Generative AI Might Just A Be Feature, Crypto is Back, Substack’s Nazi Problem

0:00Generative AI might just be a feature. Crypto is back baby. The CEO that wasn't real and the state of Substack Nazi problem. All that and more coming up right after this.

0:40Welcome to Big Technology Podcast Friday edition, where we break down the news in our traditional cool-headed and nuanced format. We have a very big show today covering some massive topics from generative AI to crypto, and of course, the travails of Substack. Here, as always, is Ranjan Roy from Margins. Ranjan, welcome. How are you doing? It's 2024, and crypto and generative AI remain not on top of mind, but I'm excited to get into it. We'll debate whether crypto and AI are two sides of the same coin or not, as some people are talking about. But first, let's get right into the generative AI discussion.

1:15Because, and let me start with this post from Fred Wilson, who runs Union Square Partners, a venture capital firm. He says, we're moving into the application era of AI. And this is from his prediction post for 2024. Well, in 2023, everyone was rightly focused on the large language models like open AI, Anthropic, Gemini, Lama, etc. We will now see AI first applications emerge in 2024 that will start to move the focus and the conversation up the stack. And we will see legacy applications embrace AI to make their products better and to remain competitive with the AI first disruptors. So I agree pretty strongly with the second half of Fred's take here, but not necessarily with the first half, because it does seem to me, and we're going to go through some examples here, that generative AI in the near term, at least, is going to be an add-on or a bolt-on to existing products, maybe just a feature.

2:14And we're not, you know, we came out with ChatGPT, that was the big application. It was like, oh, the age of AI application is here, but we haven't seen any since then. So what do you think? Is generative AI going to be an application itself or an add-on? I think generative AI presents a new paradigm where we can't cleanly separate out features and, you know, pure applications. And the reason I think this is because generative AI, I think it will and should show up as a feature. There's news around Robin AI's legal co-pilot getting$26 million in venture, like this idea. And again, legal documentation and the whole legal field feels to me like the most ripe for disruption field that exists.

2:59So I think we'll start out where existing companies and products essentially bolt on generative AI features. But I think then generative AI changes the entire way those features work. The example I always go back to is when Benedict Evans, I remember, had written that right now Microsoft is approaching generative AI as an add-on to Excel, but what if generative AI completely rethinks the way we think about spreadsheets at all? And I do think it will, and maybe Microsoft will guide us in that direction and we'll do it still in Excel and it'll look differently. But I think generative AI is going to start as a feature but end up transforming all these applications.

3:41Right, and the question is what that timeline might look like because if you think about the near term, everything that we've seen, right, ChatGPT comes out November 2022. We're sitting here in January 2024, and we're not anywhere close to that reimagination of Excel. Like, this is what it seems like it was going to look like to me, whether it's this thing, this Robin AI. And by the way, so what you do with Robin AI is it's a Microsoft Word add-in, and it's a company that's, you know, just got$26 million in funding and$43 million total. and you can use it to create contracts in minutes. You can review.

4:16This is an article talking about the funding. I think it's from TechCrunch. Review existing contracts using plain language prompts and identify and propose edits. They say that it reduces the time to create contracts by 80 % and it reduces costs by 75%. So it's a co-pilot, really, like the idea. And by the way, that's almost obvious that there's been this language co-pilot around AI, which basically says this is going to be something that helps you with the stuff that you're doing currently and not reinventing what you're doing. I'm really on the side of this assistive or this feature part of generative AI.

4:56I don't see it reinventing things like Excel or even search yet. So here's why I think this could be different. And in terms of timeline, what I'm about to say is definitely not happening in the next year or two, But the whole point of a contract always was, and for anyone, you run your own business now, I've run my own business. It was always the most painful thing, especially when you're a solo entrepreneur of trying to sift through legalese and feeling scared about these things. And all those words on that piece of paper are some kind of essentially a defense mechanism. It's some kind of programming that creates that idea.

5:36but once that stuff essentially becomes automated, the whole point of what a contract looks like changes. Maybe this is a little out there, but again, it's all those words are on the piece of paper to fit a specific way that we do business and people read, humans read that piece of paper and try to make a judgment around that. But if all that stuff is essentially programmed and being automated anyway, then why do you even need a written contract? So Robin AI, you know, and again, I don't know what that looks like in the future, but the, and I've thought about this with a lot of content that like, you already see it when an SEO driven article gets abstracted down to a Google snippet.

