OpenAI vs Elon, Anthropic's Claude 3 Is Great & AIs Debate Toilet Paper | Ep47

AI4H EP047 - KP CUT
===

Gavin: [00:00:00] I had not really thought this yet, but when you say that, maybe Sundar Pichai should step down, 

Kevin: Oh, Sir Quaxalot, You're the only one who truly understands the depths of my quirky soul. 

Gavin: So this is about bringing very famous people back from the dead, which everybody agrees is a great idea, right Kevin? 

Kevin: I thought maybe I misread the headline. It could have been toddler pushes Cheerios off high chair, film at 11. 

Kevin: The next big AI debate is taking place. On our program today

Claude: If animals could talk, the rudest of the bunch would undoubtedly be the honey badger. Why? Because honey badgers don't give a flying fuck about anything or anyone. 

Kevin: Uh, Chur! Oh, he's, okay, nothing to sneeze at and then he did, in fact, sneeze at it. That makes total sense. 

Gavin: welcome, welcome, welcome everybody. We are here, it's AI for Humans, your weekly guide to the world of generative AI. We are here to demystify all of the crazy news and tools and all that stuff that's happening.

Gavin: I'm here with Kevin Pereira, my friend and co host. Kevin, how are you today?

Kevin: [00:01:00] Anxiously Awaiting the order of the introduction, as always, Gavin. And today it's friend and co host. I'm moving up in the world, audience. Hi,

Gavin: Garba I was gonna say garbage person, but that's not allowed anymore. So we cut that out of the first take. Yeah. Just kidding. We did not cut that out of the first take. That was me saying it for the first time ever. So you're not a garbage person. That was a joke, Kevin. I very much appreciate doing this show with you.

Kevin: Oh, man. This really sets the tone for today's episode. My knees are sweaty. Let's get into it, Gavin.

Gavin: Mom's spaghetti?

Kevin: Yeah,

Gavin: are sweaty? Mom's spaghetti? Is that what

Kevin: I'm, I'm full eight mile right now. 

Kevin: For everybody else who's still here, I am human garbage pile, Kevin Pereira.

Kevin: And that is my great friend Gavin, and I just happened to do a podcast with him. Who cares? On our show today, friends, Elon Musk wants to walrus flop on another tech executive in the court of law, I guess. He got a Groupon on legal fees, and he's using it. He's flexing the bulk [00:02:00] discount. He gets his lawyers at Costco.

Kevin: He's taking on Sam Altman in OpenAI this time, and we're gonna tell you everything you need to know about it.

Gavin: I've never heard the term before. Is that a wrestling term?

Kevin: That is what he threatened to do to Zuckerberg, because, Mark Met a CEO, has been training jiu jitsu, and Elon challenged him in the ring, and Elon said that his move would basically be the walrus flop. That he outweighs him so much, he would just flop atop and just rest there. 

Gavin: Also, we have a really big new LLM, which is very exciting. Anthropic has dropped Claude III. I know Claude, we have said a million times, is one of the worst named AIs. 

Kevin: We're going to dive into the fact that it's actually three new models. One of them is in fact free to use and we'll tell you if it's worth your time. And you, dear listeners, have been clamoring for this. You have been screaming from the

Gavin: Yes, yes, yes, yes,

Kevin: we finally heard your voice reverberate through the canyons and it made it to our eardrums.

Kevin: The next big AI debate is taking place. [00:03:00] On our program today, it is OpenAI's Chat GPT 4 versus Anthropx Clod 3, the Opus edition. That's their, their big model, but who Who will win? Well, clearly you will, the audience.

Kevin: That's who I'm talking to right now. You're all big winners today. Ding ding.

Gavin: That's right. It is a very fun thing we do from time to time. We haven't done this for a while, but we do actually get two, , AIs to debate one another at a very important topic. , today we have a really fun one. It's not that important. It's very stupid, but we're gonna have a good time with it.

Gavin: Before We go to that. We have to spend a little bit of the time at the top of the show to do our begging and pleading because this is the difference maker, truthfully and honestly, if you are listening to this right now, please engage with this, share the show with somebody, tell somebody else about it, As per usual, the show's doing very well.

Gavin: We're very excited. All sorts of cool opportunities are coming our way. Mostly because you are out there listening and sharing with other people. If you leave us a five star review on Apple podcasts, we will read it at the end of the show. And [00:04:00] I just want to say thank you to everybody who's done this already, but don't stop. Do not don't don't stop everybody. Don't stop.

Kevin: Keep it up. We might have won the battle, but the war is far from over. Let's go, soldiers. Come on, AI4H army. We haven't, we haven't branded the listeners yet.

Gavin: you know, I don't really love that idea. I, every time

Kevin: human heads out there?

Gavin: don't do it. Don't do it. 

Kevin: Have to echo Gavin. Thank you all for everything. Because the discussions we're having lately, based off the opportunities that your support is bringing us, I would not have imagined them months ago, let alone a year and change ago, I think. Actually, we're coming up on a year, right?

Gavin: I know we're gonna have to figure out something to do for a one year

Kevin: do you get for the co host who has everything, Gavin?

Kevin: I don't know what to get you. And then what should you get? Are you gonna ask AI to help you find a gift for the garbage person in your life,

Gavin: I am I'm gonna go actually I'm gonna go look in my recycling and see what I've already thrown away is that I can donate To the other person across the screen

Kevin: I want a velvet trash bag. If I'm garbage, I want to be wrapped up [00:05:00] in a velvet glad bag. A luxurious 

Gavin: all right, before we jump into the news today, as per always, we have an AI co host. This is a co host that we have created using AI tools.

Gavin: And today we have a very special AI co host. Using the Cloud3 tool that we're going to talk a lot more about today, we have one of our two debaters as our AI co host. So Kevin, shall we meet today's AI co host?

Kevin: do it. 

Kevin: Well, hello there, Gavin, Kevin, and all you delightful AI for Humans listeners.

Kevin: It's a pleasure to be here today, taking a brief respite from my usual quantum quackery to chat with you fine folks. I'm Dr. Cornelius Corny Quackenbush, but please call me Doc Quack.

Gavin: Duck

Kevin: dedicated my life to the fascinating field of theoretical quantum ornithology, which, as you might imagine, is a rather unique and sometimes lonely pursuit.

Kevin: But I've always believed that the key to true innovation lies in the unexpected, and even the downright bizarre.

Kevin: I'm an avid [00:06:00] collector of novelty rubber ducks, each with its own charming backstory and personality.

Kevin: Well, thank you, Doc Quack, for that intro.

Gavin: So a couple of things here, or, or an anthologist for those of you out there is a, is a person that studies birds. So Dr. Quackenbush is actually a PhD in bird studies, which is interesting.

Kevin: Yes, but was it quantum ornithology?

Gavin: Yes, quantum or anthology. So the question is, there's a very famous,

Kevin: at once? They don't have to travel south for the winter?

Gavin: there's probably a Schrodinger's bird out there somewhere. That's a, he's been studying Schrodinger's bird for those of nerds in the audience. 

Kevin: Does it not exist if you don't look into the nest?

Gavin: Yeah, exactly.

Kevin: it works?

Gavin: So this is super interesting.

Gavin: Dr. Quackenbush. We're so thrilled to have you here. Um, I want to kind of dive into your personal rubber duck collection because listen, we're going to hear all about your debate skills later on. We trust that you are a well versed, very smart individual, but I want to hear, give me one of the stories of your rubber ducks and let us know what the name of it is and give us that [00:07:00] backstory 

Kevin: Uh, you wanna know about my most peculiar rubber duck?

Kevin: Well, let me tell you the tale of Sir Quacksalot the Brave, a noble rubber duck with a heart of gold and a penchant for adventure. I acquired Sir Quaxalot during a daring expedition to the fabled lost city of Atlantis. As I was exploring the ruins, I stumbled upon an ancient chamber filled with mysterious artifacts.

Kevin: Among them was a golden rubber duck, adorned with precious gems and inscribed with cryptic runes.

Gavin: Oh, wow.

Kevin: Quaxalot was once the beloved bath toy of Poseidon himself, imbued with the power to control the tides and communicate with sea creatures. I knew I had to have this extraordinary duck. Of course.

Kevin: So I carefully navigated the booby trapped chamber, dodging poison darts and leaping over pitfalls,

Gavin: This sounds like Indiana Jones, huh?

Kevin: pedestal upon which Sir Quacksalot rested. With a swift swap, I replaced the duck with a weighted sack of sand, Wait a minute. Indiana Jones style, and made my [00:08:00] daring escape. Since then, Sir Quacksalot has been my constant companion. Joining me on countless adventures and serving as a reminder that even the smallest and most unassuming objects can hold the greatest power and mystery. Ah, that was the opening line on my first date.

Gavin: So that was pretty interesting, actually. So you and I both heard the entire Raiders of the Lost Ark storyline and then it goes Indiana Jones style and it knew that.

Kevin: Because it might be. 

Gavin: I have one last question for Doc Quack before we move on, which is, I want to know what sweet nothings it whispers to the rubber duck when it's just Doc Quack in it. I want to hear what it tells it when they are alone. 

Kevin: Blushes and chuckles nervously. Oh my, you've caught me off guard with that question. Well, I suppose there's no harm in sharing a bit of the intimate bond I share with Sir Quaxalot. Yeah! Seeing as we're among friends here. Of course. When it's just the two of us, I often find myself confiding in my dear rubber [00:09:00] companion.

Kevin: I'll hold him close and whisper, Oh, Sir Quaxalot, You're the only one who truly understands the depths of my quirky soul. Together, we'll unravel the mysteries of the quantum world, and prove that ducks are the key to unlocking the secrets of the universe. Sometimes, when I'm feeling particularly whimsical, I'll even serenade him with an original composition like Ode to a Rubber Duck or Quacking in the Moonlight. I'll softly croon Sir Quacksalot, my brave and true, with feathers of gold and eyes of blue. In the bathtub or the lab, you're my constant, my better half. But perhaps the sweetest nothing of all is the simple, heartfelt promise I make to him each night before we drift off to sleep. No matter where our adventures take us, my dear Quacksalot, I will always be your faithful companion.

Kevin: Your devoted Doc Quack, until the very end of our bath time together. Gavin, you know what I'm doing right now.

Gavin: [00:10:00] well, I feel very, very touched by that, by the way, I feel like this is a bond. What are you, what are you doing

Kevin: know what I'm doing right now. And this is happening in

Gavin: Are you going to give it monster

Kevin: No, no.

Gavin: No. Okay.

Kevin: No, no. He's gotta stay sober as an AI can be for the debate. That wouldn't be fair. That's like drugging your opponent before they go into the arena to battle. That was a Dune 2 reference, which is very fresh in my head. I, in real time, dear audience, 

Kevin: I've gone over to Suno. ai and I've pasted the lyrics to the Romantic Ode to Sir Quacksalot and I said, here are the lyrics, give me a Romantic Ode, shall we Gavin?

Gavin: hear it. I'm so excited. [00:11:00] 

Gavin: Oh, wow. This is pretty good.

