Digital Nexus

5 | AI will fight dance you!

Chris Sinclair and Mark Monfort Season 1 Episode 5

🎙️ Episode 5: "AI Can Fight Dance You!" 🚀

In this lively episode of "Digital Nexus," Mark and Chris dive into the fascinating world of AI's potential futures. From Ethan Mollick's intriguing ideas about the future of AI to the industry's wild west atmosphere, our hosts explore the debates surrounding the journey to Artificial General Intelligence (AGI) and whether current AI models have hit their peak. They emphasize the need for organisations to buckle up and prepare for the unpredictable twists and turns of AI advancements, all while having a bit of fun with the practical—and sometimes overhyped—realities of AI.

Things get juicier as they chat about Meta's big play with the open-source Llama 3.1, debating the future of open vs. closed-source AI. Mark and Chris also tackle the latest AI mashups, like AI break-dancing and sports betting, showcasing how AI is creatively shaking things up across industries. And just when you think it can't get any better, they highlight exciting news from the healthcare front, where AI is making strides in early disease detection. Tune in for a fun-filled exploration of AI's wild and wonderful future!

 #DigitalNexus #Podcast #AI #Dance #CombatSports #TechInnovation #MachineLearning #Creativity #Robotics 

Other Links
🎙️our podcast links here: https://digitalnexuspodcast.com/
👤Chris on LinkedIn - https://www.linkedin.com/in/pcsinclair/
👤Mark on LinkedIn - https://www.linkedin.com/in/markmonfort/
👤 Mark on Twitter - https://twitter.com/captdefi

SHOWNOTE LINKS
🔗 SIKE - https://sike.ai/
🌐Digital Village - https://digitalvillage.network/
🌐NotCentralised - https://www.notcentralised.com/

YouTube Channel: https://www.youtube.com/@DigitalNexusPodcast
X (twitter): @DigitalNexus

<b>[Music]</b><b>Yeah, don't quit out there, Josh.</b><b>Ah, welcome back.</b><b>We're here for episode five.</b><b>I can't believe we've</b><b>already done five episodes.</b><b>Where's the time gone?</b><b>It's disappeared.</b><b>That's a month worth of us</b><b>chatting with each other.</b><b>It's the end of July.</b><b>I see you more than my wife and kids now.</b><b>I see you more than</b><b>my wife and your kids.</b><b>My wife and my kids.</b><b>You're all hoping my kids.</b><b>I hope so.</b><b>Your wife and my kids is worse.</b><b>No, no, no.</b><b>Shout out to you fam.</b><b>They're awesome.</b><b>We love those guys.</b><b>How?</b><b>We'll have to have them</b><b>on the show at some point.</b><b>Well, one day.</b><b>Just little Lana.</b><b>Yeah.</b><b>Just a chat.</b><b>Lana's Lana's corner and she</b><b>can share her opinions on AI.</b><b>I'll just get AI to do it.</b><b>Oh, we could do 11 labs.</b><b>Like we did AI Mark and Chris.</b><b>They're not making an</b><b>appearance for this one.</b><b>We didn't prepare for that one.</b><b>But yeah, there's a lot</b><b>more people I'm finding out</b><b>that are using 11 labs actually.</b><b>Oh, really?</b><b>Yeah, it's fantastic.</b><b>There's folks that I</b><b>was speaking to yesterday</b><b>and they use like the digital avatars</b><b>and they have like a clone</b><b>of their, you know, them.</b><b>So speaking and stuff, but the voice is</b><b>actually through 11 labs.</b><b>So it's this combination of things.</b><b>It's the AI Marker place,</b><b>the platforms out there.</b><b>It's wonderful.</b><b>I love showing the</b><b>audio you created at me.</b><b>Just so many people and it's like so many</b><b>didn't realize how accurate.</b><b>Yeah.</b><b>And how good quality of the</b><b>audio for my tool comes out now.</b><b>Like it's, they've always just assumed</b><b>that you can always tell.</b><b>You could like whenever they hear things,</b><b>there's usually that sort of</b><b>roboticness to other things.</b><b>Now that's just gone.</b><b>See, for me with my voice one, I can</b><b>tell, but then other people are like,</b><b>"Oh my God, it's really you."</b><b>But like there's little tells</b><b>and stuff because it either,</b><b>because I haven't said enough, because it</b><b>depends on the training set.</b><b>Like it needs a minute of data or so.</b><b>But if there's words or pronunciations of</b><b>syllables that I haven't said,</b><b>it will have to guess through that.</b><b>And sometimes it guesses American or</b><b>overly English, not Australian.</b><b>But anyway, your accent, it's pretty...</b><b>It's all over the place.</b><b>You know what I was</b><b>like living in London?</b><b>Like I was when Ozzy's around.</b><b>Hello.</b><b>Hello.</b><b>You sound like you're from London.</b><b>Let's go to work.</b><b>When you hang around Ozzy's and stuff,</b><b>the accent comes out more.</b><b>I do wonder if like with</b><b>language, it's a language model.</b><b>Like if you, you know, when you talk to</b><b>it more, starts to mimic you.</b><b>I think we'll see some stuff like that.</b><b>But in any case, the AI futures are here.</b><b>Or at least that's what Ethan Mollick was</b><b>talking about and stuff.</b><b>We'll dive into an article there.</b><b>But before we dive in, why don't we...</b><b>Can you tell me a bit about what have you</b><b>been up to this week?</b><b>Oh, this week has been...</b><b>We've got some interesting webinars that</b><b>are coming up all about...</b><b>Specifically AI.</b><b>Okay.</b><b>How you can apply it in your day-to-day</b><b>work life, workflows,</b><b>depending on what type of role you have.</b><b>And we'll go through some interesting</b><b>ways of integrating new tools</b><b>and understanding new tools into</b><b>organizations and businesses.</b><b>That's been a big focus for us this week.</b><b>Love it.</b><b>And a few bits of client work</b><b>specifically around AI as well.</b><b>Really?</b><b>Okay.</b><b>For...</b><b>Actually, I realize I probably can't say</b><b>anything around the AI stuff.</b><b>But yeah.</b><b>And even in our own business, we're</b><b>addressing some AI capabilities.</b><b>Actually around voice and meetings and</b><b>trying to close in a bit of a gap</b><b>where a lot of existing tools currently</b><b>aren't hitting that particular market.</b><b>I hear what you mean.</b><b>Like there's a lot of...</b><b>And maybe you're getting at this, but</b><b>there's a lot of knowledge</b><b>people are realizing that...</b><b>So before the advent of AI, we're like,</b><b>"Okay, we just go about and</b><b>doing our thing and stuff."</b><b>And like, "I have this meeting. I take</b><b>notes. I didn't take notes.</b><b>I missed it. I have to go back through</b><b>notes. I put stuff in Google Docs."</b><b>Blah, blah, blah.</b><b>Whatever you were using</b><b>to organize your knowledge.</b><b>And AI comes along and it's like, "Wow,</b><b>look at this magical box."</b><b>And then it does all these things.</b><b>Then we realize that,</b><b>"Oh, it doesn't do this."</b><b>And here's how you have to tackle the</b><b>hallucinations and inaccuracies.</b><b>Exactly.</b><b>But another big thing</b><b>people are seeing is that,</b><b>"Hang on a second. We don't have this</b><b>information in here."</b><b>Or, "We've got this</b><b>information, but it's a bit ambiguous."