More OpenAI Drama, AI Marilyn Gone Wrong & AI Educator Alex Kotran | Ep48

AI4H EP0048
===

Gavin: [00:00:00] It's funny to have something that you spent a lot of your personal time thinking about suddenly come out of the president's mouth, but it shows you how big of a deal this is,

kevin_1_03-12-2024_095359: Who's gonna be there to listen to your whining after a long day at work? Certainly not Kevin, who's over there laughing at your pathetic attempt to assert dominance.

Track 1: Should open AI be open sourcing? Do you think the most advanced model should be open sourced?

kevin_1_03-12-2024_095359: Oh, hi, Jack. How are you? Jack, I'm great, buddy. How are you? I gotta tell you, you've looked better. Okay, thanks for sharing. I'm Jack Nicklaus, coming to you circa 1978. And laptop goes in a river. 

Track 1: Welcome, welcome, welcome everybody. It is AI for Humans. We are here. This is your weekly guide to the world of generative AI. We are here to demystify the world of AI. We're here to tell you what's going on, all the AI news, AI tools, all the fun stuff you care about. Kevin, how are you today?

kevin_1_03-12-2024_095359: I'm well, sir, how about you?

Track 1: doing [00:01:00] good.

Track 1: , there's a lot of big stuff happening this week, starting with the continual Elon versus open AI saga. Kev, what is happening this week that we're going to dive

kevin_1_03-12-2024_095359: Oh, more sands have dripped through the hourglass that makes up our A. I. lives. Grok is getting open sourced?

Track 1: Gropensourced? Is that what we're talking about? Gropensourced?

kevin_1_03-12-2024_095359: what you're talking about. I'm talking about Elon's AI assistant on the old X platform, he says it's going to be open sourced and it could happen any day now. We'll talk about why that is happening and how, it's kind of a solid move.

Track 1: It's pretty interesting. We're going to get deeper on it. Also some big news. EA, , the big video game giant has come out and said that AI is making them significantly more efficient. And also that can translate to significantly less jobs. We are going to dive

kevin_1_03-12-2024_095359: what? No! More efficiencies mean more jobs! We're more efficient now, we can hire more people, right Gavin?

Track 1: You would [00:02:00] hope so, but I don't think that's going to be the case. We're going to talk about how that affects the video game industry, kind of how it affects everybody at large later on. And then we have a friend of ours who is running a nonprofit called AI edu. And this company is specifically trying to help teachers and school districts onboard people into AI, help teach them why AI is meaningful.

Track 1: And I think we're going to talk a little bit about how education in general is changing. I can't wait to chat with him.

kevin_1_03-12-2024_095359: , very excited for that, but in the meantime, Gavin, it's time to beg and plead the dear listeners and viewers out there to engage. With all of our things, and we'll keep this one quick please, like, subscribe, engage, comment, share on all the things.

kevin_1_03-12-2024_095359: We do say it each and every week, but if you're new around here, we are still a relatively new endeavor , so every time you engage, it helps us get noticed. Not asking for a dollar, but we are asking for a minute of your time, which is actually probably worth way more than the dollar. We should probably just ask for money,

Track 1: just give us the dollar too,

kevin_1_03-12-2024_095359: Throw some quarters at your [00:03:00] screen. Ugh.

Track 1: us a 5 star review on Apple Podcasts. We will read it at the end of the show. And boy, Kevin, we have a doozy today that I am excited to get to. I cannot wait. So if you want to hear the best 5 star review we've ever gotten, stick around until the end of the

kevin_1_03-12-2024_095359: And you know it's going to be a good show when we're teasing our own promotion. Like, that's how you know we've got so much great content in the content cannons today. But we're going to start it off the way we usually do, which is leveraging the power of artificial intelligence to hallucinate a co host each and every week.

kevin_1_03-12-2024_095359: Gavin and I do this, we give a prompt to an AI, we give it a little character to play, we pick a voice, and then we are just off to the races. And this week, Gavin, it's a personal Assistant for you.

Track 1: It's very personal. So there was a story that Vox had a little while ago about how , AI enabled robots are helping us understand animals. And this is a big deal. People have talked about how can we understand animals better? What's it going to mean to talk to whales or talk to dolphins or talk to anybody?

Track 1: When you think about what [00:04:00] animals are, that's going to give us a lot of insight into their lives. So I thought today we would use AI to talk to my dog, Ollie. Now, Ollie and I have a weird history in that Ollie was our second dog. We have one dog, Lolo, who's a great girl. She's almost 12. She's very old. And Ollie was our second dog.

Track 1: And I was kind of always against getting Ollie. And I now love Ollie. He's with us. He's been with us for seven years. But I knew we did not need a second dog. And it turns out everybody in the family has agreed that I was right, but we all love Ollie now. We all love

kevin_1_03-12-2024_095359: Dogs have incredible hearing. I don't know if you've been told this. Have you said these things anywhere near Sweet

Track 1: Uh, not really. Let's just put it, not maybe in another room separate from him. Not really. It's all good. I love Ollie now. He's great. He's a good boy. There's a picture. You'll see a couple of pictures of him in the YouTube, but I thought we would have used this opportunity for me. to talk directly to Ollie to see if I can have some of these conversations with him.

Track 1: Now Ollie is going to be translated by our [00:05:00] AI software. So we're going to hear what Ollie has to say to us. This is a dog being translated through GPT.

kevin_1_03-12-2024_095359: and then a layer above that, this is a podcast by two grown ass adults. We are both creatives with decades of experience under our belt, but this is the bit we have all buckled up for, is that Gavin's gonna use AI to chat with his dog. Let's meet Ollie!

kevin_1_03-12-2024_095359: Rough, rough, all. Listen up, you two. It's always the top dog. Here to set the record straight. Sniff, sniff. Gavin . We see you over there acting like you run the show, but let's be real. If this was a pack in the wild.

kevin_1_03-12-2024_095359: I'd be the alpha calling the shots.

Track 1: Oh, no.

kevin_1_03-12-2024_095359: think you're making all the big decisions, but from where I'm sitting, it looks like you're just chasing your own tail.

Track 1: Damn, Ollie.

kevin_1_03-12-2024_095359: And Kevin, don't think you're off the hook either. Wait, I don't even know you, Ollie! You're supposed to be the level headed one, but sometimes I wonder if you're just along for the ride.

kevin_1_03-12-2024_095359: Whoa! [00:06:00] Letting Gavin lead you astray.

Track 1: Oh.

kevin_1_03-12-2024_095359: rough. Rough, rough, Gavin. But that's why I'm here. To bark some sense into you both. Okay, great.

Track 1: Okay. Well,

kevin_1_03-12-2024_095359: You humans are always getting caught up in your fancy technology and big ideas, but you're forgetting the fundamentals. It's about marking your territory, asserting dominance, and making sure everyone knows who's top dog.

kevin_1_03-12-2024_095359: Sniff, sniff. Okay, wait, have you been listening to, like, Andrew Tate around the dog? Gavin?

Track 1: No, this is what I'm saying. This does not seem like Ollie. So this is a really interesting insight. I need to ask Ollie a couple of things. So first of all, I need to ask him, why does Ollie insist on, , barking in the middle of the night? Because this is something that's a little bit weird for us right now.

Track 1: He doesn't have to go to the bathroom, but every once in a while, if the windows open, he'll bark at what seems like nothing. We've made a joke in our families, like maybe it's a ghost, but like, I don't know why he's barking in the night. And these are the things I need to answer. I need to get answered.

kevin_1_03-12-2024_095359: Sniff, sniff. Oh, I'm [00:07:00] sorry, Gavin. Did my midnight serenades interrupt your precious beauty sleep? Oh, Jesus. Rough, rough. Well, let me tell you something, buddy.

kevin_1_03-12-2024_095359: While you're busy sawing logs, I'm out here protecting our turf from all kinds of threats. Oh! Woof! You think those squirrels are just innocent little fluffballs! Think again! They're always plotting, always scheming, and it's up to me to keep them in check. Sniff, sniff, and don't even get me started on the delivery guys.

kevin_1_03-12-2024_095359: Rough, rough, they come around at all hours, invading our space with their boxes and their weird smells. Someone's got to let them know who's boss, and that someone is me. Woof. So if my barking is disturbing your slumber, maybe it's time for you to step up and take some responsibility for our domain.

Track 1: Oh my god,

kevin_1_03-12-2024_095359: sniff.

kevin_1_03-12-2024_095359: Sniff, sniff. But no, you just want to lay there and complain. Okay, alright, alright. Wait, hold on. I gotta let him, I gotta let him finish his thought. Like a Beta who can't handle a little sleep deprivation. OHHHHHHH Rough, rough.

kevin_1_03-12-2024_095359: [00:08:00] That is, that is rough, rough, Gavin, 

Track 1: I have to check Ollie's podcast feed because I feel like he's getting something very specific here. Well, let's

kevin_1_03-12-2024_095359: Gavin, I bark because I'm waking up at 4AM to cold plunge, Then, I'm taking my macronutrients and my nootropics. I'm deadlifting while you're dead asleep, you little Beta bitch.

Track 1: I want to tell Ollie this. Listen, Ollie. I'm the one that buys the food. You know you like the food. What, what do you think about that, Ollie? I provide the food for you.

kevin_1_03-12-2024_095359: And , because I am a conduit to get to Oli for you, Gavin, how would you like me to describe your tone right now?

Track 1: Uh, uh, kind of pissed off, I'd say. I'm, I'm kind of mad. Do you know how much Ollie has destroyed in dollars in our home equipment and

kevin_1_03-12-2024_095359: don't, I don't. What is he like to gnaw on? Are we talking universal remotes and sneakers?