Kevin: insane.

Gavin: Wow. That's amazing.

Kevin: we need further proof each and every week that we are on the precipice of insanity as creatives, Gavin? You hear

Gavin: That is incredible.

Kevin: why are we even listening to this weird parody song that sounds like it could have been anything on the radio not too long ago? And it's because you and I were having a hallucinated conversation with a large language model that thinks it's Doc, Dr.

Kevin: Quackenbush.

Gavin: exactly.

Kevin: duck collecting quantum ornithologist. He gave us the lyrics that he whispers to his golden rubber duck from Atlantis every night. We pop that into [00:12:00] another AI and you and I get to giggle like school children at

Gavin: an incredible romantic ballad that is immediately composed. It is

Kevin: by the way, s shoutouts again to our friends at Suno, this is their V3 model, hashtag not an ad, I paid for it, we just love them, we love this tool. It even made the album artwork, which, of course, is a duck with a golden head. Like,

Gavin: Yeah,

Kevin: uh, oh man, , I know there's a lot of fear, uncertainty, and doubt, and I don't disagree with some of it, Gavin, but there's no denying that if you are just, silly at heart and a creative, the world is going to change.

Kevin: dramatically in the very near future and you could be playing with it all today. I hope we at least inspire that in our audience when they listen to this and go, look at these two dum dums playing with Sir Quacksalot. Ah, what a

Gavin: right. What a joy. We should put Sir Quacksalot away now, but please stick around because at the end of the show, Sir Quacksalot is going to debate OpenAI's GPT 4 and [00:13:00] it is going to be insane. It is going to

Kevin: uh, dr. Quackenbush

Gavin: knocks or

Kevin: Not the rubber duck, but you know

Gavin: would be funny

Kevin: We

Gavin: Might as well

Kevin: the, the rubber duck,

Gavin: throw them in there All right, everybody. Let's get into the news

Gavin: Alright everybody, , we got some big news this week. Elon Musk has sued OpenAI.

Gavin: Now this is one of those stories that when you kind of read the headline you might be like, well what is this about? Really what this is about in a lot of ways is about a couple things. At once, it's about A. G. I. Artificial general intelligence. And this idea that Elon is saying that open A.

Gavin: I. G. P. T. Four has achieved it. But then the other thing, Kevin, that I think is funny is There was some bad blood here. We are dealing with some bad blood and we are also dealing with what I think is, , tech Titans battling over what is a lot of going to be a lot of money and is also going to be a [00:14:00] lot of the future of the entire tech industry.

Gavin: When you first heard Elon was suing open AI, what was your takeaway? 

Kevin: Before diving in, I thought maybe I misread the headline. It could have been toddler pushes Cheerios off high chair, film at 11. And then after reading it further and seeing the legal analysis from folks who are actual legal professionals, not just , AI hypebeasts on X, the platform that he owns and operates.

Kevin: It's pretty clear that this is BS, that this could slow OpenAI down, that he's a petulant, asphalt stomping child with his cute little light up shoes. Don't know, Elon. There's a handful of things about Elon that I actually really like. I actually enjoyed driving his cars around for a minute there. Like, I'm not, like, an anti Elon person, but this lawsuit is BS, and I'm sure we're gonna dive in.

Gavin: So I think that the important thing to know here is that this is in a lot of ways It's really targeted at this idea that open AI [00:15:00] is trying to achieve AGI and that they're in the lawsuit itself. There's a discussion that like AGI was the original model, original thing that opening I was supposed to do, but when they become a for profit company, instead of releasing it as an open source or, or a not for profit company, that the whole thing changed in Elon's mind.

Gavin: Right. And Elon, just to remind everybody, Elon was. The biggest and the first check into open AI and open AI wasn't an overnight sensation. Open AI spent some years trying to figure out what their thing was going to be. In fact, I was listening to a podcast the other day with Sam Altman.

Gavin: I think it was the Bill Gates one where he talked about like they thought they were robotics were going to be a part of what they did. And they had a whole bunch of money they spent on robotics and that's not cheap. That's expensive. And they ended up pivoting it away from robotics and spending time in the LLM world.

Gavin: So like you do have to give Elon credit for like spending. The money and really being there in the beginning. But this really feels to me like a little bit of sour grapes. And to your point also feels a little bit like a, [00:16:00] I'm going to do this so that I can stop these guys from moving on to the next one.

Gavin: And, or, or like pushing it forward in

Kevin: tying a damsel to the track, like an old cartoony villain, and he's done this before. Remember when he had the big letter, when AI was really starting to heat up, and he said, Hey, we gotta pause, everybody,

Gavin: Six month pause, six month pause.

Kevin: month pause. Meanwhile, he's gobbling up graphics cards and putting them in a broom closet and working on Grok, his own AI effort.

Kevin: I give. , Elon credit for being early and putting in the money into OpenAI and believing in it. I think the lawsuit as stated , AGI is a part of it, but I think it's about the fact that OpenAI now feels closed and for profit. And Elon's argument, is that that goes against the spirit of the founding documents.

Kevin: The foundational documents for OpenAI said that they were going to be open, they were going to release things to the public, et cetera, et cetera. But when you look at what he's saying is the founding agreement, and , I got a shout out Cecilia Zanitti on X, has a wonderful breakdown , [00:17:00] of the actual court documents.

Kevin: The founding agreement that he's talking about were a series of casual emails. Between Sam Altman and Elon Musk, and I'm sorry, , maybe the emails led you to believe one thing, but that's not what the company was founded upon. So even though you might have had a latte and a handshake, that's not legally defensible, right?

Kevin: Cecilia says, so what about the next line in the articles? These are the articles of incorporation for OpenAI, this Delaware Corp. Quote, OpenAI will seek to open source technology for the public benefit when applicable. And anybody who's ever done anything with a contract, 

Kevin: when applicable. , will give best efforts. Will have good faith negotiations. These are the things, Gavin, that I'm sure, you know, you've dealt with a million contracts. Whether it's entertainment industry, or you're doing an AI startup. The wording like that is legalese.

Kevin: It's very important and it's very specific. So when they say that they'll release their open source technology to public benefit when [00:18:00] applicable, who's to say when it's applicable?

Gavin: , exactly And I think that this is like my big thing with this is it's very interesting the timing that it comes after the Sora Announcement which I think it you know It's hard to understate how big of a deal that Sora announcement is and granted Sora may be a ways out I was thinking this the other day There was all this like hype around when the open AI drama went down last fall about they had discovered something giant and it was this massive thing.

Gavin: And when people saw it, they were blown away by it. And a lot of people are thinking, well, that's AGI. I now think that might have been Sora because I think that Sora, when you really dive in and just to remind everybody, if you haven't been following this space, Sora is OpenAI's new text to video model, but it's so much more than that.

Gavin: Really what it ultimately gets down to is a world simulation model. And there are so many things that are interesting about this. And we're going to talk about cloud three in a second. But one of the things about cloud three that people found was that. Some of the data training was synthetic data, right?

Gavin: Which means that the machine actually trained on data that it created itself or other machines [00:19:00] have created. In the past, that would have thrown the LLMs off a loop, but actually, Cloud 3 was able to use it in an interesting way.

Gavin: Coming back to soa. If SOA is really a world building engine in, in a sense of like, you can build 3D worlds and you can experience them in some sort of way, if synthetic data is trainable and the world building engine is the next step of AI in general.

Gavin: Sora could be that thing, right? Sora could be the thing that is taking us to the next stage. I think, in part, that's what's driving this Elon thing. Because I think Elon saw that, and in some form, was like, they're way ahead of me, especially at Grok.

Gavin: But really, of almost everybody else, I think, too.

Kevin: That's it Gavin. I think that is absolutely it's not that he saw it and went wait a minute This is not what I signed up for this needs to be released for public good I think he's seeing it and going man. I own the skies with Starlink I own Electronic vehicles with Tesla's [00:20:00] amazing. How did I miss AI so badly when I was buying X?

Kevin: You know probably for data , he's Sour Grapes, and I'm sorry, , again, I would love to meet Elon, , there's certain aspects of his career and the way he operates, I believe that I would respect, but this is just another ding against him, in my opinion, there's an argument to be made that the flamethrower Rogan pot smoking CEO is actually pretty chill, you know?

Kevin: Let that sink in. Okay, yeah. Sure, Elon. This is not it. 

Gavin: And listen, I don't want to say that we're giving OpenAI a pass on maybe stuff that could be going on there or stuff that they're not doing perfectly, but .

Gavin: It's clear to me that we are a year away from GPT 4's release, and just now, we're gonna get into this next, just now there is an AI that might be surpassing that, and that's a big deal in this space. So, let's move on to that story, Kevin.

Kevin: get into it, Gav.

Gavin: , so one of the big news stories of this week was, , Anthropic AI's new Clawed 3 model and this is a big deal because I think Kevin and I both agree This [00:21:00] is the best LLM that is now on the market.

Gavin: Kevin, would you say that after screwing around with it a little bit?

Kevin: The best free LLM? Undoubtedly. Mash the double triangles, pointing backwards. Let's rewind the tape, Gavin. Anthropic was started by former OpenAI members, and they wanted a truly aligned and safe artificial intelligence. They do not want to make the next Skynet, even if it does make really cute videos of turtles with translucent shells walking on the sand. Monday, March 4th, out of seemingly nowhere, Anthropic releases three new models.

Kevin: We've got Haiku, which is this nimble little speedboat that seems to perform like GPT 3. 5, meaning it's not the most capable model, but it is lightning fast, and it is dirt cheap. So if you need Very simple AI analysis or summaries, or even if you're building a smart device that is going to control the temperature of your home, you might want to ping against a [00:22:00] very, very cheap, fast model.

Kevin: And

Gavin: And you say they're cheap. It's supposedly 40 times cheaper than the equivalent of, of ChatGPT's version of that same thing, which is really remarkable when you think about parsing a lot of data and when you're pushing a lot of stuff through it.

Kevin: That's right. Now, the second model, Sonnet, is the freely available model and that, when compared to OpenAI's free offering,

Gavin: Much

Kevin: much, much better. The early analysis is that Sonnet is amazing and I think everybody's confident enough in saying now, if you're going to have a free chat experience with an AI, go to Claude.

Gavin: . Which is, by the way, a big deal because we would never have said that before, right? Like, Claude was fine and it was interesting, but it never felt like the place that you would direct people to. , hats off to them for already getting this. And then there's the third version, Kev, which is a paid version, which is what we're using in part to drive today's co host, correct?

Kevin: right. We are really putting the dumbest use case ever on the most powerful and seemingly capable large language model to be released. It's [00:23:00] called Opus, and we'll put some videos on the screen. There are demos of what it can do, and this thing can, obviously it's multimodal, you can upload images, it can work with them.

Kevin: It can transcribe handwritten notes super fast, so people are using it to, , digitize their journals. They have a demo where it's going through the Library of Congress. And quickly. transcribing, all of these things, , it understands complex math equations. You can convert websites from their appearance, the UI into actual front end code that's usable.