6:16We never really needed that article to begin with. I don't think we need a lot of this contract language. It's all just been there because it's been built in a certain way. Right. And they basically like, there's been talk about proof of work and we might cover this We'll get into crypto. That was supposed to take care of this with smart contracts by now. It hasn't, but I went through that whole thing and didn't even say smart contract. Yeah. Wow. Very impressive. Well, you'll have plenty of opportunity coming up. But it is interesting that even if it is a feature, right, it can change things.

6:49And, you know, it sort of goes back and it can reshuffle the competitive balance of different applications and fields and maybe even the mechanizations in which business is done in certain fields. And I kind of go back to the Snapchat Plus example, which we talked about a little bit last week, where I made this bold claim. And I just wrote about this in Big Technology, thanks to your nudging, actually, about how this is like the first hit generative AI product after ChatGPT. And the point is you go to Snapchat, not necessarily. And by the way, I redownloaded Snapchat this week and I've been playing around with it.

7:25I got a notification. Your friend, this has really been a week of reawakening for me in terms of a few apps that I haven't used in a while. Snapchat and Coinbase and we'll get to Coinbase in a minute. But yeah, it's a super fun app and you don't come to use the AI features. Like you wouldn't go to use the MyAI bot, like you would go to use a chat GPT. But they have that, they have image generation, They have image expansion, which helps you figure out what's outside of the frame, what the AI thinks is outside the frame, and dreams, which you take a bunch of selfies, and it puts you in these dream scenarios like old Hollywood or a movie poster or an alternative universe.

8:07And they're actually quite good. And the thing is that it just enhances the experience. It doesn't reinvent Snapchat. What it does is it makes everything you do on Snapchat better. And that's why it's worth the$3.99 a month. Yeah, I mean, and that's exactly it. That it starts as a feature, but even if you think about it, like disappearing messages on Snapchat was a feature, but it changed the way people message. It changed even sending photos that in the past, sending a photo as a means of communication, it needed to be a quote unquote good photo. But once that disappearing feature came in, it completely changed the entire paradigm of communication.

8:45And obviously that sounds a little grandiose, but I think it did, that people no longer felt like a photo was this like, you know, eternal thing that's important and has to look good. It was just this is happening right now. And instead of writing it, I'll send you a photo. So that's why I think what they're doing and making generative AI accessible, I think, is going to totally change the way people communicate. and that'll happen across whether they get it, whether meta comes in and figures this out on Messenger. I think it will change the way people do things. Again, when you have an idea in your head, I think it was Rex Woodbury, if you read him, he said generative AI, it's an engine of imagination and that stuck with me.

9:30Is that like you have a thought in your head, now instead of describing it to someone, you just spin up a photo and send it to them. Yeah, it's pretty amazing. And another thing that might happen here is that the feature aspect of generative AI might pave the way for people to use these products in those transformative ways. It gets people comfortable with the idea of using it, and then all of a sudden it paves the way for them to start using completely different and transformative type of technology. and you actually, you know, it's kind of funny. Like I made this big proclamation in our document to prep for the show saying it's just a feature and then you drop a story and saying that actually there's funding coming in from Jeff Bezos for a product that will try to revolutionize search through generative AI and not just bolt it on.

10:20And I was like, dang, like the timing that you get to make this counterpoint is so perfect, but it is very interesting. So what is perplexity? so perplexity ai is a standalone app that essentially is trying to reimagine search in the age of generative ai and you know and it's very good and it's definitely been a buzzy on twitter type thing but it's you know jeff bezos just invested into it but it's basically you know you ask it the question as though you're talking to chat gpt but the format it gives you one of the things I really like is it's a lot more citation based. ChatGPT, you get essentially a wall of text.

11:03Perplexity is very good about, you know, structuring the information that's underlying that, which Bard tries to do, but we, I don't even want to talk about Bard. My last few queries have been so bad. Bard has really gone downhill for you, Ranjan. it's it i sorry just very quick rant it got quotes about i asked what are some famous quotes from george washington it made them up it made up i actually went through and this is what i do with my time it gave me 10 quotes and it three of them did not exist how do you hallucinate quotes from george washington like i don't know it's pretty bad google sundar if you're listening get it together, please.