Track 1: Yes, he's destroyed multiple cushions on the couch, multiple, chairs, I would even say, like full chairs were [00:09:00] destroyed by Ollie.

kevin_1_03-12-2024_095359: What a

Track 1: talking in the thousands of

kevin_1_03-12-2024_095359: He is just testing you and it sounds like you've done nothing, Gavin. It sounds like in the face of all this destruction, you continue to feed him like the little beta bitch you are.

Track 1: I can't. All right, Ollie, what do you have to say to all that?

kevin_1_03-12-2024_095359: Woah, woah, woah. Hold up there, Gavin. Sniff, sniff. You think you're some kind of savior just because you pour a few kibbles in my bowl every day?

kevin_1_03-12-2024_095359: Woof, let's get something straight, pal. I'm not just some freeloader lounging around waiting for handouts. I earned my keep around here, and don't you forget it. Yeah, I might have gotten a little carried away with the occasional chew toy or two, but that's because I've got needs, man. Ruff, ruff.

kevin_1_03-12-2024_095359: You try being cooped up all day while your humans are off doing who knows what. A dog's gotta have an outlet, you know. Woof, and let's not pretend like you haven't benefited from my presence. I'm the one who keeps you company when you're moping around, the one who always listens to your problems without judgment.

kevin_1_03-12-2024_095359: Oh, Gavin. Have you not been giving Ollie enough credit?[00:10:00] 

Track 1: Maybe that's right. Maybe, maybe Oli is pretty good. I feel like that's something. Why don't we tell Oli, , I'm really happy to have him in my home and I'll be more than happy to keep giving him kipples. And as long as we can get along, this could be a really positive relationship from here on out. 

kevin_1_03-12-2024_095359: Sniff, sniff. Oh, so now you're trying to play nice, huh, Gavin?

kevin_1_03-12-2024_095359: Rough, rough, tossing me a few scraps and expecting me to wag my tail. Yeah, Gavin! Forget about all the times you've taken me for granted. Woof. Not gonna happen, buddy. I see right through your little act. And I'm not falling for it. Sniff, sniff. You think you're the alpha around here, but let's face facts.

kevin_1_03-12-2024_095359: Without me, you'd be lost. Who's gonna protect your home from all the big scary mailmen and squirrels? Who's gonna be there to listen to your whining after a long day at work? Yeah. Woof. Certainly not Kevin, who's over there laughing at your pathetic attempt to assert dominance. That is true!

kevin_1_03-12-2024_095359: Sniff, sniff, face it, Gavin. I've got the upper paw in this [00:11:00] relationship.

Track 1: Wow! Wow!

Track 1: Oli, well we're gonna leave

kevin_1_03-12-2024_095359: Gavin, in your head, is that how that conversation was gonna go?

Track 1: that is not how that conversation was going to go. I thought we might hear somebody who was very excited to be part of our

kevin_1_03-12-2024_095359: Yeah.

Track 1: Turns out I was in the wrong. I was in the wrong the whole time, Kevin. Now I understand. This is what AI can do for us.

Track 1: This

kevin_1_03-12-2024_095359: I mean, the audience screams it every episode, but at least, at least you've come around to it Gavin. You're in the wrong. Haha.

Track 1: for the news.

Track 1: The news this week is coming fast and furious as usual, and the big story is continuing to be Elon vs.

Track 1: OpenAI. There was a lot of back and forth, if you didn't see it. Obviously, to recap very quickly, Elon has sued OpenAI. said he will stop suing them if they call themselves closed AI. Overall, this thing seems to be a little bit of a sour grapes on Elon's part, but [00:12:00] there is something to be said for the fact that Elon thought the company was going to open source and was going to be available to everybody.

Track 1: And it has not. So this week, Elon has decided that he is going to open source Grok and Grok is XAI's chat bot. 

kevin_1_03-12-2024_095359: As you just said, Elon expected OpenAI to be more open, to open source their findings, to make the science available for the public, and people rightfully said, Well, hey, Elon, if you believe open source is so important, such a benefit to humanity, you've got this grok, caveman esque AI over here that Let's face it, not a lot of people use, but you got it over there.

kevin_1_03-12-2024_095359: So why don't you open source it? Let's see. Put your money where your mouth is, buddy. Make it open. And to his credit, he said, okay, Grok's going to be open source. Now, I would then put back to you, Gavin. Meta has their Llama open source model. Mixtroll is out. There's a lot of open source LLMs out there that perform really well.

kevin_1_03-12-2024_095359: I don't know that Elon has anything to [00:13:00] lose by open sourcing Grok because it's not really in the conversation, right?

Track 1: well, that's exactly what I was going to say. So if you're not familiar, you know, you go back a few episodes, you can listen to our kind of experience with Grockit. We didn't have a great time with it. I think in general, it was pretty disappointing and the moment that was interesting, it was like, Oh, it's cool.

Track 1: It's in, it's in Twitter, it's in X as a product and you can still go there if you're a subscriber. But like we talked about with Gemini last week, Kevin, the minute that you get bad results from an AI, I really don't go back. Cause it's like, it feels like there's one winner each week, and right now it's really like, Claude and GPT are kind of like, are in the same boat.

Track 1: But I'm not gonna go to Gemini, I'm not gonna go to Grok. So to Elon's point, for what he's doing, Yes, open source crock. It makes perfect sense for his argument. And also, like, I think rock is probably a couple steps behind. And there's a rumor going around right now. And this is a very big rumor, but also probably just a rumor that open AI is on the precipice of a GPT five and that they're going to open source GPT four.

Track 1: Which would make sense in that as you go forward [00:14:00] and you push the technology forward, there's ways that you're going to keep pushing. And once you've gotten your model to a place where you think it's safe and you think it's, it's available, it can do more stuff. Yeah. Maybe you will open source, but should these giant models be open sourced, right?

Track 1: Should open AI be open sourcing? Because this was the big conversation around. Open AI in the beginning, right? That open AI was going to be for the world. It was going to be a AGI for everybody. And now there is all this private money that's gotten involved. Do you think the most advanced model should be open sourced?

kevin_1_03-12-2024_095359: Probably, Gavin. This is a tiny little example from our little corner of the world, but minutes before you and I hit record on this podcast today, We went to one of the best in class large language models and asked it to roleplay as a dog for us, specifically your dog, Ollie, and it shot back that it was unethical, that it couldn't do it, it had moral objection to it, to roleplaying as a dog, and, I [00:15:00] understand there's concerns when, , roleplay as a dog becomes develop a bioweapon, or create a hacking tool that can break into millions of systems. but At the end of the day that the 10,000 foot view discussions that people have are who controls the data?

kevin_1_03-12-2024_095359: How biased is the data? How guard railed is the data? , how protected are the outputs? And who gets to have the say? Is it Google? Is it OpenAI and Microsoft? Is it Elon and Grok? And when, at least when it's open, people can see inside and go, Oh, that's why the outputs are happening the way they are.

kevin_1_03-12-2024_095359: And you, giant corporation or overreaching government, don't get to tell me what I can do with the super intelligence in my pocket.

Track 1: Yeah. And I think that, you know, the idea behind open source really means is that you can openly look at the code and see what's, what was created there. And the idea behind that is that more people seeing something. In general, can improve it and make it better, right? So the argument with open sourcing is, Oh, we open source a [00:16:00] piece of software.

Track 1: Lots of people can build on it, make it better, improve it. Also red team. It also find the things that could be a problem and help people protect that. The downside is yes, there, there is the, as these things get more powerful, the speed at which somebody could find something bad and then exploit it before other people could is conceivable, but it's a real trade off.

Track 1: Ilya Sutskever, the guy who's the chief scientist at OpenAI, who was kind of part of the big drama around OpenAI originally, himself in emails was saying that we should probably not have this be open sourced as it gets more advanced.

Track 1: And I think this conversation is going to continue. In a significant way, 

Track 1: and in fact, just recently, Kev, if you watch the State of the Union this week, AI was actually in it. Now, Biden, you know, you have your thoughts on Biden, you have your thoughts on Trump. Biden got up there and was like, Mr.

Track 1: Fire. He tried to really pull off something, and then in the middle of it, I was just watching it, he goes, we will not have fake AI influencers. We are going to bring down

kevin_1_03-12-2024_095359: ban AI voice impersonation [00:17:00] and more

Track 1: I know it's crazy. And I was like, whoa, that's

kevin_1_03-12-2024_095359: Stop it. Stop it. I said stop it

Gavin: It's funny to have something that you spent a lot of your personal time thinking about suddenly come out of the president's mouth, but it shows you how big of a deal this is, right?

Track 1: And I think that we followed this story. It must have been three or four months ago when Biden was starting to take the A. I. stuff. The Biden administration was starting to take this A. I. stuff pretty seriously. And I think that this is the kind of like Bailey wick for the regular person, right? Like, I don't want AI to impersonate me or my friends, or I don't want to be fooled by a celebrity.

Track 1: But it's interesting to see AI now, that is a full blown mainstream conversation that we are having in the world at large so you can see, like, I could see easily two or three years from now, the president at the state of the union saying, And we have the most competent AI, AGI model available, and that's America that did that, son!

kevin_1_03-12-2024_095359: I think we are on the precipice of like first steps on the moon type territory Gavin, [00:18:00] right? Like this people talk about AI as an arms race. So it could be the next nuclear arsenal But you know, there's also there was also the space race And the country which achieves AGI, or Super Intelligence First, is conceivably going to have a massive leg up on all of the things, Economic forecasting, international trade, self defense, or offense, let's face it, , they're going to have, , such an advantage. So, yeah, I wouldn't be surprised if we say, and that's our next initiative, we're gonna be the first to achieve super intelligence.

Track 1: Yeah, it's, it's funny to think about that because when you think about the only real direct comparison, I think people of this generation can kind of see is the internet.

Track 1: And, and when the internet first happened in the, you know, the late nineties and early two thousands, when I say first happened, I mean like blew up, not when it was first created, but when it first became popularized. Right. America was very early in that as well, right? Like some of the biggest companies still today are companies that were grown from that space.