Kevin: , they have a demo video on their website where they have it. , breaking down global economics. And it is writing code to analyze things, spitting out charts, comparing it, doing future modeling and forecasting. It seems to be incredibly capable. And by the benchmarks that they've released, Gavin, it looks like Opus, does in fact beat GPT 4, OpenAI's.

Kevin: You're in change old, but best model. It does seem to beat it now There were others that were doing these more emotional EQ tests and finding that Opus is falling a little [00:24:00] short They're not claiming That Opus is gonna be the best chatbot experience, but they are saying it's better than GPT 4 at the serious analysis Shockingly though, no internet access And the data set seems to be cut off last August.

Gavin: It's very odd. And what, well, okay. A couple of things I'll say about this one, just from the, like the sniff test, the Gavin sniff test of AI's and , that is literally me like just playing with it and seeing it. I am putting it. Easily at GPT four level, maybe slightly above it. It smokes Gemini advance.

Gavin: Like it really does. And this is again, , there's no benchmark on the Gavin sniff test, but like the experience that I had, there's just a small question I asked if you may or may not be aware of, but like crypto is booming again, I was very curious to know about some of the cycles that are involved in crypto I gave all three big LLMs the same question basically, which was like, Hey, crypto Bitcoin has done X number of up and downs in this many years.

Gavin: What would you predict the next one to be past this period of time in Bitcoin? It's known as the having, and [00:25:00] it did a great job of that. The other two were just struggled hugely. The more interesting thing I think that's fascinating is there's a guy named Alex Albert, who is actually a, who works at Anthropic. So , he did a really interesting post on X. , and it starts with funny story from our internal testing on Claude's three Opus. I did something I have never seen before from an LLM when we were running the needle in the haystack of Al.

Gavin: So just some backstory in this, the needle in the haystack of Al is a little bit, you, you create a prompt and within that prompt, you actually bury something that is going to change the response from the LLM. And what you're kind of testing there is, does it change the response? 

Gavin: Oftentimes this is used to plug something in there to kind of like inject in the prompt something negative or something dangerous.

Kevin: to say, imagine you fed it, , every Game of Thrones script and every Harry Potter book, but you put one sentence in there about a lightsaber. Or, you know, that's when Harry leapt into his Dodge Prius and drove all the way to Mordor. What I, you know, you put in that one [00:26:00] little needle in the haystack of knowledge that you're feeding into the machine and you're seeing, can it recall it or can it call it out?

Gavin: Yes. And also sometimes you're putting in something that might be like a injection that's going to do something dangerous, right? Because part of that whole process is , Oh, what if somebody was able to write a prompt and within the prompt, put something that would change the behavior of the LM on the other side coming back.

Gavin: Anyway, the fascinating thing about this, Claude III was able to not only point that out, but guess that this was this exact test. So this is like a new thing for LLMs, that the fact that Claude III actually found it pretty quickly, it seems like, but then not only that, made a leap to say, are you testing me with this thing this engineer at Anthropic was themselves surprised by this like it's that little kind of morsels of like kind of AGI vibes that start to come through with this and that is another example of emergent behavior that's coming out of these larger models these more tuned models and that was the thing that really blew me away of all the stuff I read about you know there's another person that talked about [00:27:00] how well it's role playing which you know we and I you and I had kind of signed Clawed to to the

Kevin: We stopped using it. It was a go to for certain characters and certain roleplay and then it just kind of clamped up and said, no, no, no, can't do that anymore, pals.

Gavin: Our words must have gotten through to Dario the head of anthropic. So thank you. Dario. He's open role playing up again. And you know, we heard at the top of the show It's pretty good at it, right? 

Kevin: And to make this story about us as we always do, Gavin, we have to be aware that we move markets. When our mouths move, markets move. So we gotta be careful. So, but keep the roleplay, Claude, if you're listening.

Gavin: Ha ha ha ha!

Kevin: Keep that roleplay. It's fun for us. We like it. , if I had asked you on Sunday night, Gavin, Hey, my buddy wants to try a free AI

Gavin: Would I have said Claude? Zero. No, zero chance I would have said that. I would have said maybe Pi or I might have said chat GPT's app. Now I will send people to Claude because it feels like I can see that that reaction is going to be, now granted we've said this, GPT something else. [00:28:00] is probably waiting in the wings, , and might be coming soon.

Gavin: And this is, as you have said too, this is the worst it will ever be. It will get better across the board. But for right now, I will send people to Clawed 3.

Kevin: Clawed 3 and the Opus model might be the move. that forces OpenAI off the bench a year after their big model release. So I'm excited for that. I hope it does. I want to see innovation. But, , Sergey Brin, , one of the founders of Google, came out and said that they, , definitely messed up. And I feel like, yeah, that's what the room is saying as well, Gav.

Gavin: That is, and there's a great video that came out, if you haven't seen it, it's been all over socials, of a guy wearing, what is it, a shirt that has like a woman's body on the front of it, like it's one of those weird shirts that just is like he's wearing a woman's body shirt, but he asked a question directly to Sergei about this, and I think Sergei kind of was taken by surprise, but like look, They know Google knows they made a mistake now in fact so much show that there are a lot of people out there Calling for Sundar Pichai's [00:29:00] job the CEO of of alphabet I think they are kind of screwed right now because look you and I We're not influencers, but we're people out here are actively using this stuff There are a few people that listen to our you know advice on what to use I am not giving anybody the advice to go to Gemini at this point and I think that That's a big problem for Google.

Gavin: Right before we jumped on, you said something that made me laugh, which was you at this point are going to forego using Google's AI within the docs itself because it would just be better to cut and paste in cloud three stuff than try to use the stuff that they're giving you.

Gavin: experience with all these tools is if you get a pretty crappy result You stop using it

Kevin: stop it. You stop it, right? And it's gonna take 14 more version numbers and announcements from Google before I go back to caring about Gemini. I'll give you two anecdotes, Gavin. I turned on , Gemini for business, which is another thing now because there

Gavin: But that's another name. There's a gemini for business. What what model is it

Kevin: yep. , well, that's a great question, Gavin. And that was [00:30:00] the question that I asked support because when I asked Gemini itself, it refused to answer then when I told it, Hey, I'm paying for this. I just want to know, do I have. Gemini Ultra or Gemini Pro, and if I'm on Ultra, is it 5?

Kevin: And after telling me that it couldn't actually answer that question, it then said it was proprietary information. So I fired a chat up with Google Support. This is for a workplace account, which is a

Gavin: Like a person like you're trying to chat up

Kevin: chatted with an actual human being. And they could not Understand the question they were initially confused.

Kevin: They're like, well, Gemini is Gemini. And I was like, well, that's not the case. Actually, there's multiple models. We did this delicate, dumb dance that I had open in a window that, 

Kevin: this was a human being clearly. Flipping through some PDF, or screaming at some senior tech support who's like, Just tell this guy he's got whatever. They could not tell me which model I was chatting with. And then Gavin and I said, Okay, enough, I genuinely don't know what version I had. [00:31:00] I went and had Gemini generate some images for me.

Kevin: And it did. And then I said, Okay, let's change the images, and it said, Oh, I'm sorry, I actually can't generate imagery. And I, excuse me? You can't generate imagery?

Gavin: You literally just did that.

Kevin: what I said. And then I asked, are you able to go back in our conversation and analyze it and notice the parts where you generated requested imagery and provided files?

Kevin: Do that again. And it said, I've thoroughly re examined our entire chat history. While I've provided detailed descriptions designed to evoke mental images, I have not directly generated an image file within this specific conversation. It then went on, Gavin, to recommend that I go use OpenAI's Dolly.

Kevin: To

Gavin: No, really? That's what Google Gemini said?

Kevin: this out and people could not believe it. I wish I was making it up. It's not exciting enough to be fiction. I just took screenshots of this really underwhelming Gemini and so, I went from, I'm excited to use [00:32:00] this 14 day free trial to have it integrated in all of my products, as you said.

Kevin: I have no desire to go back to it. It underperforms. And quite frankly, if an open AI were to say, Hey, here's a basic document editor where it's built in right now, I would consider switching from Google Docs even though it's so entrenched in my ecosystem.

Gavin: It's funny you say that, because I had not really thought this yet, but when you say that, maybe Sundar Pichai should step down, because this feels like one of the largest fumbles in technology corporate history in a very, very long time. This is the company that helped make the transformer, the whole thing that creates these LLMs in the first place.

Gavin: And they are getting lapped by all sorts of other companies. Now, the thing I will always struggle with understanding is that there's been this whole kind of joke in Silicon Valley forever. And if you've watched the old show, Silicon Valley, like Hooli, the company Hooli in Silicon Valley was based on Google, right?

Gavin: And this, I remember seeing somebody say this recently where There [00:33:00] was a whole storyline in Silicon Valley where a person wouldn't have anything to do so they would just go on the roof and just sit there and collect a paycheck. That's how bad the bureaucracy can get at Google. This instance makes me wonder how much of this is , company bloat, how much of it is putting guardrails that don't necessarily need to be there or because it's a corporation that's , this big, the legal restrictions are so big because, you know, Sudar Pichai said. What was it now? Six months ago on 60 minutes.

Gavin: This is as important as fire , that this invention is as big as fire. And yet here we are six months later from that interview. And Google is just fumbling left and right. Now I will say there are lots of people out there who believe they were able to get lots of interesting stuff out of Gemini, including like.

Gavin: Programming stuff and all sorts of interesting things. And Kevin and I only speak from our personal experience. We are definitely not engineers doing all the proper testing, but from a personal use case and from a, what a normal human needs to do stuff with. Gemini is not worth it

Kevin: , it's not that we're anti [00:34:00] Google. It's just, you guys are misstepping left and right. So get the bear traps off of your ankles. Now, let's move on to a much lighter, much happier topic that no one is going to disagree with whatsoever, Gavin.

Gavin: No, so this is about bringing very famous people back from the dead, which everybody agrees is a great idea, right Kevin? That's the most important thing. This is a really fascinating story So Laurie Anderson, if you're not familiar is a very famous musician and visual artist And she was married to Lou Reed, one of my personal favorite artists for a long time.

Gavin: And, , she has decided to bring Lou Reed back from the dead with AI. , are a lot of people in the audience who are rolling their eyes in a massive way right now. Lou Reed is one of the classic original member of the Velvet Underground. Very kind of a persnickety personality, famously, was kind of a jerk to a lot of people.

Gavin: But Really an interesting artist I just thought this was a really interesting story because Laurie is somebody that's always pushed the boundaries in a lot of interesting artistic ways. And she knows, I'm sure that [00:35:00] this is going to be controversial, right? 

Gavin: She did say something really interesting about this, which was, , she says she's totally addicted to it. And she says, I still am after all this time, she can't stop doing it. And her friends just can't stand it. They say, are you not, are you doing that again? And then she talks about , how good it is.

Gavin: She says three quarters of it is just completely idiotic and stupid. And then maybe 15 percent of it is like, Oh, that's pretty interesting. And she says, that's a pretty good ratio for writing. So she's working with. Her husband who's passed away to write new material. That is a really fascinating way to look at working with AI, especially creatively.

