Loading summary
Transcript9 lines
- [00:00]
Jaden
We have some major news this week out of Anthropic that Connor and I are dying to tell you about. So there's a few things. One of the biggest things is that Claude's new LLM is beating OpenAI on certain benchmarks. GPT 4. This is actually very hard to do. It's kind of been interesting and something I've almost laughed about over the last year, that everyone that comes out with an AI model, they, like, put the benchmarks of themselves versus GPT 3.5, because no one even wants to compare themselves to GPT 4. We're now getting to a point where these models are getting better than GPT4. Of course, OpenAI is probably going to launch GPT5 soon, so, you know, like, you know, hold your horses. But this is amazing. Claude is doing it. Claude's an amazing model. In addition to this, they announced something absolutely insane, which is a 2.5 or 2.75, $2.75 billion investment from Amazon. So, yeah, Connor, what have you been seeing here?
- [00:51]
Connor
Jaden, where do you even begin on this? Anthropic is having itself a week. You know, this past week it was bonkers. I mean, so first of all, Jaden, we were talking about it was kind of a weird, quiet week for a little while there. And then Amazon kind of comes out and tells the world that they're doing 2.75 billion into anthropic. And they might have been reading the tea leaves a little bit just because you and I and a lot of other people got access to Claude 3 whenever that was a couple weeks ago. And I'm hearing just, just from anecdotally, but certainly I'm experiencing Claude 3 is just so, so good. And. But here's the interesting thing, right? A lot of different places to take this because Amazon, this was its largest venture investment ever. And when you're talking about that with Amazon, like, that's pretty serious. And so just the implications of that for, like, what they're offering, like enterprise users and through Bedrock and through, you know, their cloud provider, all that kind of stuff, like, they're clearly, you know, making a really, really big bet in Anthropic. Of course, Google has also made a big bet frame Anthropic early on, I think. Again, I think Google is invested. I think it was like 300 million back when the valuation of Anthropic was very low. But the thing that I'm again, sort of like the most interested in, in a way, is this chatbot arena leaderboard. And Jaden, you Kind of hit the nail on the head, right? Like, it's. You see these things and everybody's kind of creeping ahead of everybody else. But ultimately, you know, it's interesting. It gives you a good benchmark. ChatGPT4 came out a year ago, guys, so obviously OpenAI is still, you know, kind of the OG on this. But my thing for this is when I saw Claude 3 up there, it was like one of these moments in the movie where it's like, wait, there's a new leader for the first time in 40 years? Or it just felt like that moment. And so when I saw that, it. That was interesting. But even maybe more interesting was, that's Opus, right? That's the paid version, which I pay for. But then there was also Haiku, which is their small. They have Opus Sama and Haiku, and. And Haiku was beating an old version of GPT4 as well. And when you think about the efficiency of all this, and Jaden, I know with AI Box, by the way, I love the new AI Box hat. Gotta get that merch out there. But something like, you know, for developers, when they see something like this, a small, really reliable model that's way more efficient and way cheaper, even that feels really exciting. No?
- [03:13]
Jaden
Yeah, 100%. I think for me, what is amazing, like you mentioned is, of course, we're having these models that are getting better and better and more and more powerful. So this is. This is cool. And yeah, seeing Claude beat GPT4, that's also very cool. We've actually talked about this on the podcast before. We've both been very impressed with Claude as of recent, and, like, who knows what's coming down the pipe. Sam Altman recently said that GPT4 sucks. So, you know, that's kind of funny as they're obviously prepping for five, so who knows what's on the horizon? But as it stands today, I think your best bet is if you want the best results with the best tone. I found it's. It's Claude, and that's. That's surprising. I feel like I'm very impressed by the company, right? They. They've. They've been able to stay very relevant and up there, so that's amazing. But the other thing that you brought up that I'm really excited about is the size difference. A lot of times right now we're just talking about who's the best and benchmarking the best models, and people aren't talking as much about, like, what is the most efficient model, what model can run on a device on your phone, like, and I think that there's, there's some people working on this that it's very exciting. So when I see this from, from some of anthropic other models, this is, this is very exciting for me being, you know, having a software company, we're going to be embedding all of these models in. This is something we look at, we think about is, you know, how much does it cost to, to run each of these models because people inevitably make products and apps on the AI box platform. Um, it's a factor like people are gonna be. People will be using like, not the number one model. Right. So you would, you might use theoretically. Right. Like a lower version of, of Claude because it's cheaper to run. And so people take that into consideration when they're building products. Um, and so, yeah, I think it's an interesting element that I'm, I'm very excited about for sure.