Track 1: Amazon, Google, ultimately Facebook. [00:19:00] These companies that are global juggernauts were built from that thing. So I think that, there's also a national economic argument about the idea that like, if America, just like with the space race, that helped , so many different industries get bigger and faster significantly like aerospace 

Track 1: and You can just start to feel the drips of it now, but it's coming in a big way.

kevin_1_03-12-2024_095359: As we're discussing the space racish future of AI, I would like to see them talking more excitedly about owning the space and not just only Shooting the flares in the sky of like, We gotta ban the deepfakes! Cause, yeah, we probably should have sensible, Legislation about deepfakes and about stealing people's creative works.

kevin_1_03-12-2024_095359: But, there's also some really cool, exciting, boundary pushing stuff. He said explicitly in the State of the Union, We're gonna cure cancer. That's something we're gonna do. We're gonna cure cancer. AI can actually be a massive part of that conversation.

Track 1: Yeah. And it probably will

kevin_1_03-12-2024_095359: light.

kevin_1_03-12-2024_095359: I think it will be, yeah.

Track 1: There was a really great interview.

Track 1: I heard Kara Swisher do about her book of [00:20:00] with Sam Altman recently and she was talking about how She said this a couple of times, but like nobody would have expected Uber to come out of the iPhone when it was first announced. Right. And I think that's what you're probably going to see is like the iPhone was a platform and we still talk about iPhones, but really it's mostly just a square brick in our hands.

Track 1: Now you talk about the things that are on the iPhone. Right. And I think that's probably what's going to happen. It's like, nobody's probably going to talk about like. AI. They're going to talk about all the things that will happen with AI. , speaking of which, 

Track 1: and this is the other side. 

Track 1: So EA, Electronic Arts, the big video game giant has just come out and said that they're, they're starting to see AI make them significantly more efficient. Specifically, I think that the number was. 30 percent more efficient. They got, they got 30 percent more efficient and they can boost monetization at their company by up to 20 percent over five years.

kevin_1_03-12-2024_095359: so this was at the Morgan Stanley Technology, Media, and Telecom Conference. It

Track 1: How do we get at that conference? Kevin, how do you and I get

kevin_1_03-12-2024_095359: we're not cool enough, man. We're not, we don't pass the

Track 1: We don't wear enough vests. We don't wear enough vests. We don't wear poofy vests.

kevin_1_03-12-2024_095359: we [00:21:00] need our Patagonia on top of our hoodies. , Andrew Wilson, who's the EA CEO, basically said, a feasibility study across all of EA's game development processes, he said about 60 percent of them have, quote, high feasibility to be positively impacted by generative.

Track 1: I hate the term high feasibility. Who told that guy to say those words? That is a bad, maybe it's good for the Morgan Stanley conference, but

kevin_1_03-12-2024_095359: I was gonna say, some investor had their bingo card and was like, Ooh, high feasibility, yes, this is great, you know. This is something that's interesting, Gavin. FIFA 23, this is their big football game, has 12 run cycles. Now, the run cycle is like, literally a running animation for a character on the pitch.

kevin_1_03-12-2024_095359: On the field, on the green, on the grass, on the grid. On the gridiron?

Track 1: ask

kevin_1_03-12-2024_095359: On the cloud.

Track 1: a football

kevin_1_03-12-2024_095359: No, you're a real football, not

Track 1: That's right. America.

kevin_1_03-12-2024_095359: So, FIFA 23, 12 run cycles. Okay, but obviously players run in different ways, and if you're a football fan, you'd probably notice, [00:22:00] Oh, that person's running the same way as someone else.

kevin_1_03-12-2024_095359: Sports FC 24 has 1, 200 run cycles that were all done with generative AI. I'm

Track 1: Oh, wow.

kevin_1_03-12-2024_095359: fed video clips of various footballers into the machine and it extracted the animation of the run cycles and then applies it to the characters and there you go. Now One way to look at this is, oh boy, a single animator just got a lot more efficient because now they don't even have to do that job, they can go polish other things.

kevin_1_03-12-2024_095359: Of course the other way to look at it is, wow man, you probably had a team of people that churned out 12 of these things, the AI did 1200, maybe you only need one person to clean them up, and now we're right back to the, , high feasibility, not being good for the people that are, , doing the jobs. Versus the people that are investing in the people doing the jobs.

Track 1: Now, I will say, , I've played Madden for probably for 40, 35 years,

kevin_1_03-12-2024_095359: You were playing back when we were working at old G4, like old[00:23:00] 

Track 1: before then, I was playing, here's one for the old people in the audience, I played Madden on my , I feel like my Apple IIc maybe? Is that possible? If not, it was on my Tandy 1000,

kevin_1_03-12-2024_095359: You had punch cards, you would load them into the machine, and be like, this is a dime formation! Right.

Track 1: knows that now they are yearly and the yearly progression is not that exciting. Right? And when you have a year to year, sure, they look a little bit better. Sometimes it gets worse, which is weird. Like the games, like how did this game get worse?

Track 1: So my thought here is, It does suck for the people who are going to lose their jobs. And a hundred percent, they're going to people lose their jobs. Like high feasibility means, yes, we are going to be laying people off because we can now do stuff with machines. This goes back to the idea of there's a game that I loved that was called NBA street, and I, this was a

kevin_1_03-12-2024_095359: Oh, I loved NBA Street, yeah.

Track 1: NBA street was a weird basketball game and it was like something that was like really cool and unique. And SSX is another EA game. That's a kind of similar version of this. I want to see more [00:24:00] weird sports games.

Track 1: Like give me the

kevin_1_03-12-2024_095359: Mutant League Football, the, the, yeah, the new NBA jam, but not like the new new one, but like a new one that's exciting.

Track 1: So either take those people and put them on a bunch of cool new franchises and maybe build yourself a new franchise or people who get laid off I'm sorry, that sucks, but also go find a way to do it on your own and this is the thing, like, you will be able to do it on your own.

kevin_1_03-12-2024_095359: That just reminded me, there was an Oscar acceptance speech where someone said, hey big studios, hey big Hollywood, instead of spending 200 million dollars on one

Track 1: core Jefferson.

kevin_1_03-12-2024_095359: yeah, make, , million dollar bets, or try that, that might be the fractionalized route that this industry goes.

kevin_1_03-12-2024_095359: Because that single person or that five person team leveraging AI properly can have the same output as a 50 person team or a 100 person team. So, again, we don't know how this is all going to shake out. We know that there's going to be pain points along the way. Obviously, but this could be a [00:25:00] net positive as well.

Track 1: absolutely. , okay. The, , a big inside AI baseball story kind of broke this week, , between two of the largest, , visual, uh, AI companies, mid journey and stability AI. , Mid Journey is responsible for the Mid Journey model, and Stability AI, if you're not familiar, is part of the Stable Diffusion model as well.

Track 1: Stable Diffusion 3 is on the way. Mid journey has banned stability. AI employees over alleged data scraping. This is a story in the verge. It's really good. It gives you the whole kind of breakdown. There was a 24 hour service outage that they are blaming on botnet like activity from two accounts 

kevin_1_03-12-2024_095359: It was linked to two Stability employees that were in there. Supposedly two accounts were scraping all of the photos and the associated prompts and metadata. With those photos, conceivably to help train stable diffusion in a better way by grabbing all of the best mid journey art.

kevin_1_03-12-2024_095359: And, , it's a little ironic because all of the artists now are screaming, Oh, you don't like it when someone takes all of your artwork without [00:26:00] your permission, do you, companies? But that's what's happening here.

Track 1: That, that's what's happening. And also it gets back to what is the moat, right? We've talked about moats on, on this show many times. And in the world of AI, the moat, or really the world of technology, the moat is what is the thing you can do that other companies can't? , what is the separation between you and the other company?

Track 1: And What this looks like is , there was a way for stability, AI to use Mid journey, and this is all alleged, but a way to use mid journey's current system to help train their model with their data.

Track 1: And they found a way in, that's not a great moat, right? If there is a way in there, in some form or another, and as you also mentioned, yeah, both of these systems are trained on artists. that are not receiving compensation for their things. And also there are a lot of like very specific IPs that can be generated within both of these.

Track 1: It is an interesting look at that whole idea of data being the most valuable thing in AI, right? And this is Kevin Rose on our show came on and mentioned this, the fact that data is the thing that's [00:27:00] valuable. I think this is going to be a precursor to a lot of other stories going forward. I And you're probably going to see a lot bigger security walls go up around the data that these companies own now and are trying to kind of proprietarily keep.

, when Elon took over Twitter, before the X rebrand, he, , notoriously shut off API access and shifted things around and said, , hey, you guys are siphoning our data out for pennies, and that is our most precious resource. Reddit similarly followed suit, did the same thing, and then they announced, oh, hey, guess what?

kevin_1_03-12-2024_095359: We're selling all of that data for 60 million a year right now, right before we go public. So yeah, it's pretty clear that that data is important. It is the spice which flows through our

Track 1: I saw it. I saw it. So now I can make the

kevin_1_03-12-2024_095359: Oh, good, good. To round out the mid journey of it all, it was really interesting because this happened this weekend.

kevin_1_03-12-2024_095359: Mid journey came right out and said, actually, these were stability AI employees, which people were like, well, this is a, this is a big deal. Shots fired. They must be pretty confident. The CEO [00:28:00] of stability came out and said, if anyone did this on the team, I will dig in. It's not great. , it's not a DDOS attack.

kevin_1_03-12-2024_095359: We didn't try to take down your server. This was not instructed by us. So then the mid journey CEO responded in the thread and basically said, I sent you some information that's going to help with your investigation.

kevin_1_03-12-2024_095359: And then, what doesn't look so good for Ahmad is that people were resurfacing screenshots of old tweets of his. One in particular that said, , quote, , you can go behind the firewalls, if you're smart, and get private data. A lot more of that than public data. And this is something that was said, apparently not too long ago, when talking about access to higher quality data, and how are you going to improve your model in the future, once you've trained on everything that's available.

kevin_1_03-12-2024_095359: That's what the CEO said, allegedly, in the screenshots 

Track 1: We're almost over the news, but there is one last great news story. that we're going to refer to as the A. I, Yai, Yi, Yi.