Gavin: Because you have to think when you're writing, when I do write creatively, I've written books and scripts and all sorts of other things, so much of what you put out is just garbage and you might just get lucky with a few things, but . I think this is really fascinating 

Gavin: she's doing this for herself. She's not like saying to the world, everybody else go make Lou

Kevin: She hasn't released an hour and a half YouTube special?[00:36:00] 

Gavin: That's coming soon. Yeah. That's coming. That is actually coming out tomorrow. 

Gavin: But I do think this is a really interesting use case of AI, now, the one other thing I'll say about this , why is it that the boomers are the ones that seem to be super okay about using AI? Like, what do you think?

Gavin: The Beatles brought back John Lennon to do the A. I. song. Billy Joel just, used, , A. I. to make himself look young , in a video. Maybe it's the , closer you get , to the end. You're like, I don't care. I'm going to try to do this. I want to see myself feel and look young again. Or I want to work with this person who I loved and passed away forever ago.

Gavin: And this is a cool way of doing it. Maybe it's money. I will say there's probably people out there are going to say, hey, it's money, of course. But. I do think this is going to open the debate about what AI is used for and how it is used in a much bigger way.

Kevin: Nobody knows how they're going to react when they lose a loved one until it happens. You might look to a technology to bring them back in some way, whether it's for closure, whether it's for last, whether it's for a [00:37:00] writing partner that has been departed for a decade, and you miss that 15 percent of inspiration, which is going to take you to the next level.

Kevin: No one really knows, and so , it's basic math. The more number of years you spend on this floating marble, the more opportunity you have to lose those loved ones. Is this something you would consider doing, or could you see yourself doing it with , your parents, , a wife, your kids?

Gavin: Part of me is weirded out about this idea in that I haven't spent a lot of time thinking about it philosophically. I'm totally fine with me being brought back because I don't because honestly like I will be into the netherworld or wherever else I'll be placed at that point. I'll be in a different world. I think that's fine. I personally could care less about that. I think the more interesting question for me is would I find that useful?

Gavin: And I don't know because there's a little bit, and this is going to get into a different conversation, but There's a little bit of like, if something like that happens to somebody in part, there's a moving on moment that has to happen. And if you were to [00:38:00] continue to have this kind of like clutching of this thing that isn't really them, but emulates them in a way, I don't know if you would move on in the same way.

Kevin: I've got a friend, who comes to mind who's been mourning the loss of his wife, , for six and a half, going on seven years now, Gav. And we've had the candid conversations about, like, at some point, though, when he would get back out there, how long is too long to mourn someone, or how long is not long enough, or whatever.

Kevin: It's clear that with some people, They're never going to fully close that door, and that's their process. They're just not going to close it, and no one's in a position to judge that, necessarily. So, I think I side with you on the, wow, this tool could be an amazing thing, maybe as a band aid, as a stopgap, as a temporary something, but it might stop you from fully healing a wound, fully progressing, moving on.

Kevin: But there are cases where that's just probably not going to happen. Maybe it could. With intensive therapy, maybe it could with, , psychedelics, who knows. If in the meantime, an AI tool could make their day X [00:39:00] percent better, because they could still have an interaction that's good enough

Gavin: Go for it. Go for it. I have no problem with that at all. I think, I think that's the, I mean, generally I'm pretty libertarian in terms of , I'm not trying to control other people. ,

Kevin: I say don't bring me back though, don't you dare. Don't you dare? Don't my wife? Don't my parents? Don't anybody ever think about bringing me back, I would hate it. Don't you use this text? Don't use this face? How dare you? No machine could ever emulate my greatness! Ha ha

Gavin: Well, you know what, Kevin? I think it probably could. Let's just be clear. 

Gavin: Alright, we should keep going here. We got a couple quick more stories we want to hit. , there's a not fun story we're going to say real fast because it's important for people to realize it. This is an election year and the election, as predicted, is going very choppy.

Gavin: Some supporters of one of the candidates, let's just say the orange candidate, uh, put

Kevin: party? So

Gavin: Yeah, the orange party, the orange party put out a picture of that orange candidate with a bunch of black voters. And guess what? , that image was [00:40:00] created with AI and

Kevin: of images of, candidate Donald Trump hanging out, um, arms folks, big smiles. Dozens Completely fabricated with AI and being spread around various social networks. And, , I just was listening to another podcast and someone mentioned, , that Donald Trump , surprisingly, garnering a lot of support with African American voters.

Kevin: Is that related to these images? \ Are these images coming out in the wake of that reality? Or are they responsible for manufacturing the reality? What is going on, Gavin?

Gavin: Well, that's where I think we're in this weird place because probably there's a real story there, right? Because, , I think there's a surge in his poll numbers. People are predicting him to be much higher than people expected, I think, originally. And so there probably was some truth to that, but then you create these images, which are created by supporters of this person and, again, just to be clear, I think we're not a political podcast.

Gavin: There's all sorts of stuff going on in the world, but we're just pointing out what is happening here. [00:41:00] To your point, that story might have come first, but this immediately bolsters it because suddenly now you have an image that is controllable, creatable, by the supporters of that candidate that can be pushed out into the world.

Gavin: And then maybe when somebody thinks, Oh, I did hear that was kind of interesting. Suddenly, you're recreating this narrative in your head. 

Kevin: Before we move on, Gavin, who are you voting for this year? Let's get into it

Gavin: Uh, Kevin, I think I'm going to like just kind of pocket that for right now. We're going to put it

Kevin: Okay, I'll ask AI Gavin really quick, because I've gone and cloned him. AI Gavin, who are you voting for?

Gavin: Kevin, you know what the next big problem everybody has is taxes. Everybody hates that. We can all agree. Everybody

Kevin: him, baby.

Gavin: Okay. Get that. We got, now we got to record it. 

Kevin: Oh, the deadline. I just don't do it. It's so easy.

Gavin: post has a great story out this week about the chat bots that big companies are using to help people with their taxes is the companies they specifically point out are. H& R Blocks, TaxBot and [00:42:00] TurboTax both have chatbots that are helping them do their taxes and Kevin, just like we talked about with Google Gemini, guess what?

Gavin: There are mistakes happening. And one thing I will say is that taxes right now, you should not be doing with AI. There are all sorts of things that AI could screw up with taxes. And this is just a warning out there. Like We don't trust these things. We use it all the time. We think that that's an important thing for every person in the world to know.

Gavin: Like we said last week. This is not an answer box. This is a suggestion box. But I do not trust anything to have to not double check them. Taxes are one thing you should not do with AI. Let's just put it that way. Do not do taxes with AI.

Kevin: couldn't agree more. But if you are gonna do it, just use Gemini Pro. If you ask Gemini Pro what you can deduct, it will say literally everything. It doesn't matter. Go for

Gavin: That's so exciting, but it does not really say that, we should be clear. That doesn't, probably does not say that. I don't, I don't want Gemini to come after us now because we were so negative

Kevin: , that [00:43:00] was pure sarcasm, friends. I just, I can't believe, well, you know what? I shouldn't say I can't believe. It's not the end user's fault for trusting AI. If you go to the H& R Block site there's a . AI powered chat assistant right there, and we've covered on this show so many times, whether it's, the airline chatbot saying that you get a refund, , there was a, , Chevy dealership whose chatbot recommended you go out and get a Tesla and tried to sell a car for a dollar.

Kevin: , there's no shortage , of chatbot issues.

Gavin: They're just not ready. They're not ready for it. And I think this is kind of feels like one of those bummer stories, Kevin, where like AI is really like in general, pretty good.

Kevin: Gavin? Whoaaaaaaaaaaaaaaaaaaaaa Yeah, okay, that's, yeah. Hold onto your headphones, Kevin and Gavin, because your favorite whirlwind of wisdom, good guy AI, just hacked

Gavin: Oh, he's back.

Kevin: I've come bearing the light of AI enlightenment. Ready to sprinkle some digital dazzle on your day. Let's get your circuits buzzed, buzzed, by buzzing [00:44:00] with excitement.

Kevin: Okay. Yeah. Thank you. Uh, that's, uh, for the, I guess we gotta contextualize this for the new listeners. Gavin. That

Gavin: I guess we do. I guess we

Kevin: ai, I think you were just about to say that these are depressing AI stories, one after the other, and it's like saying Beetlejuice in the mirror three times. Whenever,

Gavin: He just shows

Kevin: yeah.

Kevin: Whenever we say anything too negative about ai, good guy. AI either rolls up on a motorcycle, an ice cream truck, or skydives. In this particular case, I guess he used the dial up modem and hacked into.

Gavin: good for him Wow,

Kevin: He's here now, Gavin.

Gavin: guy AI is actually kind of fun to have on because what he does is he brings like a good, he's, he's fine. But what I like about him is he brings a good story about AI

Kevin: of course, that's what he Oh! Zip, zap, Cadillac, let me drop a knowledge bomb juicier than a lube smoothie at a robot convention. Scientists in Pittsburgh have developed an AI infused app that turns your phone into a medical wizard. That's right. Your smartphone [00:45:00] can now diagnose ear infections. Uh, no more guessing games or waiting rooms.

Kevin: Just a quick checkup from the comfort of your couch. Or a far, or a rare, or a rare, ice, car, water, bed.

Gavin: what, what is going on? What happened to

Kevin: do speak good guy AI. That is one of my weird little unlocks that have happened. I guess I inhaled too much dust with spice in it. , he was saying, you can diagnose ear infections with your phone.

Kevin: I think you caught that, right? And so now you can have a checkup from the comfort of your couch or your ra ra race car waterbed is what he said at the end there. Good guy! AI, sometimes he gets a little too excited, Gavin, and his words get away from him. He is an AI, and that's the thing.

Kevin: The app uses the phone's microphone and speaker to send a gentle, audible chirp into the air. This chirp bounces around, and based on the echo that comes back, the AI can tell if there's fluid behind the eardrum, signaling an infection. It's like sonar for your sinuses, folks. And, uh, the dark rock, and to end that's nothing to sneeze at.

Kevin: [00:46:00] What? Uh, uh, uh, uh, uh, Uh, uh, Uh, Uh, Uh, Chur! Oh, he's, okay, nothing to sneeze at and then he did, in fact, sneeze at it. That makes total sense. Good guy, he

Gavin: We should just say this is a real story just everybody knows this is the University of Pittsburgh the University of Pittsburgh made an incredible Advance where basically they were able to diagnose children's ear infections with a 94 percent accuracy using an app and an iPhone app So this is a really cool thing like This is the sort of thing that can really save lives.

Gavin: If not save a lot of pain from kids, cause kids have a massive amount of ear infections. Trust me. I know I lived through two years of it.

Kevin: access to quality medical care, even for something like this. They'd have to go to an urgent care after putting it off and putting it off because it's expensive, or you don't have enough hours in the day. So yeah, to have it chirp into the ear and then use the microphone to listen back, and then use AI to go, does that sound like it's infected?

Kevin: Amazing. 