</b><b>They're realizing their documentation</b><b>because when they ask the AI,</b><b>they don't quite get the right answer.</b><b>And then it's not just the incorrect</b><b>answers or whatever, but</b><b>the missing information.</b><b>And we're realizing that there's so much</b><b>that is just not recorded.</b><b>We don't tape ourselves. We don't do just</b><b>the conversations that</b><b>we're having and stuff,</b><b>or even like team meetings and stuff that</b><b>we should record them.</b><b>Exactly.</b><b>So people are realizing the</b><b>missing information there.</b><b>I think it's wonderful because it'll</b><b>really kickstart the information age.</b><b>This is the key to it.</b><b>Until we had this tool, it was like,</b><b>"Well, whatever. We'll</b><b>just miss out on it."</b><b>So it's interesting.</b><b>And even the application of roles.</b><b>So it might be a tool that has, and this</b><b>talks to the fact that that</b><b>information might not exist,</b><b>but one, this feature set may work really</b><b>well for someone in some particular role.</b><b>Let's just say product, for example, that</b><b>you follow a bunch of</b><b>procedures and roles</b><b>and do certain things.</b><b>Those procedures and roles are different</b><b>to someone who might be in sales.</b><b>And therefore, the application of that</b><b>tool suddenly shifts.</b><b>Even though it may hit 80% of it, that</b><b>20% is super important.</b><b>So how do we close that</b><b>gap for everything else?</b><b>And that's what we're exploring.</b><b>You want to break that gap.</b><b>It's really good. I agree.</b><b>What have you been doing?</b><b>What's your work like?</b><b>Oh, apart from just</b><b>recovering from being crowdstruck.</b><b>It was like, I think, what</b><b>did we say in the last episode?</b><b>The last episode was only a few days ago</b><b>because it was delayed till Sunday.</b><b>It's like we normally</b><b>do this on a Thursday.</b><b>So it feels weird doing it now.</b><b>We've had such a short gap.</b><b>Yeah, yeah.</b><b>Anyways, back to normal is good.</b><b>I mean, the good thing about it is I</b><b>could copy lots of</b><b>notes from our last one</b><b>because we kind of shortened things a</b><b>little bit</b><b>intentionally just because it was...</b><b>Oh, exactly.</b><b>Don't be rushed.</b><b>People did like the online one.</b><b>So I mean, as much as it's</b><b>good in person and stuff,</b><b>just because we had the cartoon cut out,</b><b>like the comic book style cut outs.</b><b>Yes.</b><b>That was really good.</b><b>So I don't know how we do that with this.</b><b>I could probably apply it again.</b><b>I just like close the gap, particularly</b><b>when we do the split screen.</b><b>It's on brand.</b><b>That's good.</b><b>So it's good.</b><b>I'm going to get it.</b><b>So people had comments about</b><b>that, which is really good.</b><b>The shorts are getting out there,</b><b>although they're auto shorts.</b><b>So like the short video cut outs from</b><b>this, we're using Opus to do that.</b><b>We'll have to take a look at some of that</b><b>stuff because it's like,</b><b>come out with stuff where people looking</b><b>at it and the comment was like,</b><b>what are you guys talking about?</b><b>Because it's taking it out of context.</b><b>Anyway, or hookers there.</b><b>It's missing the element.</b><b>Yeah, it's the context wasn't there.</b><b>It's also if you like social media, you</b><b>know, having a bit of</b><b>background in social</b><b>marketing as well.</b><b>You want to make sure you're starting</b><b>with the hook and the videos that the AI</b><b>tends to punch out is misses the hook.</b><b>It just does the whole story.</b><b>Yeah.</b><b>People don't have an attention span for</b><b>like two, three</b><b>minutes worth of watching,</b><b>right?</b><b>To get what the story is about.</b><b>So they come in with this, this initial</b><b>one minute of hallucinations.</b><b>What are they?</b><b>Yeah.</b><b>And it was like, what the</b><b>hell are they talking about?</b><b>Is it hallucinations?</b><b>Strawberry?</b><b>What is strawberry?</b><b>Don't you like strawberries?</b><b>Anyway, so apart from that, I've been</b><b>really busy this week with</b><b>a couple of projects on</b><b>in the blockchain space.</b><b>We're finishing up something in the real</b><b>world asset</b><b>tokenization space, which includes</b><b>a lot of blockchain stuff has been like</b><b>pretend, not pretend,</b><b>but like proof of concept.</b><b>And you need that to</b><b>showcase what's possible.</b><b>And now we've got some</b><b>stuff where it's actually real.</b><b>There's something where there's hundreds</b><b>of thousands of things being created or</b><b>technically there's hundreds</b><b>of thousands of on on chain,</b><b>hundreds of thousands of wallets that are</b><b>being created for a</b><b>real world thing that is</b><b>being brought to blockchain.</b><b>And I'll talk more</b><b>about that another time.</b><b>Apart from that, we've upgraded psych.</b><b>So that was using super exciting.</b><b>It's funny because even</b><b>just a few seconds difference.</b><b>And let me just show you here.</b><b>Like, oops, that's another screen there.</b><b>Let's go back to that.</b><b>So, you know, folks, this is behind.</b><b>Bring it up, Jamie.</b><b>Bring it up there.</b><b>Mark.</b><b>So here is a dream that I've got.</b><b>Now that's not the right dream.</b><b>I want to take that out.</b><b>I want to go to the psych description</b><b>because here, this</b><b>description is all about psych.</b><b>It's got its features.</b><b>You can expand on that just for people</b><b>that might not be watching.</b><b>I'm in the dream section of psych where</b><b>we've got various post-it notes.</b><b>It used to be that reading all of this,</b><b>it might take a few seconds</b><b>when you ask a query on that.</b><b>A few more seconds than it does now.</b><b>And now it almost comes out like we</b><b>waited one second there.</b><b>And it just starts</b><b>spitting out things straight away.</b><b>And even reading documents.</b><b>I was explaining to a client the other</b><b>day, and we just put</b><b>this in, where I was like,</b><b>okay, well, because it's reading</b><b>documents, it's running RAG, retrieval,</b><b>augmented generation.</b><b>It's going to take a few more seconds to,</b><b>and as I'm explaining it,</b><b>it's already running it.</b><b>And it's completed the query.</b><b>So that was, it was one of the, you know,</b><b>one of the things I kept bringing up.</b><b>I'm like, you're the only way you're</b><b>going to get biggest cut through is you</b><b>need to reduce the time.</b><b>Yeah. To load.</b><b>And so this has this significantly gets</b><b>you ahead of a lot of other tools that</b><b>are in the same space,</b><b>which is a lot.</b><b>Exactly.</b><b>I mean, if anyone is using like the newer</b><b>kind of language models or</b><b>like older language models,</b><b>I mean, upgrade to the new ones because</b><b>there is better</b><b>services coming out there.</b><b>And another point to this is that the</b><b>cost is lower between</b><b>like 10 to 20 times lower.</b><b>So, you know, it is why is that?