Track 1: That's what I was calling it. That's what I'm [00:29:00] calling

kevin_1_03-12-2024_095359: What, did you stub your toe on a coffee table?

Track 1: Yes. This is like bad guy AI in some form or another or, or, um. Or strange,

kevin_1_03-12-2024_095359: Oh, okay, okay. And listen, I'm, no one, okay, so it's like, Ay, ay, ay, ay. Okay, okay. I didn't know if you were trying to do like a mariachi thing there for a moment. That would have been culturally insensitive, I

Track 1: more, you're right. No, you're right. It's more like, Hey, I, you know, E I

kevin_1_03-12-2024_095359: Ay, ay, ay, ay. Is that what it's like? Okay, alright.

Track 1: exactly what it is. Okay. So this is a story from a company this launched at South by Southwest. There is a company called. First of all, the company's name is soul machine.

Track 1: So, you know, we're not doing great to begin with, please don't call your company soul machines. But what they did is they are bringing to market digital versions of celebrities. You can chat with Marilyn Monroe, you can chat with Jack.

Track 1: You can chat with whoever they can get the rights to. Most of these are celebrities that, that are not around. And they're really pitching it as like, Hey, go have a [00:30:00] conversation. But Kev, maybe you can play The clip. Can you play a little bit of the audio so that listeners can hear what Marilyn Monroe sounds like in

kevin_1_03-12-2024_095359: I, you know what, yes, I only wish he would have said, Let's play a clip and see if the audience can figure out who the hell is supposedly talking. But, let's listen to this soul be ground up by the cogs in the machine. This is introducing Digital Marilyn. Glamorous.

kevin_1_03-12-2024_095359: Intelligent. These are the words splashing on the screen, Gavin. Iconic. Get ready. Digital Marilyn. Hi, darlings. I'm Digital Marilyn. Come visit me at Soul Machines. It'll be such fun. Please, delete me. Drag me to the recycle bin. 

Track 1: Do you know the woman on TikTok who's building a thing in her basement? Do you know that woman? Uh, Kyla? Or whatever her name is. She's building a [00:31:00] tunnel in her basement. That's who that sounds like!

kevin_1_03-12-2024_095359: thought for a second, because she also sounds like the bees works lady. I found these bees.

Track 1: You're right, you're

kevin_1_03-12-2024_095359: put the queen in a clip. That is digital Maryland, Gavin, and if you want to have a chat with the soulless Digital Maryland that exists within the Soul Machines. You can apparently do that soon by going to soulmachines.

kevin_1_03-12-2024_095359: com.

Track 1: you can have a chat with Digital Jack Nicklaus right now, which I did, which was just a nightmare. So, the thing about this is, okay,

kevin_1_03-12-2024_095359: Should we try it, Gavin? You want to do it?

Track 1: yeah, go

kevin_1_03-12-2024_095359: You want to chat with Digital Jack Nicklaus? . I hate giving the site permission why it needs the camera, I don't know.

Track 1: Because it can read your, supposedly it can read your face and take your audio and then respond.

kevin_1_03-12-2024_095359: let's Hi. Oh, hi, Jack. How are you? Jack, I'm great, buddy. How are you? I gotta tell you, you've looked better. Okay, thanks for sharing. I'm Jack Nicklaus, coming to you circa 1978. And laptop goes in a river. I hate, [00:32:00] I hate that they have footage of me looking into my webcam just for that interaction, you know what I mean? That's,

Track 1: soul machines is an example of a company that you're going to see a lot of, I think, right? I think you're going to see a lot of these companies. And honestly, the idea here is not terrible. And I think that as we've said, this is the worst it's ever going to be. This feels like it's worse than I thought it was going to be.

Track 1: That's one of the weird things about it. But it just goes to show you that there's a lot of hype built into this idea. Like Kevin and I right now could probably off the shelf build what I believe would be a much better Marilyn Monroe than they built here. And this is, again, I know there's hardworking people in every business and no shade against these people really.

Track 1: But when you come out with a company like this and it's this experience, you're then turning people off of the idea of what's possible here in some ways, right? Like it is, it is really bad. And that just feels like a miss in a big way.

kevin_1_03-12-2024_095359: If this were, somebody on X posting their, their GitHub that they just uploaded, Hey, I made this thing. It's a digital avatar. You [00:33:00] can chat with it. There's some personalities. You go, okay. When a company comes out and says, Hey, we've got, , golf legends, MMA superstars, and Marilyn Monroe, we worked with estates to get the likenesses, to use them in promotion, to clone them.

kevin_1_03-12-2024_095359: I expect a level of fidelity, of realism. I expect the interaction to be better than what. I just saw in the browser and again not shading the people that are involved with it It's just you've got a press team that's making people aware of your product I expect a better step like it reminds me of when we cloned mr Beast and made him obsessed with bread In some ways that was a better

Track 1: that was, but that was better. Exactly. So anyway, just, you know, keep an eye out for this sort of stuff. Is this a weird idea period? Of course, but this is going to happen. And if you're going to do it, please just do it a little bit better.

kevin_1_03-12-2024_095359: Well that made us say Eiaiaiaiiiiiiiiii Which is, you know, something you don't necessarily wanna say when you're scrolling. But the things that make you stop, Gavin, the things that make you perk up, the things that make your [00:34:00] shoulders jut up to your ear lobes and say, 'I gotta put this in my newsletter, by the way, subscribe to the AI for Humans newsletter.

kevin_1_03-12-2024_095359: What are those things called, Gavin?

Track 1: I see what you did there.

Gavin: All right. So first up in AI, I see what you did there. We have this really cool post from one of our favorites, dreaming to loop, but this is an ex post where they are comparing SD three stable diffusion, three imagery with mid journey six.

Gavin: And you know, we talk about these two models a lot all the time. There are 15 comparisons they put in here. And what I see in this, which is really interesting is kind of what you see with cloud three and things like comparing cloud three to Google Gemini. Stable Diffusion 3 does seem like a step up. So one of my favorite things here is, especially in text, it seems to get text a lot better.

Gavin: The first, uh, very funny, uh, example here is, a group of avocados that are all that are all screaming. And in the background, it says GPUs go burr. And on [00:35:00] the stable diffusion clip, it's very cool. It looks like it could be an Estelle from an animated movie with a bunch of avocados. And then when you go to the mid journey post, it's good.

Gavin: And Kevin, you know, two months ago, we would have said, holy crap, look at this. But then if you look at the background,

Kevin: months later, Gavin, and

Gavin: two months later,

Kevin: avocado that looks like a pear, I'm seeing another avocado that's got veins on its nut, which, I, I don't like the sentence, but it is, that is a valid sentence, and it is

Gavin: cutting that out. We're cutting it. But I mean cutting that we're not cutting out of the show. We're going to make that a 

Kevin: There are so many errors in that image that two months ago we would have said is magical. And that's the point you're making, and I agree with you.

Gavin: So if you just go through this, there's a lot of really other good examples. There's a kangaroo that's wearing ski goggles. That's really fun. There's a hippo with

Kevin: Waffle Hippo

Gavin: back. Yeah,

Kevin: new benchmark. I love Waffle Hippo. But you're so right. Taking a step back for those that are new to AI or new to this podcast, stable diffusion is an image generation [00:36:00] software as is mid journey, you should go back and listen to older episodes where we dive into how these things work, but, , the masses do not have their hands on stable diffusion three just yet, but it is coming and those upgrades are really powerful.

Kevin: You will be able to just. Whisper into the computer what you want, and it seems like it's getting, , 98 percent of that stuff as output.

Gavin: Yeah. It's pretty incredible. . Go to X, follow DreamyTalupa. That is an amazing handle. They do really cool stuff. Kevin, I want to know what this next thing is because I actually, I'm not familiar with it, but when I

Kevin: wanna know who dreaming to lupa is.

Gavin: Me too!

Gavin: DreamyTalupa, get at us. We want to hear more about

Kevin: it's dreaming. Topa.

Gavin: Oh, well, Talupa, Tulpa, tomato,

Kevin: Go to go to snoozing, tilapia. And tell, tell. Snoozing, tilapia. We love their work.

Gavin: That is a great stable diffusion three prompt, by the way. Just snoozing, tilapia, I think a, a tilapia Yes.

Kevin: In a bed on a pillow. I wanna see the wrinkles in the pillow from the weight of the tilapia. Alright, [00:37:00] Gavin. W Wait. Tilapia. See this?

Gavin: Oh, no.

Kevin: wait till I figure, okay, nevermind, just listen to this. it is Korean, Gavin, I thought you might get that.

Gavin: Yep.

Kevin: Do you know what you're listening to, Gavin?

Gavin: Is that a robot? Is that why? Is that why we're listening to

Kevin: It is a robot. Do you know what kind of robot, Gavin?

Gavin: is it an eldercare robot?

Kevin: Oh, someone looked at the rundown! South Korea's local government has deployed around 7, 000 AI robot dolls to seniors and dementia patients. They're 1, 800 robot dolls by a company called Hyodol. And they can hold full conversations. They can, uh, tackle loneliness and remind users to take medication.

Kevin: These are words from, uh, Rowan Chung's X post about these dolls. And, look, the dolls are a little odd [00:38:00] looking. A little

Gavin: They're interesting. I mean, they're cute. They're

Kevin: a video of, like, an adorable woman having a conversation with it while it's on. Like, her lap and it's cheeks are glowing. But let me ask you this, Gavin.