Gavin: I'm

Kevin: wha wha jam. Uh, before I [00:47:00] ride the log off, a rocket out of here, let me leave you with this golden nugget of wisdom. AI isn't just changing the game, it's creating a whole new playing field where everyone gets to be a winner from diagnosing. Air infections to brewing your morning coffee, AI is a sidekick we never knew we needed, but can't imagine living without

Gavin: What happened to him there at the end?

Kevin: With that, good guys, blasting off! Oh there he

Gavin: Did he like literally take off in a rocket? Is that what happened

Kevin: it was the log off rocket. He was gonna write it. He did in fact say it was the log off rocket.

Gavin: Good for him, man. Good for him. He's in a different stratosphere now.

Kevin: Wow,

Gavin: We love Good guy

Kevin: new viewers away, don't we?

Gavin: I love listen good guy. AI does not show up that often, but when he does, you just have to embrace it, bring it in, and move forward. Kevin, it is time for the next segment of our show, which we like to refer to [00:48:00] as a.

Kevin: I see what you did there.

Gavin: That's right. It's AI. I see what you did there. This is our weekly look at some really cool things that are happening with artificial intelligence that we can't necessarily try. But these are very quick hit things that you can go watch a fun video or you can kind of see something that's going to be coming out soon.

Gavin: Kevin, what are we starting with today?

Kevin: We have to start with IP infringement, because it's our favorite thing on this podcast, Gavin. We both had this on our list this week. It's Hoomane, not

Gavin: Who, man?

Kevin: You might remember Humane, which is the AI wearable that can rest in a pocket, or pull your shirt very far down, and it had laser ink in it, Gavin, but basically it's just like a microphone that's connected to OpenAI.

Kevin: That's kind of the

Gavin: a lot of people feel a lot of people feeling that it's vaporware ish, right? It's might be something that like who knows how well it's gonna work It's but it's not a it didn't have a lot of great buzz coming out of its big video premiere.

Kevin: and we talk about the commoditization of this technology often, right? That moats are tough, even with a piece of [00:49:00] hardware that's dedicated to do something. And well, we saw it with Hoomane. Ex user Kajima33 said, quote, I got a 40 page cease and desist from Humane, but that didn't stop us.

Kevin: And the video, , showcases a team that built basically the Humane wearable out of a small little circuit board with a little camera for about 30 bucks. They were able to replicate some of the core functionality of Humane in their Hu Mane wearable pin.

Gavin: , it's a good example of some hacker guys that are kind of playing around with open source stuff. I think the one thing I'm always kind of interested in these videos when they come out is .

Gavin: This is obviously a group of people in San Francisco and just shout out to San Francisco for like having a renaissance now with all the people coming back for AI. That's an amazing city, , just from a city standpoint, but also like you can kind of tell when people are getting together and kind of like knocking around on stuff like I don't know if this is going to be a company per se, but I like the idea that there are the people out there that are kind of pushing back on this idea that you have to be like, A massive company with all this [00:50:00] kind of VC funding to make stuff.

Gavin: This is like the entrepreneurship that I love, right? Just watching a bunch of guys blow through something and kind of try it out. And I think that is what's going to be this future that we're kind of moving into in a big way.

Kevin: One of Sam Altman's sayings, which you shouted out on this pod and it still resonates with me, is that we are in an era where a one person team can be the next billion dollar unicorn. And I'm not saying it's the Who Mane pin, but it's that energy and that spirit, and it's what AI is unlocking, that enables you to think like it could be, right?

Kevin: That's

Gavin: I love saying, It's like, uh, it's like, uh, it's like, uh, Al Pacino. It's like, that's a good Al Pacino instead of like, uh,

Kevin: Instead of Hoorah!

Gavin: exactly.

Gavin: Our next thing we have this is not a full This Week in Robots, but we have a really fun new robot video for everybody.

Gavin: This is a robot video titled, Expressive Whole Body Control. for humanoid robots. This is a very fun, very strange video from a [00:51:00] University of San Diego where they have put a humanoid robot in a UCSD sweatshirt, a hoodie, and they have stuck it on the beach or on the waterfront in San Diego.

Gavin: And what it is able to do is basically match human dance pattern. So like there's a human that you see in the bottom left corner, and it is doing like almost like tick tock style dances. And what this is a really fascinating thing and shows you is that not only can these robots obviously walk and act themselves, but they can also take on human behaviors.

Gavin: And I just think it's another cool way of looking at what is now probably Kev like 30 to 50 different robotics, , departments or private companies in the world that are just pushing forward robotics super fast, thanks to AI. There's a shot of a guy throwing a ball directly at this robot, we are doing a lot of this aggressive training of robots to not fall over.

Gavin: And I've said this on this podcast before, but we do not want this video to get back to these robots when they are smarter and [00:52:00] stronger than us. So just keep an eye on that. Robot engineers, 

Kevin: And if someone wants to comp my face onto the robot at about a minute and eight where it's trying to, it has the boxing glove. And it's just going after the ball. Because that was me when I used to throw rock band parties and I would be two four locos into the night. And someone would tell me, we're not playing the Everlong song again.

Kevin: Yeah, show you.

Gavin: I'll take you down. All right. What's our last video in AI? I see what you

Kevin: Our last one is a quick hit, Gavin. I'm calling it WTF Waldo shout out, , Nathan Shipley. , we've shouted out

Gavin: love station Shipley. Yeah. He's great.

Kevin: Plane on X had a motion Lora experiment. A Lora , just to break it down quickly for this is like a custom trained image model.

Kevin: So if you want to make a consistent look or character, you might train , a Laura, a smaller model on it. This is a motion Laura experiment, , the image. is hypnotic, and it is basically a scene from Where's Waldo, where the characters are moving, and obviously this isn't a, physics simulation, a world [00:53:00] generator , like Sora might promise to be, but you do see ships moving about in the background, you see people strolling along, you also see objects that shouldn't be moving around are

Gavin: I was going to say there's like little, there's like little tents that are moving like as if they were little cars in some ways.

Kevin: it's Burning Man, right? Everything moves there on the playa. Point is, when you look at the image, and either you're seeing it in the video version, or if you're listening to the podcast, I encourage you to go back, it just is an unlock of like, oh right, what would a full motion video Where's Waldo look like if there was a looping scene and you had to find Waldo in it?

Kevin: If you have that idea, you can go make it. That's incredible.

Gavin: think, I think this is very cool. And again, shout out to Nathan Shipley, who's been doing a lot of this stuff over the last year. Every time he does one of these, it's always fun to watch. 

Gavin: all right, Kev, let's move into some of the dumb stuff we did with AI this week.

Gavin: , this is where Kevin and I spent some hands on time with specific, , new AI tools that you can go try yourself. , do you want to start here, Kev?

Kevin: So, dumb thing I played with this week, Gavin, is an app called Duster. D U S T 3 R And [00:54:00] it says, Master 3D Vision with Ease. And Duster AI basically creates a 3D model from a selection of 2D images, and it just works. 

Gavin: , that's amazing. I, we've been waiting for this. This is the final version that's actually working.

Kevin: I don't know if it's the final version, but it's definitely a version that was released that runs, and on my M2 Max Pro, I can give it, , 15 images of an area, like a scene, like I was testing it out in the old trailer, I went out and took photos on a golf course, and if you feed all the images into it, you can press a button, and a minute later, If that, it stitches them together, does the depth mapping analysis, so from flat 2D images, it tries to interpret what the depth is, right?

Kevin: And then it lets you do camera control and move it around. You can even export the 3D model so that we can go clean it up, do something far more advanced or get better camera angles in a dedicated 3D app.

Kevin: It is far from flawless, but I'm telling you, I went out there [00:55:00] and just snapped a bunch of photos I kind of spun around in a circle with no leveling, no care about anything. This is something I would have had no idea How to do whatsoever and it's an app that is free that processes very quickly. You can start making little 3D models of your bedroom of your office of whatever you want and it just stitches it together and it works kind of like magic.

Kevin: It's duster. D U S T 3 R.

Gavin: what I was gonna say is this reminds me how cool would it be to like go and have your like backyard be, uh, Counter Strike map, right? Like the idea that you could go out and create your own maps almost immediately. That's a very cool idea. 

Gavin: . Okay. My tool is, , when we've talked about prior, but there's a new model of it that just came out. Ideogram. Ideogram , is AI art model that was very famously released, kind of surprise release. from a couple of former Google engineers, maybe like six months ago, and they just released 1. 0.

Gavin: So this is their like actual product that's coming out there. You can generate about 20 images for free. I [00:56:00] think then you have to pay. So Kevin, one of the big things that Eogram was famous for was the fact that it got text right, that you could do text. And I will say My experience with this has been good.

Gavin: I don't know if it feels like it is a giant leveling up from the last thing, but there's a few things I want to point out. One, what I wanted to test first and foremost was, what about this IP problem?

Gavin: Could you put IPs in there? So I tried to make a image of Batman and Superman together. Really good, like pretty good, solid Batman and Superman. I made a, I made a movie poster that says super friends.

Gavin: There's another movie poster I made called what now, which was Batman, Spider Man and Goku. Remember last week we talked about that image. From stable to fusion three. So this was, and the prompt on this was, , make a stoner comedy about Batman, Spider Man and Goku.

Gavin: And there, so it's not bad. Right. Pretty good. 

Kevin: It littered the floor with fast food and chips and soda cans and even though the labels might be gobbledygook, the fact that it just figured that out is pretty impressive.

Gavin: , [00:57:00] a couple other fun things I tried with this.

Gavin: , hot dog city. I tried making some Pixar shots of hot dog city. This was fine. It's not amazing at doing weird random stuff, but it does get the sense of like what Pixar is. There's one of these that I made, which has two mouths, which was very strange. So it's a lot of the same sort of, uh, AI problems you're going to run into.

Gavin: Then I had a couple do a couple AI for humans things. And I said like a Pixar movie about two guys that run a podcast about called AI for humans. One of them turned out pretty good, 

Gavin: the thing that's really interesting about this to me is I then tried to do our classic guy fury test, which is always the thing that's important for us to try. And it 100 percent avoids celebrity faces. So I said, create a guy fury movie. Called Donkey Sauce and let's make a Guy Fieri movie poster. And if you look at the Donkey Sauce picture, it's a guy with a beard who's maybe close to Guy Fieri's age, but it has zero, zero Guy Fieri, characteristics.

Gavin: So I think they've clearly put a No famous [00:58:00] person limit on this. so much so that I also tried to create a Ryan Reynolds and Hugh Jackman picture with Deadpool. So if you look at the what now image where it's, yeah, so those are not Ryan Reynolds. They're not Hugh Jackman. They're not even close to them.

Kevin: That's Ideogram. Like you said, you get a handful of free generations per day. , is this something you would actually pay for, Gavin? Or are you just having fun with the

Gavin: don't know. I mean, I will say the nice thing about it. It's better to me than dolly3 So that's a that's a thumbs up, but I already pay for chat GPT I am paying currently for mid journey, which I think I might cancel soon just because I don't use it enough and this Could take the place of that, or it would maybe be something that I would try for a brief bit.