</b><b>Why is the cost so</b><b>significantly different?</b><b>Give it to the newer model.</b><b>Same.</b><b>Usually, no, when you upgrade to the</b><b>newer model, it's more pricey.</b><b>Typically, but when we went to GPT 3.5,</b><b>yeah, when you upgrade to a new human,</b><b>typically, the second</b><b>child is more costly.</b><b>I'm joking like, Lana, who's not going to</b><b>be upgraded if she'd</b><b>like, she'll be cloned.</b><b>The AI.</b><b>We love kids both.</b><b>AI daughter is more expensive.</b><b>Yeah, exactly.</b><b>But what was I going to say?</b><b>The whole upgrade stuff, we typically</b><b>used to more expensive.</b><b>Sure.</b><b>3.5 going from GPT 3.5 to</b><b>4 and then 4 to 4 turbo.</b><b>Things did get cheaper and you got more</b><b>reasoning power and you got more,</b><b>not necessarily speed, but mind you,</b><b>because the older model</b><b>3.5 versus 4 3.5 was faster,</b><b>but 4 had more</b><b>reasoning and stuff like that.</b><b>More computational.</b><b>Right now, we've got 4 turbo and then now</b><b>it's on the and there is still 3.5.</b><b>3.5 is like your baseline that you use</b><b>for mundane tasks, right?</b><b>But what's ended up</b><b>happening is and it's faster.</b><b>What will end up happening is that 4</b><b>Omni, the latest, has a mini model.</b><b>That mini model is</b><b>supposed to replace 3.5.</b><b>It's smaller, it's a lot faster and it's</b><b>just crazy how they're doing it.</b><b>The whole point of this is that they</b><b>want, or Sam, I think has been on record</b><b>as saying that he does</b><b>not, Sam Altman does not</b><b>want, what are you laughing about?</b><b>I have to say, smaller and a lot faster.</b><b>Everything a person would not</b><b>want in a real relationship.</b><b>Sometimes smaller is better.</b><b>It's what you do with it.</b><b>It's what you do with</b><b>your AI model, folks.</b><b>Welcome to the adult show.</b><b>The interesting thing is that he doesn't</b><b>want tokens and costs to be an issue.</b><b>He wants that to not</b><b>be a barrier, basically.</b><b>And it has been.</b><b>When things have been really expensive</b><b>and sometimes what you're doing,</b><b>imagine that you're doing something</b><b>really complex and it</b><b>costs you $5 to do that because</b><b>it's bouncing between all these documents</b><b>and soaking up so many things.</b><b>It can get expensive, but making it</b><b>cheaper, just like mobile phone planes,</b><b>where maybe you even have a plan where</b><b>it's unlimited, I think</b><b>we're going to get to that.</b><b>This is really good.</b><b>Omni was one that there's other models</b><b>out there that we'll</b><b>talk about in a little bit.</b><b>But this was show and tell.</b><b>This is the most exciting thing because</b><b>it just unlocks so much.</b><b>Yeah, this is really awesome.</b><b>It's very exciting.</b><b>We'll have the offers for users, like</b><b>anyone that wants to sign up.</b><b>We'll give them a three month trial.</b><b>Hit us up, folks.</b><b>It's right there.</b><b>And we're going to be doing more stuff</b><b>together with you</b><b>guys to bring to market.</b><b>So I've got a question.</b><b>So when you went through this journey of</b><b>getting into this platform,</b><b>so I'm going to take a step back out of</b><b>the AI more into the</b><b>business and development,</b><b>because this is kind of</b><b>what we're doing in exploring.</b><b>It's that MVP validation stage.</b><b>Did you go through a lot of that initial</b><b>or did you just go headstrong into,</b><b>you know what, let's</b><b>just build this thing.</b><b>Like all great</b><b>entrepreneurs diving headfirst in.</b><b>Without seeing how deep the</b><b>water is has been a thing.</b><b>No, we we got validation from the market</b><b>because we had clients.</b><b>We were not going to build something.</b><b>We were just very happy.</b><b>Like in our group, like I was one of the</b><b>first ones over at Not Centralized</b><b>that was using chat TVT and the other</b><b>boys like they</b><b>wondered about it and stuff.</b><b>The other staff were looking at and like,</b><b>how is this working?</b><b>And then we started getting</b><b>gigs doing training as in,</b><b>hey, can you come in and</b><b>train us on how to use it?</b><b>We see your post about it.</b><b>Can you teach us?</b><b>Not teaching people privacy and the good</b><b>etiquette for when you're using</b><b>these public kind of models,</b><b>even if it is your own data.</b><b>But then it was like we want to use this,</b><b>but we can't because</b><b>of the privacy issue.</b><b>That was the first thing that we tackled.</b><b>And thankfully Microsoft had their model</b><b>that stays in Australia</b><b>because they've got a</b><b>copy of OpenAI's models.</b><b>From that, when we were building, it was</b><b>directly with a client.</b><b>So we were getting direct feedback and</b><b>validation as we're doing it.</b><b>Fantastic.</b><b>Which is not everyone gets to do that.</b><b>Like it's rare.</b><b>Sometimes you just have a great idea and</b><b>you want to bring it to market,</b><b>but we would have tried to</b><b>validate as soon as we could.</b><b>Right.</b><b>If we went from the cold</b><b>start, we had a bit more warmer.</b><b>But even then with some others in</b><b>different industries,</b><b>even like with finance,</b><b>like we didn't get even the way from the</b><b>financial market space,</b><b>we started with healthcare and we started</b><b>with ESG and we started with legal.</b><b>We didn't go into</b><b>finance until more recently.</b><b>And so there was a lot of</b><b>like validation directly.</b><b>And we're lucky that we've got a network</b><b>that we were able to tap into.</b><b>So I guess we would have been following</b><b>best practices as much as I want to go</b><b>YOLO, dive head first in.</b><b>It was a mix of that,</b><b>but a lot of iteration.</b><b>Yeah.</b><b>Awesome.</b><b>Yeah.</b><b>And that's the kind of thing that we're</b><b>looking at now with ours.</b><b>I hinted at it.</b><b>I want to give everything away into where</b><b>we're confident that it's something</b><b>that we can start to</b><b>roll out into the market.</b><b>But it's that initial going through that</b><b>MVP where we're validating right now.</b><b>Is it possible to do what we want to do?</b><b>Can we take these existing tools or even</b><b>just tweaking some of the things</b><b>that we have in hand?</b><b>I'm trying to be really</b><b>subtle and beat around the bush.</b><b>And then we integrate that with in a real</b><b>scenario and can we tick that box?</b><b>Can we go, yeah, cool.</b><b>This is possible for this</b><b>one little piece of story.</b><b>And therefore, can we then scale it out?</b><b>You understand this quite well.</b><b>You guys in the digital</b><b>village and even just your own work,</b><b>the whole kind of like that concept of</b><b>stigma in something, you know,</b><b>that it looks good and it's clickable and</b><b>you can do things as a prototype.</b><b>Yeah.</b><b>I think that's the kind of concept that</b><b>we should think about</b><b>when it comes to building</b><b>solutions for our clients, right?