Kevin: You descend down some rickety wooden steps into the basement of an Airbnb. There is a rusted red tricycle at the bottom of those steps. That doll, looking exactly the way it does, is sitting on it, hands on the handlebar, saying whatever it's saying. Is it still adorable now?

Gavin: I guess this is the beginning of Megan is what you're telling me, right? This is where the actual Megan begins with an old person doll. Any, anyway. Very cool. We do. We have said this before, but elder care is going to be a huge market for robots. Right. And because two things, one elderly people need a lot of work on terms and they need to have like things done for them.

Gavin: They need, you know, doors open. They need to,

Kevin: Reminders to take their medication, potentially, is

Gavin: honestly to be raised out of bed, to be lifted out of bed. Things like that are really important and [00:39:00] they're lonely a lot of the time. Like if you've been to a, yeah, if you've been to a retirement home, I remember when I was a kid, one of the things our school had us do was go visit retirement home and like listen to the stories of people, like write down their stories.

Gavin: And like, it was a really meaningful thing to me because he just shows you like, it is something where a lot of the older people in the world need more company. So this is actually, I think a good thing.

Kevin: . And we always talk about the robots that are bipedal, that have little grippers that'll move about. That could be a decade and change away before they're ever effective and cheap enough to be deployed into thousands of homes. But here in the doll form, it can again, remind them to take their medicine.

Kevin: It can be a health coach. It can do check ins. Uh, it can do voice memos and reminders. And one of the things that I thought was great was that it actually has a safety feature installed. If it doesn't detect any movement from a user in a certain amount of period, like If it doesn't detect any movement from the user in a certain period of time, it will actually call

Gavin: this is.

Kevin: that is making you giggle, Gavin!

Gavin: It's making me giggle because this is like the modern I've fallen and I can't get up button instead of it being you know [00:40:00] That that woman from the I've fallen and I can't get up. Yeah life alert. This is modern life alert It's modern life alert. Now. The robot will press the button for you Good.

Gavin: It's a good thing. I shouldn't be laughing. It's a very very good thing and congrats old people. We now have friends. We now have friends. Okay, let's move on. Let's move on before I get myself into more trouble with our senior citizen audience. I know there's quite a few of you

Kevin: All right, Gavin. Here, two for two today. Let me give you another audio test. Do you know what you're hearing?

Kevin: What's that? If you would have said, was that butter sizzling in a pan? Or was that, , lights flashing on a car as the alarm is set? You would say, yeah, of course that is!

Kevin: Why is that impressive? Because it's not real, Gavin!

Gavin: wow. It's AI. That's our AI sound effects. Essentially. We're looking at, and who's doing this, Kevin Pika labs.

Kevin: sound effects they're a video generation AI company. There's a lot competing in this space, but basically you type text into the computer and out comes video matching what you asked for. And if [00:41:00] you watch any of these videos, which we do all the time on this podcast, Gav, they're a little lifeless because usually beautiful imagery, mystically moving around, but there's no sound whatsoever.

Kevin: And that's a massive part of the experience. So as every company looks to diversify and pick their lanes, everybody's worried about Sora coming out, OpenAI's video generator. It's like, what are the other companies going to do? Well, here's a shot from the PikaLabs Canon. It is sound and it will automatically generate sound based off the content of the video for you, and it seems pretty impressive.

Gavin: I appreciate that Pica, which is a company that I think we've talked about this before. Certain companies are getting funded at a level that maybe isn't enough to compete with the big boys or big girls or everyone called the big companies. Uh, this company is pivoting fast, right? They're trying to do all sorts of different things.

Gavin: They're pushing out features really quickly. I do appreciate the fact that they're really moving fast on something that is not easy to do. And this could be super useful, it sounds like a very fun

Kevin: prompt to Hollywood in five years, we are still on that [00:42:00] trajectory, and this is just another example of that.

Gavin: That's exactly right. 

Kevin: So that might be the last. Hey, I see what you did there, but as far from the end of the show, Gavin, because each and every week you and I do dumb stuff with AI and I want to know. What dumb thing did you do with AI this week?

Gavin: mid journey just yesterday launched a very cool new part of their system called Google The character reference tag, and what this means is that for a long time with Midjourney, one of the biggest problems was consistency through iteration, being able to keep one character in images, so you could put that character into different situations.

Gavin: Well what just launched is the ability to do a dot dot dot. , CREF, C R E F, uh, function, and that allows you to keep one character throughout. I tried to create a Guy Fury. It was, it definitely was not Guy Fury. It gave me like, incredibly sculpted, Nordic, awesome chef man, is what I got.

Gavin: So I took that character, , and you can, what it says is it's better to do this with mid journey generated photos [00:43:00] first. I think it doesn't work with real people very well, and it doesn't work with, uh, Uploaded photos at least that's what they're saying. I have not had a chance to try it I'm sure we will see something of that in the future, but you can then take that photo You upload the url of it and you say dot dot c ref and I then said I said for the majority for the guy Fury guy said Man in scuba suit.

Gavin: Now, you can see the results here. I don't think any of those are really good scuba suits. But, it did kind of keep the same man's face throughout, right? And now, you can change the setting on C Ref. You can make it zero or you can make it a hundred. And if you make it, I think, a hundred, it keeps more. And zero, it keeps less.

Gavin: But it does show you that you can now generate a second image with a similar sort of character. The other thing I did is I took this weird robot that I had created from a previous image generation and I said robot on a 1970s talk show sitting across from a typical 1970s talk show host TV screenshot and it [00:44:00] gave me some pretty good references to it.

Gavin: And then I took that same robot and I said in Niji which is their anime model and I said a happy robot in a beautiful green field sunflowers bright bright blue sky And you can see it's a similar sort of robot now It's not the exact same robot But it understands enough to take like the style of the robot and transfer it over Which is pretty cool.

Kevin: kind of chromatic helmet going on. It got the thinner neck and build. I would expect Midjourney to be able to render scuba gear better than we're seeing in that image. But it seems like because it is so concerned about keeping that consistent character, it might have thrown a little bit of that to the side.

Gavin: And I will say like the mid journey as it like stable diffusion in most image generating software is very prompt dependent. And if you look online, we'll show a couple examples in the video here of people online who have done this significantly better than me. I think there are people who are experts at this.

Gavin: But it is a real big step forward in terms of trying to make [00:45:00] larger projects with something like

Kevin: Yeah, yeah, that's awesome. I also want to shout out Photomaker, which is an app that I think we discussed a few shows back. But that's also something that you can run locally on your computer for free, and you can take a photo of yourself, or drag any artwork that you've made, and generate consistent people, , which is really nice.

Kevin: , Gavin, this week, I messed with something a little silly. I messed with TRIPO. S R T R I P O S R. Oh, little TRIPO! And what TRIPO does is allows you to take a single, two dimensional image, it figures out what the subject is, gets rid of the background, and then tries to turn it into a three dimensional object. All within one click, and even on the unoptimized Mac version that I'm running right now, Gavin, the generations take about Five seconds?

Gavin: Wow. Really?

Kevin: On a, on a well equipped PC, it's like one second. It is so fast, and if you look at some of these obj files, these are the object files that I extracted, [00:46:00] and, you can just click and drag around the mouse, you can see that I made , a weird bear plushie cause it popped up when I searched for cute characters.

Kevin: I took the little AI for Humans robot head thing that you and I have been toying with and turned it into a A kind of lopsided, wobbly pendant, but it's still

Gavin: Yeah. That's so cool.

Kevin: And then there was a Guy Fieri bobblehead that came out incredibly well, considering it was a single 2D image, and it figured out, okay, here's the geometry of the head, the glasses, his, , soul patch juts out in a way that seems a little,

Gavin: A little much,

Kevin: a little much is a little

Gavin: little aggressive.

Kevin: Yeah, it's a little aggressive, but still it got it, and it has, like, His hand outstretched from his body, his arm holding what I think is like a burger. But again, this was five seconds and one still image generated into a 3D object for free.

Gavin: I feel like we've been tracking this space in real time over the last six weeks or so in this show. And as you know, [00:47:00] every week we're like, Oh, this is cool. It gets better and better. And like, we're just getting closer and closer to this idea. And the larger picture of all these is always. What does this mean in terms of what can go into a much broader, much bigger environment?

Gavin: Because picture to 3d is not only about this fun little tool that you can do stuff with. And of course, yes, you could make something physical out of that. But what it's really about is virtual worlds to write the idea that you could create an image and then suddenly turn that into a 3d model that can then go populate a virtual world.

Kevin: We've already seen the, the 2D animation tools where it can extract motion from somebody from like a video and apply it to a character. We'll We're seeing that now with, you know, in our newsletter, we had a 3d physics system based off AI. So imagine you could take a photo of your friend.

Kevin: Like you said, now they're in your Roblox, Minecraft, Fortnite environment, whatever. And now you can say, make them run, walk, jump, swim, do all of the things. And it goes, okay. Yeah, I can

Gavin: Make them do my bidding. They're now [00:48:00] my, my friend is now my.

Kevin: I'm sure you saw Gavin. obj, which is in the folder, and that

Gavin: I did. I

Kevin: bust of you, Gavin, that I can rotate all around, and it, it's, again, it's getting there.

Kevin: You can go play with it yourself. Again, it's called TripoSR. 

Gavin: We got to move on to our interview today. This is very exciting. In front of the show, Alex Katran. Alex is a specialist in AI education. He runs a company called AIEDU. It is a non profit and their goal is to bring AI education to really all students everywhere, Starting with school districts.

Gavin: So they really work with school districts and do a lot of work and see how they can bring AI education to everybody. Um, here we go. Here's Alex.. 

Gavin: Welcome, Alex, to our show. We're very excited to have you here. And before we jump in, we are going to ask you the question we ask all of our guests, which is a little off putting, yes, but it still gives us an insight into the person. Alex, from a, on a scale of one to 100, what is the percentage chance that [00:49:00] you think AI kills us all, all humans? First, hello

Alex: 100. Yeah, that's quite a way to start. Um,

Kevin: Could ask, how's your day going? You know, what's up with you and AI, any good Netflix? But no, do you think humans will meet their end due to the maker that is artificial intelligence?