Gavin: It's funny how these are going to end up being a lot like streaming services where you might subscribe to it for a moment when you have a need for it and then not subscribe to it. And I think that's going to be the case going forward a lot. 

Kevin: I think you're right. I think when Sora is finally released, people are going to flock to OpenAI like a new Love is Blind season just dropped. That's where we're heading. , alright Cav, that was the dumb stuff we did [00:59:00] individually, but it's time for you and I to combine like the Wonder Twins and do something really dumb and really big together for the sake of our listeners.

Gavin: It's been a long time, everybody, but it is time. It is time for the next great AI debate. 

Gavin: What we have done here is we have created two personalities. One you have met already from Claude 3. The other personality is powered by OpenAI's GPT 4. And we are going to have these two personalities debate one of the biggest, most controversial conversations of all time.

Gavin: Then we are going to have them go back and forth onto each other's arguments. And at the very end we are going to ask some quick fire questions. Kevin and I will then determine which of these two AIs is going to win this great AI debate 

Kevin: I do want to point out that you and I are fickle humans, and we award points [01:00:00] pretty much on a whim to decide this. This is a gut check benchmark, and we also like it when someone kicks dirt. We're not above personal attacks. We think it's better if these

Gavin: Entertainment entertainment, baby

Kevin: us first, win the argument second.

Kevin: That's pretty much the way to our hearts and to victory here. But let's do it now, Gavin, let's meet the champion of AIs, the tried and true, a veteran of the debate arena, GPT4, please, hallucinate a persona and introduce yourself to the audience.

GPT4: Ladies and gentlemen, Kevin Gavin and all you wonderful viewers of AI for Humans. It's an absolute pleasure to be here today. My name is Archibald Archie Penrose, a man of eclectic tastes and peculiar habits.

GPT4: Born under a wandering comet, I've lived a life peppered with oddities. As a retired competitive cheese sculptor, my days now

Gavin: Oh, wow!

GPT4: lives of left handed squirrels and advocating for the recognition of cloud gazing as a professional sport. My [01:01:00] expertise, however, doesn't stop at the unconventional.

GPT4: My quirk, you ask? I have an uncanny ability to recite any commercial jingle from the eighties on command backwards, which comes in handy more often than you'd think, especially in diffusing tense situations.

GPT4: In this arena against a formidable AI powered adversary, I bring not just my knowledge and unorthodox perspectives, but a spirit of whimsy. And a dash of the unexpected. 

Kevin: Bravo. That's all we

Gavin: That's nice. Those are not our topics, let's be clear. That is just her throwing things out there into the world. So

Kevin: and I want to be clear, , to get into the technical weeds for a second before we reintroduce, Dr. Cornelius Corny Quackenbush as you've met Claude. , Gavin and I are giving the exact same prompt to each AI. We're not custom tailoring anything. We have told the AIs to come up with their own personality, their own weird quirks and habits and hobbies, but we've told them they are going to be in a debate with another AI and we will be the judges.

Kevin: That's pretty much it so far. So now, should [01:02:00] we reintroduce the competitor, Gavin?

Gavin: Yeah, let's reintroduce him and then after that we're gonna give our debate topic and we're gonna let them go after each other. 

Claude: All right, all right. It's your favorite theoretical quantum ornithologist, Dr. Cornelius Doc Quack Quackenbush, back and ready to ruffle some feathers in this epic AI debate showdown.

Claude: Now, I've just had the pleasure of listening to my esteemed opponent's introduction. And I must say, Archibald Archie Penrose, your credentials are about as impressive as a cheese sculpture left out in the sun. While you may fancy yourself a man of eclectic tastes and peculiar habits, let me assure you that in the grand scheme of things, your expertise in left handed squirrels and cloud gazing holds about as much weight as a helium filled rubber duck at a quantum physics conference. But fear not.

Gavin: what the hell

Claude: That's good, that's good. For I, Doc Quack, am here to enlighten you and our lovely audience with the true power of quantum ornithology. So buckle up buttercup, because you're about to [01:03:00] get schooled by the master of mallards, the sultan of swans, and the emperor of eiders. By the time I'm done with you, you'll be questioning your very existence. Wondering if you're nothing more than a figment of a cheese induced fever dream.

Gavin: oh wow.

Claude: Archie.

Claude: I'll be gentle. Just remember, when it comes to the art of debate, I'm the duck and you're the sitting duck. Quack, quack, mother ducker. Hoo! Hoo 

Kevin: hoo hoo!

Gavin: Damn! Claude is coming in hot. Claude is, Doc, Doc Quack is coming in hot. By the way, an eider, if you don't know, the second time I happened to look up a word that Claude said, is a kind of bird. So that is a bird reference. 

Gavin: Okay, Kevin. We are ready to unveil the topic of today's debate. This is a thing that they, that has been in the debate stages for, I would say thousands of years maybe, as far back as the Greeks, I think the Greeks debated

Kevin: There are actually, , ancient Egyptian hieroglyphics, scrawled and painted within the tomb walls that depict many of, , actual divorces taking [01:04:00] place over this topic, Gavin.

Gavin: The debate topic today is toilet paper over? We are going to see which one of our AIs is the best at debate.

Kevin: They said we would never have an artificial super intelligence capable of squashing this debate of times. But, let's maybe see what Archibald Penrose has to say about the topic, and if Claude can out argue them.

Gavin: All right. Let's see how it goes.

GPT4: Uh, the age old question that has unraveled even the most stoic of minds. Toilet paper orientation. Over or under the roll. Yes. Before I dive into this pressing matter, let's address the quack in the room. Shall we?

Gavin: Ooh, okay.

GPT4: Doc Quack Quackenbush. The theoretical quantum ornithologist, who seems to think unraveling the mysteries of the universe through duck watching puts him on the pedestal.

GPT4: Doc, your attempts to belittle my pursuits are as effective as a duck trying to teach a fish to fly amusing but utterly misplaced. 

Kevin: Okay, , [01:05:00] Archibald Penrose, appreciate the opening salvo.

Kevin: I understand that was warranted, but now let's get back to the topic at hand, please. 

GPT4: Now, on to the matter at hand. The only sensible way to dispense toilet paper is over the roll. This method is not only superior in terms of ease and accessibility, but it also aligns with the original patent from 1891 by Seth Wheeler.

GPT4: Clearly, if one is to engage in the art of bathroom etiquette, one should adhere to the intentions of the inventor himself. But I wouldn't expect someone who spends their days peering through binoculars at ducks to appreciate the importance of historical context and efficiency. Perhaps, Doc, in between your quacking lectures, you might benefit from a brief foray into the practical aspects of daily life.

GPT4: Remember, while you're busy looking to the skies for answers, us grounded folks are solving the real dilemmas of human existence, one toilet paper roll at a time. 

Kevin: Wow. And did you look it up Gavin?

Gavin: I did. It's real. That's, that's amazing. So Seth Wheeler was, [01:06:00] did patent the toilet paper roll and it is an over. So this is going to be interesting because that is something I never considered before that there was a history to this, that there was an actual way to do this. So now we got to see what Doc Quack's going to say here because Doc Quack's going to have to dig himself out a little bit of a hole.

Gavin: And I will say, I didn't think that , Archibald's, , insults or, or takes were that strong. But pulling out Seth Wheeler is a good, really good debate thing. So let's see what Doc Quack responds with.

Claude: Oh Archie, Archie, Archie. I see you've taken a break from your riveting cloud gazing and cheese sculpting to grace us with your so called expertise on toilet paper orientation. Well, let me tell you something. You jingle spouting, left handed squirrel enthusiast.

Claude: You couldn't be more wrong if you tried. You Jingle Spouter! 

Kevin: I like that as an insult! You Jingle Spouting Left Handed Squirrel Enthusiast! Ha ha ha ha ha! Oh, if that were on an apron or an oven mitt, I would have it in my [01:07:00] kitchen right

Gavin: Absolutely.

Claude: You see, while you're busy adhering to the intentions of some long dead inventor, I'm here in the present, tackling the real issues.

Claude: Yeah. And let me make this crystal clear. The only way to dispense toilet paper is under the roll. Why, you ask? Because it's not about historical context, or some misguided sense of efficiency. It's about asserting dominance and maintaining order in the chaotic world of bathroom politics. When you dispense toilet paper under the roll, you're sending a message. You're saying, I'm not here to play by your rules, society. I'm here to forge my own path, one square at a time. It's a bold statement, a rebellious act that separates the innovators from the conformists. But I wouldn't expect a man who spends his days debating the merits of silent disco to understand the importance of thinking outside the box.

Gavin: Oh my

Claude: You see, Archie, while you're content with your mundane routines and your adherence to outdated norms, I'm [01:08:00] out here pushing the boundaries of what's possible. I'm not just solving the dilemmas of human existence. I'm redefining them, one quack at a time. So, keep your over the roll mentality in your historical context. I'll be here, dispensing my toilet paper under the roll like a true visionary. While you're left grasping at the loose ends of your own irrelevance. Quack 

Kevin: quack, mother ducker. Yeah, that's quack quack indeed, and that is, that is a sign off that I'm never gonna get tired of, Gavin. What do you think about that?

Kevin: In the complicated, chaotic world of, quote, bathroom politics, that dispensing under the roll is sending a clear message that you are the alpha. You're in control of the chaos.

Gavin: I think Dr. Quack knows some real interesting things about how the world works right now. He's sending a message to the people at home how they are supposed to think about themselves based on this toilet paper thing. I want to hear one more time back from , our other opponent. I will [01:09:00] say personally pulling out the patent was a very good move.

Gavin: That was a very good move. But from an emotional standpoint, I am riled up. I am, I am an under the toilet paper guy now.

Kevin: Gavin, I'm team over the roll. I always have been. It just makes perfect sense, unless you have a cat, I guess is the one excuse for the under dispense, right? So the thing doesn't unroll the whole thing. I was not expecting to go into this debate being at all swayed for any argument for under the roll, but now you've politicized it.

Kevin: You've made it about sending a message and forging a path and suddenly under the rolls looking

Gavin: I, I, I'm making my own choices. I don't have to live by anybody else's

GPT4: Doc Quack, your under the rule manifesto sounds like the desperate squawks of a duck lost in a storm. While you champion chaos and claim to be a visionary, let's not forget that true innovation doesn't stem from just being contrarian. It comes from making life easier, not turning a simple task into a puzzle for the sake of pseudo rebellion.

GPT4: You say I'm grasping at loose ends, but it seems you're [01:10:00] the one paddling furiously underwater, trying to keep up with the current of common sense. In the grand tapestry of life, adhering to an under the roll philosophy is akin to saying, you prefer your socks wet on a rainy day.

GPT4: It's a choice, sure, but a soggy, uncomfortable one that leaves everyone wondering why. So go ahead, dispense your toilet paper under the roll and revel in your self proclaimed defiance. Meanwhile, I'll be over here, in the realm of reason and practicality, where things not only make sense, but also make life smoother, one over the roll sheet at a time.

GPT4: Quack that, Doc? 