</b><b>Either with existing or like made up.</b><b>I would take it a step back.</b><b>It's not even about looking good</b><b>necessarily, but can it,</b><b>does it fit in where they,</b><b>a journey that someone is going through?</b><b>So if you have an instance of</b><b>someone who is doing something,</b><b>there's a bunch of steps that they go</b><b>through to do something.</b><b>And then can you integrate some kind of</b><b>solution into that story,</b><b>into that journey that that person is</b><b>completing and can you</b><b>make that experience better?</b><b>And that's the thing that we've got.</b><b>We understand the journey because of our</b><b>experience and our</b><b>customers' experiences in the market.</b><b>And we can really</b><b>rapidly then, okay, cool.</b><b>This is kind of what our, the tools</b><b>journey needs to be like.</b><b>And now we're going, okay, cool.</b><b>Can we take that prototype?</b><b>Can we actually apply</b><b>it in a real scenario?</b><b>And that's what we're testing right now.</b><b>I like it.</b><b>Which is the, yeah.</b><b>So we're going through the sort of real</b><b>traditional MVP rapidly.</b><b>We've got four weeks.</b><b>We're doing in four weeks.</b><b>We're really sure</b><b>we're like shoeboxing this.</b><b>Okay.</b><b>To hang by the end of four weeks, we need</b><b>to have a definitive yes or no.</b><b>Is this possible for us to do?</b><b>And then being the app store.</b><b>That's great.</b><b>From there, let's go.</b><b>And I think that's a really</b><b>good way to do things, right?</b><b>And I don't see enough businesses.</b><b>It's almost like that sandbox prototype.</b><b>Google time, you know, setting aside a</b><b>bit of time to just test this out.</b><b>Yeah.</b><b>A lot of times people, because they don't</b><b>purposely make that</b><b>time and I don't get it,</b><b>but you need to do that.</b><b>Otherwise you just get stuck and others</b><b>will be able to outcompete you.</b><b>Yep.</b><b>If you've got the money, but if you</b><b>don't, the biggest issue I</b><b>see with that is I see a lot</b><b>of businesses, they go headfirst into the</b><b>tech or entrepreneurs into the tech.</b><b>So they start building something without</b><b>understanding or</b><b>considering the experience.</b><b>I don't want to say the UX in total.</b><b>You don't have to design things out.</b><b>You don't have to map it out.</b><b>You don't have to have</b><b>every single screen shaped.</b><b>It's mainly about understanding the</b><b>journey and then the</b><b>application of your solution</b><b>into the journey to be</b><b>able to validate something.</b><b>And you can do that visually or you can</b><b>do that with tech and AI.</b><b>My God, shortcuts that process.</b><b>Yeah, that's it.</b><b>To the sky.</b><b>And we'll talk about some tools in a</b><b>moment as we go through</b><b>this today session around how</b><b>things that are coming out into market</b><b>that are really helping</b><b>with that overall experience.</b><b>But that if you dive headfirst into tech,</b><b>you are shoestringing yourself.</b><b>Yeah.</b><b>You're putting yourself in a box.</b><b>You're putting yourself in a box.</b><b>You're not understanding 100%</b><b>what it is you're developing.</b><b>You end up doing things that are</b><b>unnecessary and</b><b>therefore taking longer to...</b><b>I just think of a scenario where you go</b><b>to someone that is</b><b>doing the complete opposite</b><b>of what we're recommending and then</b><b>you're asking, "Okay, so</b><b>have you tried to understand</b><b>the problem before you've done the tech?"</b><b>And they go, "The problem?"</b><b>It is just a cool idea.</b><b>It's like, "Ah, so..."</b><b>Yeah, but I see you said AI and stuff.</b><b>Yes.</b><b>I get it in one way.</b><b>Flip side of that is</b><b>that there's limited...</b><b>Not all businesses have</b><b>full 100% kind of view.</b><b>That's why you have</b><b>management layers and stuff like that.</b><b>But it is very easy to just go, "Oh,</b><b>let's see what AI does and stuff there."</b><b>And then it gets</b><b>misinterpreted to, "We need to do AI."</b><b>And it's like, "No,</b><b>that's not what I said."</b><b>It's like that in that Silicon Valley,</b><b>like Gavin Belson, the CEO of the...</b><b>Hulley, the fake</b><b>company, leaves and he's like...</b><b>He says, "He's actually</b><b>had some breakfast stuff."</b><b>And he's like, "Oh, the bear is...</b><b>There's a bear with honey</b><b>in it, like a plastic bear."</b><b>And he tells one of his assistants, "The</b><b>bear is sticky with honey."</b><b>And then they interpret that to mean</b><b>something about a product.</b><b>And so they form two factions, like one</b><b>that thinks it means this,</b><b>one that thinks</b><b>everything's that in any way.</b><b>It actually was a bear.</b><b>It turns out it</b><b>actually was a honey bear.</b><b>Like it was actually food related.</b><b>But can I dive into</b><b>possible futures of AI?</b><b>So we follow along with a</b><b>guy named Ethan Mollick.</b><b>He is a teacher of many things.</b><b>He's got a great book</b><b>out called Co-Intelligence.</b><b>I highly recommend reading that if you're</b><b>in the bookstores or downloading it,</b><b>you know, off Audible, et cetera.</b><b>But he had an article on his blog and he</b><b>talks about a few things there.</b><b>Significant debate is out there about the</b><b>future trajectory of AI.</b><b>Some experts believe</b><b>it's exponential growth.</b><b>Others think that we've reached our peak</b><b>and it will be slow.</b><b>Either or, I think we're</b><b>going to definitely improve.</b><b>One thing that he does mention, and we</b><b>just kind of talk about it then,</b><b>is that organizations</b><b>are not adequately planning</b><b>for the potential impacts</b><b>of continued AI improvements.</b><b>And what I mean is that, you know,</b><b>they're just thinking about things,</b><b>how we do things now.</b><b>How does AI work into that?</b><b>Rather than if we were</b><b>to complete the overhaul,</b><b>how could we rethink about our processes</b><b>if AI was at the center?</b><b>And another thing that was quite</b><b>interesting is that he</b><b>says that there's a lack of</b><b>comprehensive, what he calls like a, I</b><b>get it right,</b><b>non-technical documentation.</b><b>So for current AI systems, and this leads</b><b>to misunderstandings about capabilities.</b><b>We know there's a lot of</b><b>misunderstandings, like the whole</b><b>counting strawberries,</b><b>counting Rs in the letter</b><b>strawberries and people not.</b><b>And we're going to get a short on this</b><b>again, that would just be misinterpreted.</b><b>Because like counting</b><b>how, like doing math.</b><b>There you go, folks.</b><b>I got it early.</b><b>Doing math is not what</b><b>these models are good at.</b><b>They have to do</b><b>function calling and whatnot.</b><b>But just people knowing that there should</b><b>be this non-technical documentation</b><b>that just teaches people.</b><b>I don't know if people are trying to</b><b>educate, and you guys are too,</b><b>but we're just saying it's important.