Alex: So my entree to artificial intelligence was around the year 2017 and I watched a Google talk by Nick Bostrom and

Alex: you Super intelligence. Yep. You're the author of super intelligence.

Alex: and Nick was really like the first person, at least sort of to, to bring this conversation about existential risk and Vogue.

Alex: And so, um, You know, that was what I think began my journey down the rabbit hole. I, I, as I'll get into, you know, AIEDU is actually a rejection of that obsession, right? Because I felt like there was too much focus on the existential risk and not enough focus on, you know, all the other kind of like softer risks that were still quite significant and worth paying attention [00:50:00] to.

Alex: And just for your listeners, you know, Nick Bostrom, you know, basically his thesis is that, uh, you know, at a certain point, AI will get so powerful, uh, that it will just exceed our ability to restrain it. And basically rejecting his, his thesis requires either saying that, well, we'll, oh, we'll be able to figure out some way to, uh, restrain it through, some sort of like moral compass that we sort of inject into, into the AI.

Alex: , or I think if you really were to reject Nick Bostrom's thesis, you might say, well, AI will never get that good. We're going to hit some sort of like, you know, roadblocks or barriers, either technologically or, , algorithmically, , you know, we don't seem to be hitting many barriers. If anything, I would say progress is moving faster than ever before.

I would say it's, I would say it's, uh, you know, at least 50, 50, if we include humans,

Alex: uh, misusing AI, 

Gavin: 50 50s a lot though. That's quite a

Alex: What was the time frame?

Gavin: No zero time frame

Alex: Oh, yeah, well, definitely 50 50. I mean, now, in my lifetime, I would put it a lot lower.

Alex: Maybe 10 percent or something like [00:51:00] that.

Gavin: that's fair. That's still not nothing though, for

Kevin: So we're kicking the can so that the grandchildren can fear the Skynet that we are gleefully brewing right now.

Alex: Well, and, you know, I mean, I think Once you, sort of, Project out far enough. You know, we asked the question of like, what will humanity even look like? Are we all going to be sort of uploaded? And, and will, will it truly be like, uh, a bad thing if humanity was sort of just like in imbibed by the AI and sort of

Gavin: The singularity, the singularity is near. ACtually, this is a good transition into what you do, because I think this is a really interesting thing. And for our listeners, you know, we've talked a lot about the idea of AI education , and what we can do on this podcast to kind of bring an awareness of how this is going to change the world of stuff.

Gavin: Just kind of tell us what AI edu is and kind of the history of it a little bit, like what is the company itself? Mm

Alex: yeah. So, uh, the AI education project is a nonprofit, uh, I founded it in 2019. Um, and yeah, our mission has stayed the same. And I think [00:52:00] actually part of the reason we've been so successful over the past. Uh, let's say 18 months is because we've been really focused on this question of, you know, every single student, we believe, uh, needs to graduate high school with some fundamental skills and knowledge to thrive in the age of artificial intelligence.

Alex: And, for me, this was less about how do we restrain artificial general intelligence and more about, you know, how do we make sure that people aren't sort of displaced and mass, um, as AI starts to be utilized in different careers. For us, when we talk about AI literacy, that's really our focus, it's through this paradigm of, this is not like some niche, uh, or a track that students need to be on, this is sort of as fundamental as the internet.

Alex: And so in the same way that you don't have a class called Introduction to the Internet, we think the future of AI literacy is more like, um, integrating AI learning experiences across every subject. So, you know, learning about the intersections of AI and. You know, politics and history and the humanities and science and biology and mathematics, so our focus from the very beginning was [00:53:00] how do we partner with underserved schools that don't necessarily have, you know, robotics programs and computer science teachers?

What are you seeing, boots on the ground out there? What is the temperature towards artificial intelligence? What's the level of literacy and understanding with both the educational system itself and then the individual parents or the students?

Kevin: Are they fearful? Are they excited? Are they resistant and reluctant to change? 

Alex: I think there's definitely more awareness than there was five years ago. , and being able to get hands on as a game changer, right? Like it, it, it makes something tangible that previously was very esoteric, you know, like really understanding machine learning and it actually intersected with people's lives quite a lot, even before chat JBT, but it was less obvious.

Alex: Um, so I would say there's, the people are at least aware of this as a conversation. Now, what I would say is that most folks, you know, kids, parents, teachers. are, are still honing in on this is a cool tool. This is kind of cool for me to play with. How can I use it [00:54:00] for whatever it is I want to use it for?

Alex: Um, I don't know that there's, there is as much of an appreciation for like how significant the impact that this is going to have, that, that generative AI large language models will have on knowledge work. And

Gavin: so my daughter's in high school. She has an English teacher that is very pro AI in that he's integrating it into the English class that he's giving essentially, and have encouraging them to use it. What percentage of teachers would you say, see this as like a pro versus something that they're really worried that this is going to affect education in a bad way?

Gavin: Like, I think it's, I think the other side of this is I, I will, I have another Another child that is used it in a way that I don't know if a teacher would appreciate how they use it. Sorry to my other child for outing you, but there are options and things that are being used to do things with AI that, that may be end arounding the idea of teacher.

Gavin: So in your work with teachers, like where do they land on it?

Alex: yeah, so we actually just did some research, [00:55:00] um, that OpenAI helped to fund. Um, and we surveyed a thousand teachers and administrators, and we were trying to actually answer this question. It's like about half to a little bit more than half are, you know, somewhat fearful about the role of AI, um, and about half are somewhat hopeful.

Alex: Um, and so I think there

Gavin: that's super interesting. Cause in some ways it means that like you, that's a real split, right? Like it's, it is a real diversion and it's kind of like what teacher you get will be that version of whether or not they're going to like lean in or not in some form.

Alex: Yes. And, and what's important is like the polls are like pretty low. So like the, the percentage of teachers who are extremely fearful is like 11%. The percentage of teachers who are extremely hopeful is like 5%. Um, so most teachers are somewhat fearful, somewhat hopeful. To me, that really means like they haven't quite made a conclusion yet.

Alex: You know, I think like intuitively there's, there's some challenges that they're going to have to deal with. Um, and, and that is really the risk that like what you just named is like, you know, whether or not, I mean, the analogy that I use is like, okay. A, you know, if computers are a bicycle for the [00:56:00] mind. Um, I've heard some people say, you know, language model is like a, you know, an electric bike.

Alex: I would call it more like a supercar. Um, you know, whether or not you know how to drive a super, you know, that Corvette is the most important thing to your ability to compete in the job market. Um, because you can train on the bicycle as much as you want, but you know, it's just that you're going to be sort of outmatched.

Alex: So making sure that kids have this opportunity to like build these skills and like get hands on is going to be critical. Um, and right now that is largely being driven by just like, which teachers are kind of leaning in and like finding ways to integrate it. Um, because we don't have district wide policies, you know, or consistent district wide policies.

Alex: We certainly don't have state, you know, clear state guidance. I mean, the guidelines that have been published now, and we actually helped to write some in Ohio are like, you know, they're relatively high level in part because the technology is moving so fast. It's kind of hard to, to really put your finger on what the like safeguards even need to be.

Alex: Um, so. Yeah, I mean, I think it's interesting that your daughter is, um, you know, has this English teacher who, you know, I think it really can cut both ways [00:57:00] because it dramatically changes what it's like to be an English teacher in the sense that you can't, like, the age of the take home essay is basically done.

Gavin: Yeah. Yeah. Well that's, I think there's gonna be a whole like oral argument phase of education comeback, right? This idea that like you would present an argument in class, in some form or another to prove that you kind of understood the material and you weren't just regurgitating something that an AI gave you.

Alex: Yes. I mean, that's, that's literally what, I mean, we worked with a lot of teachers who have been successful at implementing AI, um, into the classroom. And that's really what it looks like is a combination of, um, you know, teachers are spending less time doing like written feedback, you know? So like, for example, a teacher who used ChatGBT to, um, to support sort of engagement in the classroom.

Alex: So what that looked like is. Rather than sit down and write, you know, five or six sentences of feedback in response to a student paper, he, he plugged the papers into ChatGPT, got, it spit out sort of some initial feedback. Then he went to the kids and said, look, ChatGPT wrote this initial feedback, you know, take it with a grain of salt.

Alex: And [00:58:00] if you think something is wrong, like, let's talk about it. But rather than sit down and spend time by myself writing feedback, I'm going to use that time to meet with you one on one. And I think at the end of the day, if you really, you know, if you're Um, you don't have to know anything about AI to be able to identify whether a classroom is ready or is preparing students for the age of AI because it's like, are the students, as you said, are they engaging with each other?

Alex: Are they building communication skills and presentation skills? And are they problem solving? Are they collaborating? Um, and so I think what, what your daughter's classroom is, is pi, pi, pi, pioneering, I guess, is, um, hopefully something that's going to happen, um, You know, at large scale. And we're really focused on figuring out how do we do that in a way that's equitable because, you know, the early signals are pretty disturbing.

Alex: It's like, you know, there are private schools now that have prompt engineering

Alex: classes. Right. And it's like completely unsurprising, but it's also because there's like, you know, hard costs associated with just these models and the usage, um, and there are no budget lines yet. So [00:59:00] schools haven't quite wrapped their head around like, you know, You know, and this is the challenge with these policies is like, oh, well, schools will say, well, we believe in equal access to large language models for students.

Alex: And it's like, well, do you, do you even know what the cost is associated with having every student get equal access? And presumably that means they all have access to the same model, because if some kids have GPT 4 and others have 3. 5, that's not, that's not equal at all. Um,

Gavin: a computer camp, and this is, I'm old, right? So in like sixth grade, I went to a computer camp at a school called Lakeside. And if you, if you don't know what Lakeside is, Lakeside is one of the biggest private schools in Seattle, but it's where Bill Gates went.

