium.

(Puneet at 00:30:22) And, yes, that could be somewhat challenging.

(Joel Beasley at 00:30:25) That is exactly my thought. It's we've always done this—horses, industrial revolutions, all of that—but it was stretched out. Right? Now I believe it could decimate industries overnight or in a quarter. And, yes, I 100% agree with you, and I'm a very optimistic person on this.

(Joel Beasley at 00:30:46) I think ultimately, we—this all ends up really well. Right? But I'm curious, and I'm going to ask some more people from maybe, like, government or policy, somebody that specializes in this, is what do we do when it happens so frequently, so fast that the unemployment happens and is so high because we're waiting for the—or unless if there's some principle where the ecosystem follows the innovation at the same rate of speed that the innovation occurs. Maybe that's something. I don't know if that's true.

(Puneet at 00:31:19) Yeah. I think, you know, I think we have probably seen that come up. We have talked about it in the abstract so far. Right? I think as you mentioned, you know, that was Elon Musk talking about, you know, what, five years, six years ago, he said, no, it could be dangerous. There should be some guardrails. I think we're getting to that juncture. So, yes, I think ChatGPT, if you are, you know, if you were that one person, as you said, if any one of these companies that used to kind of do this, curate this manually, yes, there's an immediate threat and impact. And because it could be on a large scale, then I think somebody will have to think about some guardrails. But that's kind of where the debate, I think, gets even further interesting because, you know, guardrails are really the antithesis of innovation.

(Puneet at 00:32:09) Right? And one of the reasons why the area, generally speaking, you know, United States and the area in particular where innovation has thrived is because those guardrails are not there, and it lets people kind of be creative and dream of things and go and chase those. And not everything was successful, but I think in large part because that autonomy existed. So I think that is a big question. I'm personally of the belief that, you know, free enterprise is the way to go.

(Puneet at 00:32:41) I think if you bring in government too soon, you can basically put the brakes on something because, you know, just kind of the fun goes away. Right? But there's a moral aspect to this is what you're calling out. And I think we are going to find ourselves again in some uncharted territory where, yes, I mean, if you ask the person who's going to be impacted, I mean, we're talking livelihood.

(Puneet at 00:33:04) Right? So yeah, I think it's interesting. And I think from a moral lens, it makes sense. You know, put some guardrails. It cannot happen too soon because that would just cause a level of unrest that, you know, we have to maintain a certain kind of peace and quiet and decorum that cannot be challenged on the back.

(Puneet at 00:33:28) So no matter what kind of moral productivity we can get, so that's sort of my thoughts.

(Joel Beasley at 00:33:33) Yeah, I agree with you. And for me, as I said, I am an optimistic person. I think it's all gonna work out, and new jobs will come just as people were scared that the Internet would kill the mail system. It did the exact opposite.

(Joel Beasley at 00:33:49) And I think it takes definitely a lot of faith, you know, to get through that. And so the way I calm my mind down—well, first of all, I'm stealing your "ecosystem follows innovation" and resetting the equilibrium, because when you said that, it was incredibly clear. So that helps me feel good about processing this as a whole. Right? So that is brilliant.

(Joel Beasley at 00:34:13) As far as the people and their jobs and all of that, you're exactly right. I don't think government really ever steps in too early. They're almost always on a delay. And I do agree though, if you bring that in and you start forming boards, you're just gonna stifle innovation and make everything super weird. But yeah, you've given me quite a bit to think about in regards to ChatGPT.

(Joel Beasley at 00:34:37) I enjoy—you've got a very sharp perspective, and it's different than mine, so I like that.

(Puneet at 00:34:43) Yeah, very cool. Yeah, look. Yeah.

(Puneet at 00:34:44) That was just my impression. First, I was just blown away, but my mind went immediately into overdrive. I said, man, what's going on here? Right? So it's not something that's just been, you know, dropped from the sky and onto an awesome that you can unpack.

(Puneet at 00:34:57) I mean, at the end of the day, there is technology. So yeah, I mean, that's sort of my view. I think it'd be worth—I hope somebody next—I don't know if you might have an opportunity to find out, you know, what are the sources behind ChatGPT? I think they have something exclusive there. Yeah.

(Puneet at 00:35:14) They have to.

(Joel Beasley at 00:35:15) Mhmm. I think it's gonna do well. And then GPT-4, once it's connected to the Internet and you can—look, the moment this thing becomes real time, it's gonna change even faster. Right?

(Puneet at 00:35:26) Yeah, I know for sure. And I think that's where they're going because, I mean, $10 billion, I was just thinking for a second. Okay, why? Yeah. I mean, they got all the money in the world, but why $10 billion? I think a lot of it is now going to just about scaling the infrastructure. I mean, I don't think we're done with cloud computing. Yeah. So now they are—now the tentacles kinda go out.

(Puneet at 00:35:45) I saw that graph. I didn't compare it to ChatGPT-3 versus 4. I think 3 was just like a blip, and 4 is like a, you know, almost this giant of a ball in terms of the data and how much information it'll be able to process.

(Joel Beasley at 00:35:59) Did you see the one about ChatGPT and the guy is asking him a question, and he says, no, my wife knows this. And it's like, oh, your wife? I guess your wife is right. Have you seen that one?

(Puneet at 00:36:09) No. Oh my god. She knows.

(Joel Beasley at 00:36:12) I don't have it exactly on top of my head, but he's asking the question. He's like, no, that's wrong. And he's like, my wife says so. And then ChatGPT says, well, I guess, you know, I'm only trained up to 2021, so I guess your wife is right.

(Intro Narrator at 00:36:24) Yeah. And I think specifically, it was like a question about, like, what two plus two equals. And then the guy was arguing it was five because his wife told him or something like that.

(Joel Beasley at 00:36:32) And then GPT is like, well, I guess you're right. I guess it is five.

(Puneet at 00:36:36) Yeah. That's why it's trained. You don't wanna argue there.

(Joel Beasley at 00:36:38) I know. Right? Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn, or send me an email [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.