</b><b>It's 100% important.</b><b>And as you hinted at, it's exactly what</b><b>we're touching on a webinar.</b><b>And that's not an</b><b>intentional sneaky plug there.</b><b>It's just it's a literal</b><b>observation in the market.</b><b>And I sort of, the way I look at it is</b><b>like there's sort of four facets of it.</b><b>It's understanding what you need it for.</b><b>So what are the things</b><b>that you're doing that can be,</b><b>have improved the experience upon to help</b><b>fast track your thing?</b><b>The second one is then understanding the</b><b>capabilities or understanding</b><b>how those tools then fit into your work</b><b>and what they're capable of doing.</b><b>So exactly what you touch on is</b><b>mathematics, for example.</b><b>Your chat GPT is your, for your metas,</b><b>your other tools that are all coming out.</b><b>We're touching on that momentarily.</b><b>They're LLMs.</b><b>Their whole purpose is language strategy,</b><b>as you can break it down to.</b><b>They look for pattern recognition.</b><b>They're not mathematical tools, and they</b><b>need to call something else</b><b>to get mathematical results.</b><b>And even when they call something else,</b><b>they're using their language model</b><b>to generate Python, which is still, which</b><b>could be wrong to then do that,</b><b>to do that solution.</b><b>And it's getting more accurate and there</b><b>are tools that are coming out.</b><b>So that's the second thing.</b><b>And then the third thing is then risk.</b><b>So what, similar to what Psyc is trying</b><b>to solve, it's the</b><b>risk against your data,</b><b>understanding the risk against your</b><b>results, understanding how</b><b>to make sure you're checking</b><b>and validating and security and all those</b><b>types of things that bring risk,</b><b>which can also be a risk to the business,</b><b>which to the culture</b><b>of your organization,</b><b>risk to the processes and work that you</b><b>do risk to how you're perceived</b><b>by your customers and clients as well.</b><b>Internal and external.</b><b>Exactly right.</b><b>And four.</b><b>And number four is then how to utilize</b><b>them more than</b><b>understanding the capability,</b><b>but understand how to utilize them and</b><b>that's how do you avoid hallucinations.</b><b>So what are the terminology I'm hating in</b><b>the market right now,</b><b>which is prompt engineering.</b><b>So how do you prompt the LLMs properly to</b><b>get the most accurate</b><b>results and how do you then</b><b>validate those results?</b><b>And they're the sort of four key things</b><b>if you to understand the</b><b>application of LL to AI within</b><b>your business, within</b><b>your life, workforce, etc.</b><b>In my mind was like</b><b>one, two, three, four.</b><b>Yep, I am better than an</b><b>LLM because I can count.</b><b>But I think it would do</b><b>well at that kind of stuff.</b><b>But it's interesting what you mentioned</b><b>there because, you know,</b><b>on the prompt engineering</b><b>side, for example, like there's many a</b><b>debate on I'm seeing a debate.</b><b>I don't know about others, but like where</b><b>I hear people talking</b><b>about prompt engineering</b><b>and then others question, well, shouldn't</b><b>the AI just give you</b><b>what the best prompt is?</b><b>And I think it's a mix of both.</b><b>We don't stop people from making bread,</b><b>artisanal handcrafted</b><b>sourdough that we all did during</b><b>COVID.</b><b>But we don't stop people from doing that</b><b>because just because we've</b><b>got automation in the bread</b><b>making process, right?</b><b>Yeah, it's choice.</b><b>And I think that there's going to be some</b><b>people that will just want</b><b>to handcraft their prompt.</b><b>And some people that will just go,</b><b>actually, I want the AI to</b><b>tell me and I'm going to learn</b><b>from what it shows me as like best</b><b>practice, like even seeing things.</b><b>The underlying, you know, this is</b><b>interesting, the underlying custom</b><b>prompts that when you</b><b>ask query, people don't realize this,</b><b>like, why does Claude sound</b><b>a little bit this way versus</b><b>chat GPT sounding that way versus others?</b><b>They've got custom instructions, like a</b><b>page or whatever it is</b><b>of how they're supposed to</b><b>take your query and then</b><b>answer that to their model.</b><b>And that is something you can't change.</b><b>You can kind of play around with it.</b><b>That would reveal in many ways, Claude,</b><b>it seems a little bit friendlier.</b><b>Sometimes when you ask it a question, it</b><b>doesn't just go off and do it.</b><b>It'll actually ask you a few follow up</b><b>questions to clarify.</b><b>Like I was doing coding stuff with it and</b><b>it was giving me,</b><b>okay, there seems to be an</b><b>error there.</b><b>Can you try this code</b><b>and tell me the results?</b><b>I'm like, what?</b><b>What the heck?</b><b>It's asked because I would just normally</b><b>do that with chat GPT.</b><b>Yeah, he's a fix.</b><b>Try this.</b><b>No, but this one, it goes, do this first.</b><b>We'll do an interim step</b><b>first and we'll go back and forth.</b><b>Now that happens because it's got the</b><b>custom instructions.</b><b>But anyway, I was digressing from the</b><b>prompt engineering side.</b><b>I think all of this is education.</b><b>It's a good point though, because the,</b><b>and they are getting smarter doing that.</b><b>There are tools that you can now ask for</b><b>appropriate prompts.</b><b>There are, Claude, as you highlighted,</b><b>even chat GPT are now</b><b>giving you multiple responses</b><b>based on assumptions that it's making.</b><b>You're like this or that one?</b><b>Yeah, exactly right.</b><b>But like, I mean, I wouldn't necessarily</b><b>want it to, I want it to clarify.</b><b>I wouldn't necessarily want to go, hey,</b><b>you should ask the question in this way.</b><b>Because you think you're removing it away</b><b>from its intent of trying to be this</b><b>conversational tool.</b><b>Like if you just, if you wanted to give</b><b>you the answer to go</b><b>to Google search things,</b><b>like why, why are you in an LLM?</b><b>Well, yeah, potentially and stuff.</b><b>And the other thing is that I think what</b><b>you just said there with conversational,</b><b>that touches on a point where some things</b><b>you definitely need as conversational,</b><b>like AI agents, bots that sit on your</b><b>website, or even just internal stuff.</b><b>There's a new starter and you've got data</b><b>in a room that they can like talk to.</b><b>But sometimes it's not conversational AI</b><b>you need, you need like workflow type AI.</b><b>So whether you're having agents that</b><b>you're giving instructions</b><b>to when you don't actually</b><b>need to talk to them, or it's just like</b><b>you've got a, and this is where the</b><b>dreams thing comes in.</b><b>It's for ideation.</b><b>Like I don't need something going back</b><b>and like, yeah, I can see a</b><b>conversational by helping there.</b><b>But we're using that a lot more for like,</b><b>okay, I just got a meeting with Chris,</b><b>I'll put these ideas in, let</b><b>me look at all my old ideas.