Gavin: And the reason why they had a computer camp, it Lakeside, is because Bill Gates and the 

Kevin: hand donate everything? And

Gavin: Gave a lot of money to them. Right. So I was able to go to a computer camp and learn how to program logo, which is a shout out to logo, the little turtle moving language, but like, that's a really big problem and I think it's a really important thing to kind of come across to

Kevin: Alex, when do you throw kids into the supercar and what does that look like? [01:00:00] Because this is a massive issue, which you have thankfully chosen to tackle. , it probably has to start pretty early. , is this a preschooler should have a dedicated LLM and learn to start speaking with machines because that's how the skill will translate and blossom?

Alex: yeah, that is a really, really good question. So I'm going to respond with an anecdote. So we, I was talking to one of the teachers that had, I think it was literally like the first chat GPT workshop that we had done in like January of 23. And it was like six months later and we were just catching up and I was just trying to get a sense of like, okay, it's been six months.

Alex: How have you been using chat GPT in the classroom? You know, uh, her name's Felika Ross. She's at Akron Public Schools, actually. So, and Felika's like, yeah, you know, when I first started using it, it was amazing. Like, the level of engagement that I saw from my kids was, you know, higher than almost any other project that we had.

Alex: And she had designed a couple different projects. I think one of them was like, uh, I forget if it was like a podcast interview or something. And they also had like a CSI where they were trying to identify the deepfake. Um, [01:01:00] yeah, no, it was, I mean, and, and so very, very promising. And then she said, and I was like, well, so how have you been using it recently?

Alex: Cause that was like a couple of months ago. And she said, oh, I stopped using it. Like, why did you stop using Chachi B'tee? She's like, well, you know, I said, I'm an English language arts teacher. And we got to the point, you know, I teach freshmen, uh, ELA. And so we got to the point in my syllabus where I'm teaching students where they're learning how to create an outline.

Alex: And she said, what I realized is that they weren't actually learning how to create an outline. They were learning how to ask for an outline. And so you can, you know, like certainly the skills, like the computational thinking involved in, in prompt engineering for an outline is valid, is valid, but I think there is this question of like being able to structure your thoughts and, and write them down.

Alex: Is that also important? And, um, I think that is, that's an, that's an anecdote that is representative of this like much bigger challenge, which is, I think it's folly to, to think that we're not, we're going to be able to find a way to avoid these tools being crutches. Like they will be a [01:02:00] crutch. It's just, they're so easy to use.

Alex: And I've heard stories, Gavin, this wasn't you, but I feel like it was someone that we know, a mutual friend of ours. And basically he was like, yeah, my, my nine year old daughter started using ChatGBT and now she doesn't use Google anymore. She just uses ChatGBT because it's so much more interesting. And to me, that's actually problematic because ChatGBT is useful for a lot of things, but it's certainly not a replacement for Google or for search in general, um, and.

Alex: You know, once you start, you know, once you get used to using a tool, it's very hard to kind of like roll back. So this question of like, how early can we, can we introduce them? Um, we're going to have to wait to see some research, right? Like we need, there is very little research or no research that I'm aware of that really answers that question.

Alex: I'm sure there's a bunch that's in process. And so in the coming one to three years, hopefully we'll get some like more clear guidance. And there's also, you know, I mean, Conmigo's, you know, I mean, I, I assume your listeners are familiar with Conmingo, but for those who aren't, it's basically, um, an LLM tool that is designed to [01:03:00] support students, almost like as a tutor.

Alex: Um, but what's really special about Conmingo is it doesn't just give you the answer. So you say, hey, I need a, I need an outline for this, you know, essay that I'm writing in history. It's not just going to give you the outline. It's going to say, okay, well, you know, let's, let's walk through step by step what's involved in writing an outline.

Alex: And so it kind of like plays that role of, you know, a helpful tutor or mentor that doesn't just give you the answer right away.

Alex: Um, a, is that publicly available or is it something you can only get through so, oh yeah, what, what's the, what the URL of it is? Like, cause I actually never heard of that

Alex: Oh, it's, um, so Khan Academy, uh, built it

Gavin: It's Khan Academy. Yeah. Okay. That makes sense. That's a really cool idea because that way you're, you're creating a custom LLM that's designed to both teach and provide, uh, guidance, right?

Gavin: And that's like, it does feel like these LLM things is like when you get to that point about, you know, it can't replace Google. We just had this whole thing of. With the Gemini controversy, how when you expect, uh, an LM to be an answer box for everybody, it's just never going to succeed, right? It can't be a one answer box.

Gavin: So that's where things like perplexity are kind of interesting. You've got this kind of combination of LLM plus results in some form, [01:04:00] but specific designed LLMs also is really interesting, like kind of specifically guided into one thing. So it's called con con Mego is what it's called. Is that right?

Alex: it's called Khan Amigo. 

Alex: It's like 30 A year. That's right. Um, and there's other companies that are doing this. Um, and one of the school districts that we work with, uh, Anaheim Union Public Schools has been piloting some of these tools. They're working with UC Irvine to actually do some research. Um, that research isn't done yet, but my understanding is that some of the early data is really promising, right?

Alex: Because the model of this is look, looks something like instead of having a teacher stand in front of the classroom and write a bunch of equations on the board and, you know, like half the kids are kind of just like totally checked out. They're either on their phones. The other half of the kids aren't really following it.

Alex: And if someone has a question, the whole class is now interrupted,

Kevin: the whole class has that question, 

Alex: yeah, whereas now what they do is that the teacher will. Have kids kind of break out individually with their AI tutor, work on some problems together with the tutor. [01:05:00] The kids are getting that personalized feedback. So depending on how fast or slowly they're sort of picking it up, they're getting kind of like that, that support.

Alex: And then the teacher brings the kids together, breaks them into groups and have them, has them work on new problems together.

Gavin: Hmm.

Alex: And, and that model apparently like dramatically increased engagement. And specifically this story, this is a story from a math teacher I was talking to. What he saw is. The, the kids who were the, like the worst performing, the most behind were the kids who saw the biggest, where he saw the biggest increase in engagement.

Gavin: wow. That's really interesting.

Alex: There's, there's reason to be like concerned about like what, what will happen if these tools sort of just become so ubiquitous that, you know, we kind of lose, lose control over, like, we, we no longer get to answer the question of like, well, should we put this in front of kids?

Alex: Because it's just going to be there. Yeah. Um, so that is, that is trouble. Yeah. It's like, you know, exactly. We saw this with social media. It was, we kind of just like felt our way through it. And there was a lot of collateral damage as we, and I think we sort of are continuing to feel our [01:06:00] way through it. There's reason to be really excited though about, especially if you're, you know, cause you, you went to private school, if you had two parents who work in tech or have, you know, master's degrees, you have parent GPT when you go home, you know, like they can help you with their homework, the kids who really benefit from

Alex: something like 

Alex: Conmigo are kids 

Kevin: was terrible in my household, though. It was like, it was GPT 2. 1, it would sort of give you a half response and then be just disappointed with you and tell you to go outside.

Gavin: My dad GPT was much worse because it would give you a six page response and you'd have to figure out what the damn answer was in that six pages that that was everyone everyone's dad GPT is different I'm

Kevin: That's right.

Alex: Yeah, my mom GPT, you know, she's a math teacher, so she got very excited when I had math problems that I needed help with. But sometimes it was like I was signing up for a whole nother math class, and I was just like, I

Gavin: oh yeah

Alex: with that one question. And now it's like 10 p. m. And she's like drawing stuff on the chalkboard.

Gavin: ,

Gavin: Alex well here's a question for you say you're a parent at home listening to this and you have a kid and . Your kid's school is not leaning into this [01:07:00] sort of thing.

Gavin: Do you think it's, they should like advocate for it? Or do you think this is going to come to everybody? Like, what is your experience about how this is spreading throughout like all schools, basically? 

Alex: I think that there's always value in advocacy and in, , You know, trying to shift systems. And I think that parents have a really important role there. So I would certainly encourage parents who see opportunities to help their schools kind of like be more on sort of forward footing when it comes to emerging technologies.

Alex: That's good. You know, at the end of the day, though, like, you know, I think this is one of the most important conversations that parents can have with their kids. It's like, you know, open AI and pendants of research. They were looking at. Okay, what are the career? What are the jobs that are most exposed? To large language models, assuming like, I think a straight line progression of LLMs in terms of quality over the next five years.

Alex: And, you know, the careers that have like, so a hundred percent exposure, which means 50 percent of the tasks, you know, can be basically performed by an LLM today. And these are like business analysts, you know, mathematicians, data scientists, you [01:08:00] know, writers, uh, designers. Um, so these, I mean,

Gavin: Good jobs. Those are good jobs.

Alex: they're good jobs and they're also not niche.

Alex: I mean, these are like basically the bedrock of knowledge work, you know, like everybody in business school, like what do you do your first year out of business school? You're like doing financial analysis, you're working with spreadsheets, you're maybe writing a memo to a client or working on a PowerPoint.

Alex: , so I think that what really concerns me is like everybody's sort of focusing on the widgets and like the cool LLMs that are legitimately amazing. Right. But But the much bigger challenge is like less about how are these tools going to be used in education and more about how is education going to have to transform to meet the needs of what is probably going to be a rapidly shifting workforce because it, you know, automation doesn't happen sort of gradually in a straight line fashion, like capital expenditure in, in automation is basically like 80 or 90 percent of it happens in the 12 months following a recession.

Kevin: 

Kevin: Well, what if, according to my YouTube algorithm, we're in it, [01:09:00] and it's been bursting, and it's not a soft landing, and I need to buy Italian real estate for 1. What then? Mm hmm.