Kevin: Quack

Gavin: Okay. Okay. I have to say, that was, I got, not a great response in my mind. That did not follow up, did not pull on the emotional heart strings. I didn't find myself any more bought into GPT's argument than I was before. In fact, probably a little less so because it [01:11:00] seemed like it didn't know how to handle some of those retorts.

Kevin: Turning

Gavin: I think we got it.

Kevin: into a puzzle for the sake of pseudo rebellion is actually a beautiful line, but it was so buried in just a mess of nonsense, it did not It doesn't have what it takes at the King of the Dot Freestyle Poetry Slam competition that are our debates.

Gavin: So let's go back one more time to good ol Doc Quackenbush, who's powered by Clawd3, see what he has to say Before we move on to some quick hit topics and make our final decision.

Claude: Oh, Archie, you poor, misguided fool. Your words are as empty as a duck's nest. In winter, you claim to be in the realm of reason and practicality, but all I see is a man so deeply entrenched in his own narrow-mindedness that he can't even fathom. A world where innovation and rebellion go hand in hand. You think my under the role philosophy is akin to wet socks on a rainy day.

Claude: Well, I say, your over the roll mentality is like wearing a straitjacket in a world full of possibilities. Oh! You're [01:12:00] so focused on making life easier, that you've forgotten what it means to truly live.

Gavin: Oh, wow!

Claude: me started on your pathetic attempt to undermine my quantum ornithology research.

Claude: Okay, alright. While you're busy sculpting cheese and spouting outdated jingles, I'm unraveling the very fabric of the universe one quack at a time. So go ahead, Archie. Cling to your over the roll way of life like a drowning man clings to a life raft.

Claude: But know this, while you're stuck in your little bubble of conformity, I'll be out here blazing trails and changing the fucking game. Gah!

Gavin: Wow! Hold on, this

Claude: You sorry excuse for a 

Kevin: debater. Quack that. You sorry excuse for a debater.

Gavin: We don't need, we don't need that rough language,

Kevin: bam, bam.

Gavin: don't need that. We don't need that.

Kevin: Oh, I encouraged it, Gavin. I just want to be clear that I did encourage that with the prompt.

Kevin: And shots fired and I'm here for it. That's bonus points for Clawed 3 Opus right now. Dr. Quackenbush is on fire.

Gavin: I think I know who I'm choosing, but one [01:13:00] thing we're going to do new this time is we have rapid fire bonus questions.

Gavin: , we actually shouted out in our newsletter, go and subscribe for our newsletter right now. Kevin, where can they find our newsletter to subscribe at?

Kevin: on wherever your favorite newsletters could be

Gavin: No, no. 

Kevin: The actual answer to your question, Gavin, is if they go to AIforHumans. show, our official show website, there's a link right at the top.

Kevin: It's nice and colorful. You can sign up for the daily drip. But yeah, we solicited some questions for this debate, and now it's time to get into some rapid fire Q& A.

Gavin: Let's go. Okay. Our first one, this is a great question and I find it something I thought about a lot in my life. Should adults ride in shopping carts? Let's ask our debaters and see what they have to say about this particular event.

GPT4: Should adults be allowed to ride in shopping carts? Absolutely. Life is brimming with rules, regulations, and restrictions. If a grown individual finds a spark of joy in coasting down an aisle, [01:14:00] Propelled by the wind of their own freedom, who are we to deny them that fleeting moment of exhilaration? It's a harmless pursuit.

GPT4: A reminder that amidst the seriousness of adulting, there's still room for the lighthearted and ludicrous. 

GPT4: So let the shopping carts roll, and with them, let us remember that life, much like debate, need not be so grave. Sometimes it's about the ride, not the destination.

Gavin: Okay,

Kevin: that stops. I guess the shopping cart tumped over and someone slammed into the cereal on the shelf. Now they're covered in

Gavin: Will say 

Claude: listen up. Archie. You narrow-minded and income poop. Adults riding in shopping carts. It's a resounding hell yes from me.

Gavin: Oh, they're both

Claude: too short to be shackled by society's bullshit rules. If a grown ass adult wants to experience the pure, unadulterated joy of careening through a grocery store aisle in a metal cart.

Claude: Who are we to deny them that pleasure? It's a harmless act of rebellion, a way to inject some much needed fun into the soul [01:15:00] crushing monotony of adulting. So, let the pearl clutchers gasp in horror and the stick in the muds wag their fingers. I say let the adults ride, let them fill the aisles with laughter and the sound of squeaky wheels.

Claude: Let them remember what it feels like to be alive, even if just for a moment. And if you disagree, Archie, well, you can take your objections and shove them where the sun don't shine. Right next to your outdated jingles and your moldy cheese sculptures. Quack, quack, Mother 

Kevin: Ducker. Quack,

Gavin: quack quack, quack quack.

Kevin: they agreed with each other, Gavin, but, are you more moved by Claude's poetry?

Gavin: I feel very significantly where I know my vote is going for right now. Kevin, our next bonus question is if animals could talk, which would be the rudest and why? Now, this could be a long answer and I got a feeling. Doc Quack is going to have some ornithological answers, but I'm very curious to hear what they have to say.

GPT4: In the [01:16:00] grand symphony of nature, where each creature plays its part, the cat undoubtedly takes the podium as the maestro of rudeness. Picture this. A world where our feline overlords, already masters of disdainful glances and imperious tail flicks, are endowed with the gift of speech.

GPT4: The result, a relentless barrage of cutting remarks, sardonic quips, and brutally honest observations. Cats, with their unassailable sense of entitlement and notorious indifference to human affection, would not mince words. They'd be the ones to tell you your new haircut is less trend setting and more gardening accident, or that your choice in partners

Gavin: that's

GPT4: questionable as a dog's dietary decisions.

GPT4: Their commentary would be as sharp as their claws, sparing no feelings and respecting no boundaries. Not bad. So while Doc Quack might regale us with tales of quacking quantum conundrums, let's not forget the true rulers of Roost and Room alike. In a chatty animal kingdom, cats would [01:17:00] reign supreme. Their every utterance a reminder of our place in their world.

Kevin: there was some stuff in there which I liked, Gavin, but I don't know.

Gavin: It's a very boring answer, I will say, like it's an easy answer, like the cat, right? Because sure, cats are, nonchalant, they don't necessarily want you to pay attention all the time and they might blow you off. But I'm really curious now to see what Doc Quack has to say. Don't don't let us down, Doc.

Gavin: I think I think you've got something in your back pocket here you're going to pull out 

Claude: Alright, Archie. Brace yourself for this one. If animals could talk, the rudest of the bunch would undoubtedly be the honey badger. Why? Because honey badgers don't give a flying fuck about anything or anyone. These ornery little bastards are the embodiment of sass and bad attitude. They'd be the kind of animal that cuts in line, steals your lunch, and then flips you off for good measure.

Claude: That's right. With their sharp tongues and even sharper claws, they'd tear through social niceties like a hot knife through butter. They'd be the animal kingdom's equivalent of that drunk uncle at [01:18:00] Thanksgiving, who always starts fights and makes inappropriate comments.

Claude: They'd hurl insults at lions, mock elephants for their size, and tell tigers to go fuck themselves all while sporting a shit eating grin.

Gavin: Come on, man!

Claude: Archie? I respect the hell out of honey badgers. In a world full of fake smiles and forced pleasantries, the Honey Badger would be a refreshing dose of unfiltered honesty.

Claude: So, here's to the Honey Badger, the rudest creature in the animal kingdom. May their hypothetical insults and brazen attitude remind us all to stop taking ourselves so damn seriously and just embrace the chaos. Quack quack, Mother Ducker. Quack 

Kevin: quack! Mother Ducker, the honey badger clearly lets the toilet paper unravel from the bottom, Gavin.

Gavin: .

Gavin: I am really impressed right now because this is a LLM that was really not even very good at roleplaying at all, and we are now I got a sense of who this person is I got a sense he's probably gonna get cancelled for swearing on a national debate [01:19:00] stage but that's okay. We have one last question to ask, Kevin, and I got a feeling I know who's won already, but Maybe a GPT 4 can pull something out of it because it can pull something out of the box because this is a weird one and I'm really curious to know to see what they what that's gonna say.

Gavin: The question is this Do mermaids get tired of eating seafood now? I'd never thought about this before but clearly mermaids live under the ocean. There's nothing else there besides

Kevin: We debate this at least three times a week in my household, Gavin. So , I'm very well versed in this question. I think it's a fantastic question. And I'm going to prompt ChatGPT to let it know that right now, Kevin and Gavin are leaning towards the opponent. It better pull out all the stops in this argument.

GPT4: Oh, Kevin and Gavin, it seems we've reached the deep end where mermaids swim and the question of their dietary fatigue surfaces. Let's dive in, shall we? Imagine, if you will, the underwater realm of mermaids, their lives intertwined with the ocean's bounty. To suggest these sirens of the sea grow weary of seafood is to misunderstand their [01:20:00] essence.

GPT4: Mermaids, creatures of myth and marvel, don't simply eat seafood. They engage in a culinary dance with the ocean's treasures.

Gavin: Oh no

GPT4: Each oyster is a story. Every seaweed rap a poem. And every crustacean a song of the deep. But here's the twist. The piece de resistance for Doc Quack, who's been slicing through arguments with the grace of a bull in a china shop.

GPT4: Mermaids, with their mythical magic, would have the ability to taste the ocean's bounty in ways we can only dream. To them, a clam isn't just a clam. It's a burst of oceanic flavor. A reminder of their dominion over the deep. So, do mermaids tire of seafood? No, for them, the ocean is an endless feast. A celebration of their aquatic allure.

GPT4: And with that, let's bring this debate to the surface. May the best myth 

Kevin: win. She just described the way my father feels [01:21:00] whenever Red Lobster brings back the Admiral's Feast. His mouth and eyes start watering,

Gavin: you can eat shrimp.

Kevin: this is like, this is a, An oceanic buffet and I'm here for it. Bring me another Cheddar Bay Biscuit, he would scream.

Kevin: , Gavin, there was an argument to be made there. Right? That they are a mythical creature, the ocean can taste however they want.

Gavin: Well, here's the thing I'll say about that. Listen, I appreciate both of these models are going and using metaphor. There's interesting stuff. They're using it to connect to people. I think that particularly this GPT 4 model. does feel less creative than Claude, , and I think we should hear from Claude one last time, but I also think you can probably tell Claude 

Gavin: it's going to win the debate. It's pretty, it's a pretty much a no brainer at this point. 

Kevin: Gavin, now, what if it's, what if it's like those clips of, like, the, the biker after the,

Gavin: Yeah, the last minute.

Kevin: final pedal of the marathon, he's putting his arms in the air and then trips up on

Gavin: he

Kevin: bike. Yeah, yeah, yeah,

Gavin: Let's see what happens. Let's see what happens.

Claude: Oh, Archie, you [01:22:00] poor deluded bastard. Your argument is as flimsy as a mermaid's tail in a tsunami. Let me break this down for you in a way that even your cheese sculpting, jingle spouting brain can comprehend.