</b><b>Does this crossover?</b><b>Okay, I run the AI to</b><b>synthesize everything.</b><b>I'm just keeping these posts and notes in</b><b>something that is</b><b>powered by AI, AI rather than</b><b>Google Docs.</b><b>So there's different types of usage to</b><b>your point earlier before.</b><b>And I think people understanding that</b><b>along with the risk</b><b>and how these things work</b><b>is going to be massively important.</b><b>And so yeah, we're very happy to</b><b>contribute to the stuff</b><b>that you guys are doing there.</b><b>So we'll be talking more about that on</b><b>the show soon, I'm sure.</b><b>Well, I think you could relate it almost</b><b>to if you're having a</b><b>conversation with a person.</b><b>Yeah.</b><b>Right. And how many times have you had a</b><b>conversation with someone?</b><b>If you said something and that person has</b><b>completely misunderstood the question all</b><b>the time or the story or yeah.</b><b>And do you expect that person to be like,</b><b>Oh, did you actually mean this?</b><b>When you, when you ask him that question,</b><b>you let, you know, you</b><b>just expect the person</b><b>to understand and they go away and do it</b><b>and they do it wrong.</b><b>Yeah. I'll have five things, Chris.</b><b>So why would an LLM know that what you're</b><b>asking it is, is wrong or right,</b><b>or should be set in a different way to</b><b>get the best response?</b><b>Like, especially when all of its data</b><b>comes from how actions and our,</b><b>our content that we're sharing, it</b><b>doesn't have that ability</b><b>to, to do that realistically.</b><b>It does what you tell it</b><b>to do or what ask of it.</b><b>Based on the context, based on the</b><b>context, you give us the</b><b>highest probability answer.</b><b>I just wonder about that, whether you can</b><b>ask it to give me the</b><b>lowest probability answer.</b><b>You probably could ask.</b><b>Yeah.</b><b>The thing is, and this is the next part I</b><b>was going to say is you can go,</b><b>can you, is there a</b><b>better way to ask this or?</b><b>Or you change the temperature.</b><b>Change, yeah.</b><b>You get it to be more, a</b><b>bit more out of the way.</b><b>Yeah, exactly right.</b><b>And then, yeah, there are.</b><b>Give me the most racist,</b><b>incorrect, rude response you can.</b><b>Just another weekend at Chris's.</b><b>ChatGPT will not give</b><b>you, your responses.</b><b>You're a very nice bot.</b><b>Yeah, exactly. You're very nice.</b><b>But I think it's really interesting</b><b>seeing articles like that.</b><b>Just going back to Ethemonics, definitely</b><b>following, you know, the things that he</b><b>sees and shares because</b><b>it gets a lot of follows.</b><b>And I think people are also, he's the</b><b>thing like why you think</b><b>it's interesting to follow</b><b>folks like that, the mini celebrities in</b><b>the space, because</b><b>they'll also get sent stuff</b><b>and be able to try things</b><b>out before others potentially.</b><b>So I think it is important to check out.</b><b>And then I had.</b><b>Definitely, just enough of all I'm doing,</b><b>definitely read his book.</b><b>If anyone out there wants to do it, it's</b><b>a very powerful read.</b><b>Or get the eye to read it.</b><b>Or get the eye to summarise it.</b><b>We're exactly right.</b><b>Yeah, we'll do a deal with Ethan to just</b><b>have it as part of Psych.</b><b>And you can just talk to</b><b>Ethan while he's in there.</b><b>But anyway, I digress.</b><b>The interesting thing, another</b><b>interesting thing I saw was, and I'll</b><b>share the link to this,</b><b>but it's a new channel that I've started</b><b>watching where they're doing very,</b><b>what I would call practical takes on AI.</b><b>And I like it because, you know, the</b><b>thing I think is called Neat Code.</b><b>Neat Code.io.</b><b>And I haven't watched his other videos,</b><b>but in this one, it</b><b>really struck out to me because</b><b>whilst he might be a bit critiquing of</b><b>AI, that is a word, folks, critiquing,</b><b>it was done with a sense of pragmatism,</b><b>where it's like, yeah, I do</b><b>believe that AI is going to</b><b>change our future and stuff.</b><b>But just given how things have worked</b><b>out, just like we see</b><b>with other technologies,</b><b>hype can be a little</b><b>bit of a virus, right?</b><b>Definitely.</b><b>It can oversell things.</b><b>But he did say that, you know, whilst</b><b>there is the overselling,</b><b>people usually stop</b><b>at the argument there.</b><b>It's like, oh, NFTs are this bad.</b><b>People are scamming this</b><b>and they're overhyping bad.</b><b>But it's like, read beneath the headline.</b><b>We've seen it when search first came out.</b><b>We've seen it when blockchain came out.</b><b>We've now seen it with AI.</b><b>The myriad of times that this happens</b><b>every single way when people</b><b>jump onto things too early.</b><b>There's a silver lining,</b><b>though, with all this kind of stuff.</b><b>People would jump on</b><b>early, and this is human nature.</b><b>But we think that that's all bad.</b><b>It's not actually all bad.</b><b>Yes, we should watch out and protect</b><b>ourselves against the scams</b><b>and be realistic about the</b><b>claims that people are having out there.</b><b>But because of all the hype, it brings</b><b>people into the space that</b><b>weren't necessarily there.</b><b>It brings, A, the</b><b>pragmatic people in there.</b><b>Not all, but like it brings a lot of</b><b>people that will take</b><b>this space forward that would</b><b>not have looked at the space otherwise.</b><b>We, for example, would not have been in</b><b>the blockchain space had</b><b>it not had the hype that</b><b>it had because we were brought into work</b><b>on a project that got</b><b>funding because of the</b><b>hype that it had.</b><b>Now we're building</b><b>stuff that is pragmatic.</b><b>The attention that this kind of stuff</b><b>gets can be a bit of a silver lining.</b><b>They also talked about the concept of</b><b>companies in the space that,</b><b>they go, "Well, why</b><b>did Google invest in AI?"</b><b>It's like, "Well,</b><b>they've got two options.</b><b>Either not invest or do invest.</b><b>If they don't invest,</b><b>there's only one path from that.</b><b>If they do invest, then it is either it's</b><b>successful or it's not successful.</b><b>So if they do invest and</b><b>it's successful, great.</b><b>If they do invest and it's not</b><b>successful, they can try</b><b>because they've got the money too,</b><b>and they should be trying to pursue that</b><b>and push rather than just</b><b>resting on their laurels</b><b>with search.</b><b>Now, if they don't invest, they watch all</b><b>their other competitors go out there and</b><b>they will make something of it.</b><b>The guarantee is that AI will be a part</b><b>of our future, right?</b><b>Given what it can do.</b><b>They're forced to.</b><b>They have no choice but to invest.</b><b>People always wonder like,</b><b>"Oh, why is Apple doing this?"