11:47We're rooting for you a little bit. But perplexity. So again, reimagining search, a lot of traction. And I have been using it a lot. In so many queries that you start, you realize that if you're not hoping to necessarily end up on a website, even though it can direct you to a website and just get good information, but if you need to click out, it's already doing it. And it does it in a very clean format. The information from everything I've tried is very good. So it's just what it looks like and how search can act. I think it's a reminder, and especially as search has degraded over the last number of years and the top half of your search results screen are ads, I think it's time to reimagine it, and they're starting to.

12:35So you use perplexity. I feel like after this discussion, I'm going to have to download yet another app because it sounds interesting to me. Are you using it to replace search? How's your experience been with it? Yeah, so definitely to replace search. And I had already started with the chat GPT iOS app, using it to replace basic search queries. Again, if I'm like, if I'm looking for a restaurant or looking for something very up to the moment, I definitely, even though Bard should do it, I do not trust it. Perplexity. So it's more evergreen information. it's definitely my starting point. And again, it just presents things in such a better way than search does.

13:15Yeah. So first of all, I'm intrigued by the idea that there are companies that are trying to reinvent search already through generative AI and Jeff Bezos is investing. So it gives me at least some indication that this could be real. Another thing that happened in this story, as I read down, jumped out at me, which is that it's being invested in by top Google people, including Susan, well, Susan Wajiki, who's no longer the CEO of YouTube, but the former YouTube CEO, and Jeff Dean, who's a senior vice president on AI at Google. Jeff Dean is investing in perplexity? Yeah. This is from the Wall Street Journal.

13:51They both made personal investments before the recent round of funding. I mean, this is, I don't understand how that happens. Sundar, get your house in order, man. Come on. I mean, it is AI search. How do you have the guy working on AI in a search company invest in a competitor? I didn't even know that that was half possible. That cannot be a lie, Sundar. Come on, man. But yeah, yeah. No, this is, it is. And another very interesting thing, though. So I've been using Perplexity a few months. So Microsoft just launched an iOS app for Copilot. And it's interesting because you had already talked about, I think, from a branding standpoint.

14:31And I don't know if they're going to be able to own the term, even maybe from a copyright perspective. But I agree. Copilot is the best way to describe this technology to most people. And the fact that that's what they call it in the app and in the feature and all their existing pieces of software. But so as I've been talking up perplexity, I've played with Copilot for the last week. And it's pretty much as good. And it does things in almost the exact same way. Yeah, it's excellent. This is what I've been saying on the show for months at this point, that I am a Bing boy, and it's now called Copilot, but Bing Chat is excellent.

15:11Yeah, and okay, I guess previously Bing Chat, one had, and I remember at first they made you like download Microsoft Edge, which I did not want to do. But just even again, I guess this is a reminder that like user interface and all of these things matters so much. I've always believed ChatGPT's greatest trick was the fact that it kind of streams the letters and pretends it's writing to you rather than, even though if you ever use the API, it can return it as a block of text in far less time. But it creates the illusion of thinking happening. And so user experience and interface definitely matter in all of these.

15:55And the thing is, Copilot is good. it's it's a little it's a little more crowded and busy than perplexity and uh but it's still very good and again the way they present all the information but but i think this also gets us back into the question of are these heavily funded buzzy startups what chance like how do they fight against microsoft and google and apple and whoever else because again here you have the buzz one of the buzziest startups in the valley funded by Bezos and even Google's own AI search people. Yet Copilot basically can replicate the quality, at least from what I've seen, in almost the exact same UI right away.

16:40Does that mean they just kind of disappear as Microsoft pushes Copilot across every word in Excel and Office user? Exactly. I mean, so this is from the TechCrunch story. It says, with Copilot, you're getting access to OpenAI's GPT-4 technology for free, which is pretty significant because OpenAI's GPT app runs GPT-3.5 technology and charges for access to GPT-4. You also get access to Dolly 3 for free. So it really is quite remarkable that Microsoft, which has put all this money into OpenAI, is effectively offering something better for cheaper for free. Sorry, that, you're right. Right. That's actually a far more salient point.