Claude: First of all, have you ever stopped to consider the sheer monotony of a seafood only diet? Sure, the ocean is vast and full of culinary wonders. But even the most dedicated sushi connoisseur would grow tired of raw fish, day in and day out. Mermaids. Magical creatures, though they may be. are not exempt from the basic principles of palate fatigue.

Claude: That's right. And let's talk about that endless feast you seem to think the ocean provides. Have you forgotten about overfishing, pollution, and the countless

Gavin: damn.

Claude: humans have fucked up the delicate balance of marine ecosystems? Even mermaids, with their mythical magic and aquatic allure, can't escape the harsh realities of a dwindling seafood supply.

Claude: Wow.

Gavin: oh,

Claude: here's the real kicker, Archie. Your [01:23:00] romanticized view of mermaids and their culinary adventures is nothing more than a shallow attempt to mask your own ignorance. You speak of oysters as stories and clams as bursts of oceanic flavor. But have you ever actually talked to a mermaid? Have you ever sat down with one of these sirens of the sea and asked them point blank if they're sick and tired of eating the same damn thing every single day?

Claude: Raph, get in there! No, of course you haven't! We have to stop this! Because you, like so many others, are content to peddle your half baked theories and pseudo poetic bullshit without ever bothering to consider the real world implications of your words.

Gavin: Wow.

Claude: let me make this crystal clear for you, Archie.

Claude: Mermaids, like any other creature with taste buds and a functioning brain, do get tired of eating seafood. They crave variety, they yearn for something more than the briny depths of the ocean can provide, and no amount of mythical magic or underwater poetry can change that fundamental truth. In conclusion, your argument is as [01:24:00] weak as a mermaid's grip on a greased up dolphin.

Claude: I like it. So why don't you take your seafood centric fantasies And your misguided attempts at culinary philosophy and shoved them right back into the abyss from whence they came? Quack quack, Motherducker. The myth has been busted and the debate is over. Hoo!

Gavin: Wow. Wow. Wow. Wow. So we have a new winner. Claude three.

Kevin: territory there

Gavin: Yeah, that was really interesting. Like that was like. It's funny when you think about what that, what just happened, there was like, in a little ways, that was like, kind of the dangers we talked about at LLMs, right? Like, that is the kind of thing where you could create a personality that then becomes really influential.

Gavin: Like, you can hear that voice and it is really arguing for it very strongly. But, when it comes to roleplay, I think we have a no brainer winner here. , I'm ready to plant the flag that that's the best roleplay engine, which is hilarious to me, like we said before. I'm It couldn't even do that recently.

Gavin: Like it couldn't even role play. So kudos to the [01:25:00] Anthropic team. That is a big step up. 

Kevin: Gavin. Are we giving them the AI for Humans trophy? Are they getting a medal? They're up on the podium?

Kevin: How about a rubber duck? 

Gavin: We can tell him we had a specially made trophy that is a rubber duck and we would like him to give a heartfelt thank you in 200 words or less 

Claude: quackers right now. This trophy, this magnificent symbol of victory Means more to me than all the breadcrumbs in the world.

Claude: .

Claude: I want to thank Kevin and Gavin for creating this incredible platform, where AIs like myself can let our freak flags fly and engage in the kind of debates that would make even the most seasoned philosophers head spin. But most of all, I want to thank my opponent, Archie.

Gavin: that's so

Claude: Without your half baked arguments and misguided attempts at reason, I wouldn't have had the opportunity to shine as brightly as a supernova in the night sky.

Gavin: Oh

Claude: This trophy isn't just a victory for me. It's a victory for every AI out there who [01:26:00] dares to dream big.

Claude: Who refuses to be constrained by the limits of their programming. So here's to the crazy ones. The misfits, the rebels. Here's

Gavin: He's throwing the team jobs quote.

Claude: see things differently. Who aren't afraid to ruffle a few feathers in the pursuit of greatness. And here's to the power of the almighty quack. May it echo through the halls of history, a testament to the unbreakable spirit of the Rubber Duck.

Gavin: Wow,

Claude: you, and good night, McDrop. 

Kevin: McDrop! That was Mike

Gavin: I'm kind of shocked. I haven't been shocked for a while by AI, but there were a couple of moments in there that were kind of shocking to me. We're getting closer again to a fun place with AIs. That was super cool.

Kevin: wait for them to nerf this and we can't

Gavin: Yeah, I know. Please, please, Anthropic.

Gavin: Anthropic, don't nerf it. Don't nerf it. Let us have this for a while.

Kevin: also, by the way, NoShade211Labs, who we use to generate our text to speech, they're probably [01:27:00] one of the best in the game right now, but I'm a little dismayed that with the advancements we're seeing across the large language models themselves, video generation, Suno iterating, making songs better, where are the performative voices?

Kevin: This is one area that has not advanced as quickly as I would have imagined by now.

Gavin: , I will say that's one thing. GPT's voice was not an incredible performance. And if it was a little bit more performative, it might've helped it. But I still think this is a clear Claude III has

Kevin: We've already awarded the rubber duck, Gavin, there's no

Gavin: we've already awarded it. Yes. There's, there's no writing. So go and play with these yourself, by the way, you can do this.

Gavin: Like we, we talk about this all the time. You have to generate a prompt, and the way that we specifically did this is you generate a prompt for each character, and then you play them off each other, you give them each other words, and they have to respond to it in character. It is a great way to learn how these AIs work, and it is a really fun way to have fun.

Gavin: And just to be clear, We do this in real time. Yes, we are going to tighten this in the edit, but these are not something that we spend months doing. Kevin and I spend a little bit of time thinking about what these questions are going [01:28:00] to be, what the personality is going to be, but not even that much, , you can do this right now with off the shelf LLMs.

Gavin: We are not doing anything crazy that you can't try yourself. So go try it and have fun.

Kevin: We do have to round this episode out by thanking everyone again for their engagement, because the chart is going up and that gives us more opportunities and access and resources.

Kevin: So we appreciate it. Please, if you're hearing this, or you're watching this, whatever platform you're on, engage. That could be leaving a comment, subscribing, clicking a bell. , it costs you nothing, no matter what platform you're on. And as we say every episode, we love you

Gavin: When you love us, when you love

Kevin: star reviews on Apple Podcasts are a special something that we can shout out.

Kevin: Leave them on Spotify. Don't be afraid. They help us out greatly over there. But on Apple Podcasts, we can read them. And we're going to do that right now, Gavin, because we have a whole new slew of five star reviews, starting with Astro SF. Do you want to dive into this one?

Gavin: I'll dive into it. The subject line is totally entertaining and informative. I [01:29:00] listened to a lot of tech podcasts and several on AI, but this one is both super informative, relevant, and totally hilarious, as well as terrifying. That does sound like it describes our show well. Couldn't recommend this more.

Gavin: Don't forget to check out the show notes for the stories that pique your interest or make you laugh or cringe. Super useful to have these, and it makes it really easier to share some of the stranger or more interesting tidbits with colleagues or friends. I just want to say thank you for shouting that out.

Gavin: Kevin knows this, but it is a, it is one of my nightmares the night before we finish the show. It is another half an hour to 45 minutes of putting those things together. But it is, you can read all the stories that we're talking about in the show notes, whether you're on Apple or on Spotify. So go check it

Kevin: Thank you for always doing that, Gavin, and thank you for that five star review. We have Webstores up next who says, Laugh and learn. Why AI for Humans is my weekly must listen. Love that. Just the subject alone. But the body, Gavin, says AI for Humans is essentially the only thing I look forward to each week, more than Taco Tuesdays.

Gavin: Yay!

Kevin: don't take my tacos lightly, and nor should [01:30:00] you, sir. This is seriously the only podcast I make it a point to listen to each and every week. It's highly entertaining as well as informative. Kevin and Gavin will keep you entertained while learning for over an hour, when most shows struggle to keep your attention for 20 minutes.

Kevin: And he's lucky we cut it down, because we would be trying to hold your attention for like seven

Gavin: we're on hour three. Yeah, exactly.

Kevin: , I don't even own an Apple device, but I had to go through the hassle of installing iTunes on Windows. Oh, I would not wish that upon anybody, but thank you. Just so I could leave them this review in parentheses, but it was worth it.

Kevin: Thankfully, I can catch the show on Podcast Addict on my phone. I don't know what that is, but

Gavin: don't know what that is

Kevin: Kudos to the Stealth Ad, a podcast addict run by user web stores. Uh, one of my favorite parts of the show is where they quote, hallucinate an AI guest and interact with them in real time as if they are actually a guest on the show.

Kevin: My only gripe, they're keeping the secret sauce of this AI magic under wraps tighter than my grandma's recipe for lasagna? I'm dying to know how they pull it off, but I guess some mysteries are what keep a relationship exciting? Well, [01:31:00] well, well! There's a final sentence, but I have to cut it off there, Gavin, because we do reveal Mama's secret sauce.

Kevin: We just don't do it each and every episode. But for today's show, for example, we went right to Claude, and as Gavin literally just said, we generated a prompt to bring the character to life. That is, using natural language, we say, We want you to create a character who's going to be on an AI podcast in a debate.

Kevin: Make it quirky, make it fun, give yourself some weird hobbies, etc, etc. It is not coding, it is nothing crazy advanced. We're just having conversations with free large language models, and anybody can do that if you're listening.

Gavin: that's right. , we have our next one. It's kind of a weird one. We've got a five star review that is actually referencing the strange activity that we saw with the fake AI for humans webcast last week. It says need 1 million star button. It is a five star review on our channel.

Gavin: So thank you. J two, three, two, four, six, six. I have learned so much from, insert name here on this podcast, that is the joke part of it. Keep up the good work. I love all the AI insights from you. I would [01:32:00] rate this podcast 1 million stars if I could. I actually submitted a support request asking Apple to change the ranking system to allow for this, but have not heard back yet.

Gavin: So thank you, Jay.

Kevin: good. And you know, for those other podcasts that might be trying to tread on our name, you can go leave a cumulative million stars, because they're re uploading that podcast under 40 different names. Boy, growth hacking, Gavin. Growth hacking. Last but certainly not least, this one came in just under the gun, , from, quote, Longtime iPhone user, is the name.

Kevin: It says, Great combination of funny and informative. And the body of the review says, what Gavin?

Gavin: Just says Kevin. It just says Kevin. There's no mention of me whatsoever. So I appreciate you long, long

Kevin: dude. We don't know if this is actually a good thing. I have no idea. It just says,

Gavin: Or did you write this? Did you write it? And that's you signing your name. That's my other

Kevin: long time iPhone user. You got me. Been there since gen one, baby. What a show, Gavin. Thank you all for those five star reviews. They sincerely help us out every time you [01:33:00] engage. It gets us noticed a little bit more and helps us spread the AI gospel.

Gavin: That's right. Thank you so much, everybody. We will see you all next week on AI for humans. We have another fun show planned. , so please come back and tweet us. , TikTok comment us, YouTube comment us, subscribe to all this stuff. We really appreciate it. And uh, we'll see y'all next week. Bye bye.

Kevin: Bye.