</b><b>They kind of have to because when you</b><b>look at the options,</b><b>it's like game theory,</b><b>the best strategy is to actually invest</b><b>and do something,</b><b>however big or small that is.</b><b>He talks about that, which I find</b><b>fascinating because a lot of</b><b>people just look at outcomes</b><b>and go, "Okay, well, that was bad."</b><b>You've got to look at, or</b><b>maybe not even the outcomes.</b><b>They're just judging things early.</b><b>I think we have to put ourselves in the</b><b>shoes of the customer, big, small,</b><b>the person that's presenting something,</b><b>what are they getting out of it?</b><b>Maybe that's part of the education that</b><b>you guys are going to do, and</b><b>I'll be a guest lecturer for.</b><b>Guest lecturer, Mark.</b><b>Is that what we're going to get to?</b><b>You get homework.</b><b>Everyone gets homework, you can use AI.</b><b>Love it.</b><b>But I'll just raise the</b><b>bar in terms of the homework.</b><b>If anyone is a teacher out there, if</b><b>you're doing homework and</b><b>you're letting students use AI,</b><b>it can't just be the old homework that we</b><b>were doing like, "Hey,</b><b>create an essay on this."</b><b>No, because AI is the way it is.</b><b>Create me an interactive presentation.</b><b>Figure it out.</b><b>Use AI to do that.</b><b>Actually learn how to code.</b><b>It's a history lesson, yeah, but you can</b><b>actually make it interactive.</b><b>Interesting.</b><b>It's a good thing that it has put, it's</b><b>flipped education,</b><b>thrown it onto its head</b><b>by introducing these accessible tools to</b><b>pretty much fast track how</b><b>people and students learn</b><b>some negative, obviously in a massive</b><b>negative way, but then there are</b><b>positives that come out</b><b>of it as well.</b><b>But it's proof that the system needs to</b><b>change and it has to</b><b>adapt, unfortunately.</b><b>Like we had with that place.</b><b>Which it should have done centuries ago.</b><b>Exactly.</b><b>Yeah, they should have been using AI when</b><b>they were using AI because...</b><b>Yeah, we'll go with that.</b><b>What were they doing?</b><b>They weren't thinking.</b><b>You've got some news.</b><b>Oh, there's a lot of news.</b><b>That's a couple of big things,</b><b>surprisingly, this week.</b><b>Firstly, I think the one we were</b><b>discussing recently is Meta.</b><b>Big old Meta.</b><b>We had a Meta discussion and then along</b><b>came a business that</b><b>decided I'm going to launch an AI</b><b>tool into the market.</b><b>Who knew?</b><b>Who knew this would happen?</b><b>But yeah, so Meta, good old Facebook, Mr.</b><b>Zuckerberg, has launched the open source,</b><b>well not launched, but they've announced</b><b>their own open source,</b><b>LLM, to challenge all these</b><b>other competition that's out there.</b><b>So, Llama with a ridiculous name.</b><b>It's a Llama 3.1.405B, which</b><b>I think is the current state.</b><b>I don't know if that's just the...</b><b>The B is a billion and it is.</b><b>But yeah, open source, super</b><b>powerful, reasonably quick.</b><b>It's probably not as good as the GPTs and</b><b>Clords who have had</b><b>quite a foothold in the</b><b>market for a while.</b><b>But I think given the backing that a</b><b>business like Meta has,</b><b>it's not going to be very long</b><b>before they probably</b><b>catch up in that space.</b><b>But interestingly, regardless of the tool</b><b>itself, it's what Zuckerberg has said,</b><b>regarded Mr. Zuckerberg.</b><b>Mr. Zuckey, Mr. Zuc.</b><b>And that he's intentionally wanting this</b><b>to go full open source with the...</b><b>I guess hinting to what Open AI was</b><b>originally going to be, but</b><b>being a tool for the people in</b><b>the market rather than a tool for</b><b>ourselves to be a cash hunger business,</b><b>which they obviously are.</b><b>Which I thought was very surprising.</b><b>It's an interesting sentence, like it's</b><b>going to be more open</b><b>source than Open AI.</b><b>As Clo-Clo's Meta is bringing out an open</b><b>source model to rival</b><b>Open AI's Closource.</b><b>It's like, sorry, did you just</b><b>hallucinate right there?</b><b>I'm on drugs right now.</b><b>Are you having a...</b><b>Can you feel the left side of your body?</b><b>Are you having a heart attack, Mark?</b><b>But you know, I find it fascinating and</b><b>stuff, and I've got a few takes on it.</b><b>And they've had Llama</b><b>out there, older version.</b><b>This is a great one.</b><b>And they were an R, not WER.</b><b>Average are best.</b><b>But they're open source.</b><b>And yes, there are French companies that</b><b>are also doing open source stuff.</b><b>Mistral is one of those.</b><b>And they've also come up with a newer</b><b>model as well, which people are going</b><b>through and assessing.</b><b>And speaking of assessing people, you</b><b>should definitely check out,</b><b>this will be in the</b><b>links, artificial analysis.</b><b>It's a group that's come out of Australia</b><b>that's getting a lot</b><b>of press more recently,</b><b>because they do analysis of</b><b>these tools as they come out.</b><b>Guys like Andrew Ng and even</b><b>Lex Friedman tweeted about it.</b><b>So how great to get someone that we know</b><b>from the space here in</b><b>Australia, from Sydney,</b><b>doing their thing there with analysis.</b><b>But going back to the open source thing,</b><b>he wants it to become</b><b>the standard for industry.</b><b>And I can see, I think they're going to</b><b>do some versions of models.</b><b>Like imagine a model that is trained on</b><b>consultants for startups,</b><b>and then another model</b><b>trained for legal, another model.</b><b>The biggest thing, when I look at the</b><b>spectrum of hallucinations,</b><b>when you just have a very general model</b><b>and you're asking it questions,</b><b>you'll get random answers sometimes,</b><b>that's the hallucinations.</b><b>When you have RAG, which sits on top</b><b>where you're loading documents,</b><b>and it's that, you know, general model</b><b>plus your documents,</b><b>and you're even pointing to which</b><b>documents and even better</b><b>when it's an expert pointing</b><b>to documents along the</b><b>spectrum, less hallucinations.</b><b>The ultimate in terms of low level of</b><b>hallucinations is</b><b>when you train the model,</b><b>you fine tune it.</b><b>So them having these models</b><b>that are now more capable,</b><b>and we're seeing models even being able</b><b>to run on smaller</b><b>machines, like we don't.</b><b>Now this is a question for Nvidia, right?</b><b>Like do we need</b><b>bigger and bigger machines?</b><b>If we can now run these</b><b>smaller, that's another episode.</b><b>But the open source thing I think is</b><b>fascinating because even without AI,</b><b>open source has always been a debate in</b><b>the software industry.</b><b>You can potentially</b><b>develop faster with context.</b><b>I think it's contextual.</b><b>You can develop faster in some cases, but</b><b>sometimes you need close source</b><b>to actually do the</b><b>thing that you need to do.</b><b>So it's an interesting kind of battle</b><b>that's ahead, but we're here for it.