- [04:52]
Connor
Yeah, same. And it's, it's just funny because, you know, I think Ethan Mollick had pointed, the Wharton professor had pointed out that Bloomberg had spent something like $10 million on an LLM that, you know, can't even beat like 3.5 or something like that. And it's, it's this, it's this thing where I remember the original, not, maybe not original, but last year's interview with Lex Friedman and Sam Altman where they were talking about GPT4 and saying, Gosh, this kind of looks like AGI almost. And now Sam Altman is here a year later saying, like, it totally sucks, which is a pretty big. Now, this is also what I love. Just side sidebar. What I really like about Sam Altman is his, the quickness in which he talks about things not working. So you and I have talked about with the plugins on ChatGPT and how pretty quickly after that Altman said, I just don't know if there's a good product market fit because people seem to want GPT technology built into their, their Canva or their kayak or whatever, but they don't need to go to a separate thing. And I think he was dead on. So just the ability for him to call out his own product. Now, first of all, it shows unbelievable self confidence because he must know that GPT5 is coming in. I've heard May or June or something like that. But with Claude, yeah, you and I, you hit the nail on the head about it being having great tone. And this is the thing that I always, you know, this is the AI Applied podcast, right? And so we always try to think, like, how does this apply to you? And I'll just say, how it applies to me. Jade and I each create a ton of content. I think there's a lot of people out there who do that too. So when people say, oh, is it beating, you know, GPT4? I love those banners, too. I love headlines. I'll put them up as clickbait, like, oh, it's killing. Blah, blah, blah. Like, I just. Because I love it.
- [06:34]
Jaden
Yeah.
- [06:34]
Connor
But in reality. In reality, GPT4 doesn't go away even now because it has so much more. So it's not about replacing. Right. It's about, you know, because you don't have code interpreter or Dall E or, you know, the ability to browse the web or any of that stuff with Claude. What you do have with Claude is an absolutely uncanny tool that can match your brand and match your voice. And also, I don't know, Jaden, like, it's. It's funny, I never thought that this would matter to me that much, but the user interface Claude just feels kind of, like, warmer and kind of a little easier on the eyes and things like that. I know that kind of sounds a little dumb, but, like, you know, when you're staring at the screen as much as we are, I think it kind of matters. GPT is starting to feel a little harsh. Maybe I always have it in dark mode, but anyway, I guess for you. Do you think of these things when you think about what you use? How much are you breaking up? Like, you use this tool for this, but this tool for this. How do you think about that?
- [07:29]
Jaden
No, at 100% is. Is coming down to that. And I think that maybe that's where we go in the future and maybe there's not. And I actually think that's a good future. It's not just one dominant player that everyone uses for everything. I personally, I have a bunch of different tasks that, yeah, I have like, a tab of ChatGPT open in my browser that I'm like, I go there and use. But then there's like, other tasks. Like, for example, when you and I are like, oh, you know, like, we want some more context on the story we're going to talk about today. And we'll just go to Perplexity and, like, we'll get some bullet points and like, oh, okay, well, what about this? And we're like, sharing links and looking at it. Like that kind of, like, research thing. Yeah, that's 100% perplexity. And it's going to Be, it's going to be GPT4 for, like, me helping me write, like, descriptions for my podcasts or something. So I use, yeah. Different tools for different things. I think this is going to be more common and I think this is a very healthy future. I don't want this all to be consolidated into, you know, one company dominating. I want there to be a lot of competition. We're going to get better products as consumers when this happens. As far as Anthropic goes, I think this is going to be a big player to keep an eye on. Obviously, we've all been saying this, it's one of the top funded. I realized that over the last year, Anthropic has actually closed five different funding rounds. So there are five different deals are worth about $7.3 billion. So this is very impressive when we consider the fact that OpenAI raised $10 billion from Microsoft for 49% last year. And so, like, they're getting very close to that. Right, 7.3 billion. Now, it's not definitely, it's not like all cash because some of it, it's like they're obviously working with Amazon because they're getting like, you know, cloud. But that was kind of the deal with OpenAI as well. Like, in the $10 billion they got from Microsoft, it's like, yeah, here's $10 billion, but you got to be on Azure and we'll build you custom stuff. But so it's kind of like these cloud providers are also like, dishing out money that they know is going to come back to them. Some people say it's like, you know, it's, it's like rigging their books or whatever because it's like, here's some money, but you definitely have to give it back to us. And then they counted all this revenue, but, you know, whatever. That's an accountant's problem, not mine. But I do think, yeah, no big player. They're raising almost as much money. And the thing that I like about Anthropic is, and I hate to be, I don't know, a pessimist or whatever, the thing that I do like about Anthropic is that they were, they were scrappy enough and they were able to go and, you know, cobble together a measly $7.3 billion, but they got it from different companies. And I actually think this is a big play because they got 300 million from Google's Cloud platform. And I actually like this because when you have, when you have this situation, you don't have 49% of anthropic necessarily being owned by one company that it feels like can have maybe more control over the company than you'd like. And that's the only thing with OpenAI that's kind of tricky is like with 49% being owned by Microsoft, they can make a lot of decisions. They have a lot of sway into what's going on there. And so I think that's something that a lot of people are paying attention to.