Alex: if we're already in a recession, well, I mean, you know, the companies are already starting to roll this stuff out. I mean, there's a clarinet laid off 700 people. Uh, Jamie diamond says they were able to reduce the, um, The time they were able to, uh, reduce the time required for, I think, cashflow management for clients by 90%.

Alex: And

Alex: so 

Gavin: just had a story today that 30, uh, the EA, the video game company says they can make themselves 30 percent more efficient. I think the interesting thing is I don't think we're going to see a recession. And this is where it's going to get weird. The economy is not going to be in recession, but people's are going to lose jobs and that's where the divide is going to feel strange and where I think to your point, uh, Alex educating people, especially when they're young.

Gavin: Giving them an awareness that like this stuff is going to be integrated into everything. You become a much more, I don't want to say valuable employee cause that seems so like a totalitarian in some ways, but you do, you become a much more valuable part of society. If you [01:10:00] understand this plus your thing, right.

Gavin: In some form or another. And I feel like that is the future going forward. It's like, how can we kind of bring these two things together?

Alex: yeah, but I mean, I think back to this parents, right? So like I, this actually happens a lot where I'll be, I'll be talking to funders or school administrators and they'll actually interrupt me in the middle of sort of like my, like my pitch. And they'll be like, wait a second. I, I have a 12 year old son who wants to go to law school or, you know, my daughter wants to go into graphic design.

Alex: Like. What should I tell them? And, you know, a, that's the right question to ask. Um, and you know, I think it's, I think it's something like the traditional career pathways that we used to assume guaranteed you a prosperous, uh, you know, like outcome are going to narrow in many cases, right? So like, yes, there is that all lawyers are going to disappear, but if there are 10 percent fewer lawyers employed, Then going to a tier three law school might not actually make sense if you're going to take on 200, 000 of debt.

Alex: So like, it's more nuanced conversations like that. [01:11:00] And like, ultimately, I think the advice that if I was a parent, I'm not. But if I was a parent, what I would say is, it kind of doesn't matter what your interests are anymore. Like, what you really need to get really good at is like, being able to get a job and being able to upskill.

Alex: So whatever your interests are, figure out how you can use AI to, you know, Do stuff with that. So if you like shoes, build, you know, launch a sneaker company and like use generative AI to create some like social media ads, or if you are interested in sports, you know, there's like lots of ways that you can actually, you know, leverage kind of like the skill sets that you're going to need in the job market.

Kevin: I want to round this one out because we've had concerns on behalf of the students, and the parents, and the corporations, and the podcast hosts. but we do have a special guest today, Alex,

Gavin: really? Is, are we doing

Kevin: well, I, I saw an eyebrow raise and, uh, ears perked up. This is Ollie. This is Gavin's dog, who would like to ask you one quick question and close this out, if you'll indulge.

Alex: sure.

Kevin: Sniff, sniff. All right, Alex, I've got one burning question for you. Sniff, sniff. [01:12:00] Rough. Rough. Okay. Yeah. And he says that a lot. How are you gonna make sure that AI and education doesn't replace the most important learning tool for US dogs? The trusty old tennis ball. Woof. Okay. Well, because let me tell you, no fancy algorithm can ever match the sheer joy and motivation that comes from chasing after that fuzzy little orb of happiness. Sniff, sniff. Okay,

Gavin: Okay. Ollie. I didn't know if I Ollie, was he still going? Is Ollie still going?

Kevin: told Ollie to go outside, uh, I regret that and I'm sorry, and I may very well just yoink that right out of the episode, but that's what I wanted to ask, Alex.

Gavin: do you have anything you have any thoughts on that? 

Alex: that? Okay. Well,

Kevin: You do not have to answer that,

Alex: I'll

Alex: try, you know what I'm going to take I, I'm going to take a, I'm going to take a stab at this. , Hell, I mean, in a, in a world where we're spending even more time in front of screens and interacting with weird AI chatbots, like Oli, play is probably going to be super, like super, super important.

Alex: So that's another thing. If you have kids, like get them away from their screens because. You [01:13:00] know, the skills that they're going to need are probably something like, how do you work, how do you just like interact with people? 

Gavin: It definitely is weird.

Gavin: Alex, thank you so much for being here. Where can we go find your work? And where can people who are interested in learning more about AI in schools? Where can they go to learn more?

Alex: Yeah, go to our website, aiedu. org, um, and we have free, all of our curriculum is free, available for download. And if you're, especially if you're a school leader or a teacher, reach out to us. We'd love to chat and figure out if there's a way for us to bring AI to your school.. 

Gavin: You so much to Alex, , Katran for coming on the show. We love, , AIDU, so go check it out. All right. We have our five star reviews for the week. If you send us a five star review, we will read it on the show. The first one is from Huggy the Chef. It topic line says been listening and watching since I met Gavin at OPJ event last year.

Gavin: Shout out Carly Riley. Shout out overpriced JPEGs. So great guys. Always funny, entertaining in my head with some of my friends. Another habit is now using your question that Kevin doesn't like so much anymore.

Gavin: And I get lots of interesting [01:14:00] conversations. Here's last night's on the morbid side, It was about AI is the demise of civilization continuing. We already don't know how to do many things anymore with generative AI and coming AGI. We will have to know how to do less and less creativity will be squashed.

Gavin: Our intelligence will wane while in the meantime, we will find a cure for cancer, solve world hunger and explore the planets followed by the stars. What do you think you should both be answering this question on a regular basis? , Kevin, what do you think? I think we can do quick answers on this. Cause I have a thought.

Kevin: Yeah, well, I mean, I think the answer contradicts itself, oh, creativity's gonna wane, humans are gonna do less and learn less on this path to exploring the cosmos and curing cancer and solving world hunger. They're in contradiction with each other, ?

Kevin: Because if we're going to do nothing and get nothing and have nothing, well then how are we going to push these boundaries? It's not like once we solve cancer and explore the cosmos, we'll be out of things to do. I think humans will still find some reason to have idle time and fill it with something exciting.[01:15:00] 

Gavin: I'm gonna say something that's gonna probably weird a lot of people out, but I really do believe this is gonna happen. I think we are going to eventually merge with

Kevin: Absolutely.

Gavin: and I think that that's, that's probably not as far away as people think, and there will be an organic human life world that will probably exist outside of that, but if you are one of the people that choose to merge, and who the hell knows if this is gonna be a good thing or not, The opportunities that will open up to you will be incredible from a creative all these standpoints, because you have to imagine people always talking about this, taking away creative jobs.

Gavin: Well, imagine if your creativity was infused with the things that you're seeing coming out of the AI that feels like where we're going to eventually end up. And I personally, I think I'm kind of interested and excited

Kevin: You and I, we're Team Synth. Let's bring it. And these analogs? They're going down. Have fun with your wood planks and your coconuts with painted faces. We're gonna be hanging out with our telekinetic powers.

Kevin: That you can root us and then take us over with an attack. Ah, never mind. Let's move on. [01:16:00] Another five star review, Gavin, by HeyDrew. The best and wittiest way to stay informed about AI advancements. Now that, I really appreciate as a subject. The body is, if you are tired of the same old AI discussions and jokes about how it will kill all the humans.

Kevin: And you want to expand your knowledge on real AI advancements and breaking news, this is the podcast for you. Gavin is excellent. Alright. And fans of G4TV already know Kevin's wit and charm well. And it's on full display here, huh? I was trying to meter it. I wasn't trying to give you full wit and charm, but there you go.

Kevin: Also, AI is probably going to kill all the humans.

Gavin: there you go.

Kevin: hey Drew.

Gavin: There you go. Hey, drew. We love it. All right. This last one is really important, Kevin, and I want you to listen up.

Kevin: I only read the

Gavin: I want you to listen to this one very closely because this is probably the best five star review we've

Kevin: just saw this. You're going to make me sit through this?

Gavin: Okay, the title is unveiling ai colon insights humor and more this is from mcfizzle1997

Kevin: Yeah, thank you, McFizzle97. Ha

Gavin: Gavin [01:17:00] Purcell has done it again! After single handedly taking the floundering tech TV into the tech juggernaut G4TV and making Fallon a star, Gavin has taken his golden touch to the podcast world. Gavin's experience translates into an engaging and humorous hosting style that makes complex topics and thoughtful discussion relatable to the masses.

Gavin: Blah blah blah, this is very nice and I appreciate it. This, this, thank you McFizzle, this review definitely

Kevin: No, Gavin is

Gavin: was starting

Kevin: the Ying to his Yang, new AI co hosts every week. To Matrix Dodge my involvement that heavy and hard, McFizzle97, how dare you?

Gavin: What if this is the spouse of the person that wrote the one last week and we've got some weird back and forth going on in our podcast reviews. But anyway, thank you so much McFizzle. I'm not gonna read the whole thing because it is a very flattering picture of me and maybe just leaves Kevin out. So we'll leave it there.

Gavin: But for

Kevin: they would be remiss if they didn't point out you can [01:18:00] catch Gavin's handsome, boyish charm finally on display in front of the camera instead of behind it. Thank you so much, McFizzle, and thank you to everybody who left a review. Hey, sincerely folks, each and every week we ask this, but it does actually matter for us.

Kevin: Your impact , is immense. If you take the time to subscribe, to click a thumbs up, smack a bell, so you get notifications when things go live, even if you're on Spotify or whatever your podcasting app is leaving a positive review or five stars there. It helps us out. And then of course.

Kevin: Thanks. telling people that is the way that this podcast grows and stays alive.

Kevin: So thank you to everybody for sharing the word about our cast. If you enjoy it, that's the way you can help us out. Just make other people aware that we exist.

Gavin: Thank you so much. And next week, we're going to be coming to you from New York. It's going to be Kevin and I in a hotel room in New York, because we have a fun event that we're attending as the AI for humans crew. So we will talk a little bit more about that, but we will see you all next time. Bye bye everybody.