17:22Like, what does that mean in this whole drama that OpenAI, majority of their revenue, vast majority, and we've talked about consumer versus enterprise a lot, is consumer and chat GPT plus for 20 bucks a month. And if I can get that free on Microsoft Copilot right now, like for them to do that to their own investment, are they doing that because they've lost faith after the whole Altman drama? That is interesting. There was always going to be tension in this partnership, always, right? They're offering similar products. ChatGPT is, you know, potentially competing with some Microsoft products. But the fact that Microsoft is now offering such a competitive product, I don't know, Ranjan, it seems to me like there's going to be more tension between these two companies.

18:13yeah it's not just a competitive product it's a and again if you use the product better and free it's the the same product that and it's open ai's entire revenue stream for free yes big techs big tech is still playing the big tech game well i feel like if you off if you spend all that money and you get 49 of open ai's revenue but you don't own any of open ai which is the case for microsoft you got to find a way and this might be the way find a way yep yep so let's let's go back to our friend uh fred wilson who in his 2023 post uh talking about the age of ai application is also talking about how we're going to see a big year for crypto and web3 and for a while i thought like okay we're not going to hear about web3 anymore you have the sam bankman freed scam and everybody has gotten wise to the fact that this technology is largely overpromised and was pushed by many hucksters.

19:15But we're starting to see the return of quote-unquote serious people talking about Web3 again. And Wilson says that 2023 was the year Web3 held its ground. Okay, that's a very interesting description of what happened. And 2024 will be the year that regulators and lawmakers come to terms with AI. He continues, AI developed for over 40 years before its coming out party. I think it will take Web3 less than half of that time. Satoshi gave us a playbook to build a decentralized internet stack back in 2008, and I feel quite confident that we will have massive mainstream applications running on this decentralized stack well before 2028.

19:53I think we will see mainstream decentralized applications emerge in 2024 as we now have inexpensive and fast transactions and simpler user interfaces. And this is where he says, AI and Web3 are two sides of the same coin. AI will help make Web3 usable for mainstream applications and Web3 will help us trust AI. Together they will lead to a more powerful, more resilient, more trusted, and more equitable internet. Ranjan, we've been talking about how like AI is not Web3 because it has actual use cases. Is it the case that Web3 is just going to take a little bit more time to get there? Or is this sort of the last gasp from, you know, boosters of a technology that have a lot of money in it and are desperately trying to make it happen?

20:38I mean, it feels a bit like every other, every industry and company has tried to put AI in its name or company description. And that sentence to me felt, it starts to feel a little bit. The part that I definitely took issue with is that AI took 40 years and Web3 will only be half that because we started in 2008. One thing with crypto is I remember hearing endlessly, this is like the internet in 1994. This is like the internet in 1970 already, like anything to try to justify the lack of a killer use case or any kind of use case. What are we now 15, 16 years into it? So I think trying to almost brag that the timescale will be halved for this relative to AI, I think is a bit disingenuous.

21:28the only thing based on what we said earlier, where he says Web3 will help us trust AI, there still is this massive trust gap that will only get worse with generative AI. And provenance and context will become more important than ever. And right now, it's funny because like, you know, is this image real or not? Who made this image? These are the exact questions that Web3 has always promised us it can answer. And also I keep saying Web3 buying into the marketing term, let's call it crypto or, you know, blockchain. I think actually the greatest rebranding ever was blockchain into Web3 because everyone got tired of blockchain as a term.

22:12But that's what it was put. That's what it is supposed to do. Like, and it should be doing that already. Like the idea that there's not a mid journey plugin blockchain slash web 3 plugin that tells you who made what, when it was made, what the prompt was, whatever else. Again, the idea that we're going to see that in 2024 when nothing has happened on that side, I think is wishful thinking. But I also think at least as the promise is written, I mean, it should help us there. So here's another place where I take issue with Wilson, and that's him saying that it took 40 years for AI to have its coming out party.

22:52Well, all along the way, AI has had real use cases that may not have been generative AI, but computer vision and machine learning rankings, natural language processing, all this stuff has been used to power everything from translations to things like Google Photos, figuring out, even to self-driving cars. This is already in primetime technology that just took another leap forward. Whereas with Web3, you're just not quite there. Yeah, I agree. I think that's a good point that you're right. He almost equates AI and Web3, not generative AI specifically. And AI, Google search itself has been rebuilt around AI and machine learning.