- [10:16]
Connor
Yeah, I think you're right. I mean, because we don't, we don't associate Anthropic with Amazon quite as readily. I mean, we do a little bit, but not quite so much as OpenAI and, and Microsoft. And it's. OpenAI and Microsoft are, by the way, huge, huge fan, obviously. But it's also like it's starting to feel a little boy band ish, where it's like, you know, Justin Timberlake wants to break off and do his own. You know what I mean? Like, they're, it's starting to feel like they're frenemies in a way. Right? Just because, you know, how they share revenue and all that kind of stuff is getting like a little strange to me. But. But yeah, with Anthropic, I mean, I agree. So do you remember, Jaden, we did a story not so long ago where Anthropic was projecting. I might have this number wrong. So apologies, folks, don't take my wor for this. But like, they were projecting something like revenue at like 800 million for, for the year. I don't think it was air. It was like, it was like for 20, 24, whatever it was. And I think I, and you and others were like, where are they getting this number? Right, because this is before Claude 3. They had lost the advantage, their competitive advantage around Context Window, which was their big thing. I don't think people think enough about the ethical frameworks and safety and stuff like that to, to do that. I don't think we're at that stage of the movement with AI and that we are with, you know, buying organic and stuff like that or the green movement. I just don't think we're there with AI where that's really going to impact consumers. I think consumers are much more concerned with obviously speed, accuracy and efficiency. And I'm thinking that, okay, all of a sudden that 800 million feels like, oh, yeah, they're totally going to do this. But I also think that. And somebody on my LinkedIn timeline, I think it was Jason Wilde, like from Microsoft, was formerly from Microsoft, saying This was like, it's like, it's like the guy running like the. Or the woman running like the 10K. It's like if you know you're not going to win, you sprint in the beginning. And it feels like anthropic right now is sprinting. They're getting a name for themselves. Everything else because GPT5 is just around the corner. Anyway, like, you know, I'll wrap this up because I know we're going a little long here. Jaden, talking about this, but like the thing that's exciting to me is that if you are listening to this podcast, you're listening to AI Applied. What we always encourage you to do is take, don't get wedded to one, you know, thing, one tool. You know, I mean like GPT4 has been awesome. But look, I mean Claude, 30 for brand, voice, all that kind of stuff. Try it out, get in there. See what tools work for you. You don't need a million at the moment. Perplexity. Claude, GPT are kind of my go tos. And now you.com too. Shout out to you.com spoke with Richard Socer or some about some of that stuff. Anyway, I think, Jaden, the way that you and I see it, I don't want to speak for you is look, keep an open mind, test these tools out. You'll find what works for you.
- [12:51]
Jaden
Yep, 100%. And I definitely have different tasks and I use different things for and anytime and like lest you get hung up on the fact that. But I don't want four tools that all cost $20 a month. Yes, you actually do because all of them save you so much time and energy. So like just make the investment. Especially if you're using this for anything related to your career or your job. Like, this is the best investment. If it came down to it. Cancel your Netflix. I get Claude, you can quote me on that. Anyways, thank you so much for tuning into the podcast today. If you got any value out of it, if you learned anything new, please consider following us, subscribing and leaving us a review. Wherever you get your podcast, really, we really, really appreciate it. Apple, Spotify, wherever it helps us bring on amazing guests. We recently just talked to some phenomenal CEOs and we're bringing more on in the future so they'll look at their reviews. They want to hear what you guys have to say about it if you're enjoying the podcast. So it means a lot. But we hope that you all have an amazing rest of your day and.