23:40So almost every major technological use case, there's, I mean, very often some kind of AI component to it. But I think also, again, it's going to be an interesting year because I think if you've seen Fidelity now has announced the idea of a Bitcoin ETF with a fee of just around 0.39 % and is waiting for SEC approval. So there's energy back. There's energy back in crypto, in Bitcoin, and everyone's trying to set this as a year. Bitcoin's held strong at like, 45k right now. So I don't know. Do you think Bitcoin ends the year higher or lower than 46k? It's always such a crapshoot to predict what the price of Bitcoin is going to be.

24:27But I do think that there's like looking at the fact that Bitcoin is, I am astonished that it's up to, it's like 43 ,000 today. I'm astonished it's back there. I thought it was over. You know, when it went down to 17 ,000, I was like, you know, see it, you'll never reach your, you know the heights again and then i you know so i was researching our segment for today and i was like all right screw it i'll open my coinbase app to see how much i'm down and those numbers were green baby like things look good and it's gonna make people like say oh this crypto thing like i'm already getting fomo for not having invested again at the depths and i'm happy to be in you know and the positive numbers again um but it is it's very interesting that this thing just will not die and it's about to be like you mentioned it's about to be traded through etfs which is going to bring mainstream investor you would imagine more mainstream investor interest in the app in in the um you know in the coin especially as uh it's had this surge i mean it's on the way up and it's going to be listed as an etf you know that's that's um some serious momentum so on one hand, I look at this Web3 thing with some skepticism.

25:38And on the other, I'm just like, well, as long as the price of I mean, that's really what this has been about, right? As long as the price of Bitcoin is as high as it is, it's a little bit hard to dismiss. And maybe we will find use cases for this blockchain technology, as long as the money is there, which will attract more investment and more development. See, this is where I disagree. I think, and I've thought this for a long time, the focus on price is what distracts from any utilization around the technology. And that's why maybe we could have had that use case. But because the focus on price was the core, it brought in the wrong people, they focused on the wrong things.

26:20And I actually think I'm shocked that the Bitcoin, I guess the Bitcoin community, but the Web3 community from everything I've seen is like incredibly excited about the ETF. Because think about what an ETF does. It abstracts away the entire technology and just makes it a vehicle for speculation. And that's literally what the definition is. It gives you exposure without having to touch the underlying asset. And that's exactly what I would think you should not want with this, because the whole point is now you're getting focus on it again, everyone getting involved. Casual investors, their first exposure to this, or maybe their fifth or sixth, is through something that completely says the technology, we don't care about that.

27:07We only care about the price. So I actually think the Bitcoin ETF idea is it's a negative for any potential Web3 development, serious application killer use case. Yeah, this is a good moment also to plug that I I have Albert Wegner, who is Fred Wilson's partner at Union Square Ventures, coming on the show next Wednesday. So I'm going to do that interview on Tuesday, and I'm definitely going to bring this up to him. It should be interesting. So look out for that in your feeds next Wednesday. Oh, we're not done with crypto, folks, because as crypto rises, there will continue to be more and more ridiculous stories about the way that this technology or this speculation has been used to scam people.

27:50So we have two big stories. one we're going to end with but the first let's talk about now because as i've gotten to know ron john one of the things i've learned is that whenever it comes to a big scam or confidence game or con uh this man lights up and he dropped a story about pig butchering in our document this week and i was just like oh boy this is going to be so exciting to talk about because this one is a fun, fun scam. Well, I think we're going to have to say, while my eyes are lit and I'm smiling because this is such a ridiculous story, it's a serious one. So I just finished reading Number Goes Up by Zeke Fox, which honestly was one of the most fun business books I've read in a long time, where it's written in kind of like, almost like a mystery style travel the globe story format, but the underlying business writing is very smart and very good.

28:51And so his focus is on Tether. And we haven't really touched on it that much because I guess last year it wasn't quite the story it was in 21 and 22. But again, this stable coin that underlies so much of the entire crypto infrastructure, and then how Tether, and And then kind of he goes through Tether through many cycles. And he even had interviewed SPF, Sam Bankman-Fried, in person a number of times, covers that, covers the rise and fall of FTX. And in the end, Tether stays standing. And then obviously the question is, where is it being used? Other than pure speculation, why is it being used?