</b><b>We're loving the fact that it's not just</b><b>one side and</b><b>everything, we have to go close.</b><b>And then a few months</b><b>later, you get this.</b><b>This is an exciting day.</b><b>Yeah, exactly right.</b><b>I mean, and then I guess adding to that,</b><b>you did touch on Mistral,</b><b>who have just done a bit of a press</b><b>release around their</b><b>latest Nemo model, 12B,</b><b>which is pretty exciting.</b><b>Our standard, I think Gemma and Llama</b><b>were actually referenced in it in their</b><b>usual context windows.</b><b>They have about 8,000 tokens at once.</b><b>Yeah.</b><b>This is generating</b><b>128,000, which is massive.</b><b>Absolutely massive.</b><b>There's million token ones out there and</b><b>we'll even get to much more.</b><b>But 128 seems to be the standard.</b><b>People complained about Llama not being</b><b>that high before</b><b>saving that was open source.</b><b>So now it's 128.</b><b>Who's hitting the millions?</b><b>I forget, maybe anthropic, but there are</b><b>others out there that are</b><b>definitely in the millions.</b><b>Yeah.</b><b>But the interesting thing as well, I</b><b>think, is that when</b><b>you're not the incumbent,</b><b>so open source being the lead there and</b><b>then Google with this Gemini model,</b><b>when it gets to like Meta looking at</b><b>this, you can see potentially why they</b><b>would go open source,</b><b>because what better way to combat the</b><b>paid for versions than if you were to go,</b><b>"Hey, here's our model.</b><b>You should come to us and</b><b>look, we're making it free."</b><b>It's an interesting play that I think,</b><b>one that I think they can</b><b>afford and one that I think they</b><b>only do now.</b><b>If it was Meta was chasing the others.</b><b>I mean, so they are, but if they were not</b><b>as big as a massive company,</b><b>do I think that they</b><b>would go open source?</b><b>I don't think so.</b><b>They maybe want to try to make some...</b><b>I don't know.</b><b>I mean, you look at the other, the French</b><b>company that recently displayed,</b><b>is that what they called...</b><b>UTI?</b><b>Yeah, Katai.</b><b>We talked about it in our last episode,</b><b>we showed a bit of a video on it.</b><b>And I mean, if you think about that,</b><b>they're open source.</b><b>They're zero company,</b><b>like coming from nothing.</b><b>It's a thing that businesses do.</b><b>I think it's all about</b><b>the laurels, their culture,</b><b>what do they want to be as a business?</b><b>Yeah.</b><b>I think you're probably right, Mr. Zuck</b><b>is a little bit more money hungry.</b><b>So it'd be, in his early</b><b>days, probably would have made it.</b><b>That's why he's doing his</b><b>image and stuff like that.</b><b>It could be.</b><b>Maybe that's it.</b><b>Maybe that's it.</b><b>It's more human now, which is great,</b><b>rather than where he's from.</b><b>It's exciting.</b><b>A lot of new releases with other tools</b><b>that are existing in the</b><b>market and talking about big</b><b>companies.</b><b>We kind of hinted at Figma before.</b><b>Just...</b><b>I mean, they've been</b><b>talking about it for a long time.</b><b>And they've already got tools in the</b><b>market like FigJam, which</b><b>do have AI integration over.</b><b>Some really powerful stuff that allows</b><b>you to, whether it's</b><b>grouping contextual information</b><b>within the dashboards or creating images</b><b>on the fly, all that type of stuff.</b><b>There's a lot of interesting stuff that</b><b>helps you facilitate</b><b>workshops and things.</b><b>But Figma has been a little bit behind on</b><b>the release of AI</b><b>intentionally to try and catch</b><b>things up.</b><b>So they officially launched and it's</b><b>rolling out slowly to a</b><b>bunch of people who are paid</b><b>users within the platform.</b><b>I've been excited for this, but at the</b><b>same time, there's been a bit of</b><b>controversy around it.</b><b>Why? Yeah, tell me more.</b><b>So obviously, two big...</b><b>Yeah, Pokemon.</b><b>Two big things that you can do...</b><b>Two biggest things that you can do is</b><b>obviously text to image generation,</b><b>which you can then prototype something</b><b>and then prototype to code to find code,</b><b>which is really cool.</b><b>They're like the two big things.</b><b>It's done the front-end coding for a</b><b>while, but AI has given</b><b>that a bit of a bolster in</b><b>that space for it.</b><b>But the reason it's a bit of controversy</b><b>is because where is it</b><b>getting all of its information</b><b>on how that text image should come from?</b><b>Like what is it pulling from?</b><b>Clearly from</b><b>everyone's history of using...</b><b>Figma, exactly right.</b><b>And so what Figma hasn't done is</b><b>communicated that to its users and has</b><b>hidden the fact that</b><b>there is a setting deep within your</b><b>account and your soul</b><b>that you need to turn off in</b><b>order to say, "Don't share my details."</b><b>And people have known about this.</b><b>Turn off trauma.</b><b>Okay, turn off trauma.</b><b>Therefore, your data is hidden from...</b><b>But that hasn't been the case.</b><b>And so now they've been able to roll this</b><b>AI and it is pulling insights into...</b><b>Which like for me, I'm like, it doesn't</b><b>bother me, but I can imagine</b><b>those who sit much more on that for the</b><b>high level creative, individual</b><b>creativity, where that...</b><b>Their uniqueness is now out there.</b><b>Yeah, exactly right.</b><b>That is crazy.</b><b>Oh man, it's so hard.</b><b>For example, you're</b><b>using their app, right?</b><b>Exactly.</b><b>Like it's their app.</b><b>It's a setting in...</b><b>I know you can argue it wasn't as clear,</b><b>but another way it's there.</b><b>It's clear as day in another way.</b><b>Yeah. It could be argued that if you didn't</b><b>read the T's and C's, then...</b><b>But now if they...</b><b>It would be a problem for Figma.</b><b>I see if they clearly state and highlight</b><b>that they would not</b><b>do that sort of thing.</b><b>They definitely would not have done that.</b><b>Exactly.</b><b>But like they would have...</b><b>Their lawyers would have slammed them</b><b>down if they had</b><b>something like that written in.</b><b>But they definitely</b><b>haven't made it obvious.</b><b>And I feel bad for like folks like I have</b><b>some creative friends, right?</b><b>That whatever they're doing, images,</b><b>writing, this, that...</b><b>And it's not just creators</b><b>that are getting infected.</b><b>Even other like coders and stuff that,</b><b>you know, they think</b><b>that they're under threat.</b><b>I don't think they are.</b><b>But for certain things, we're just going</b><b>to have to like lift.</b><b>If everyone can now do what a</b><b>professional person</b><b>using Figma could do before,</b><b>to a varying level of success as the AI</b><b>improves, then the</b><b>ones that are really good,</b><b>they're going to have to</b><b>change how they do things.</b><b>People that were really good at abacus</b><b>usage, when everyone could use abacuses,</b><b>we just have to move on.</b><b>Go invent the calculator.</b><b>No, it's hard.</b><b>It is very hard.</b><b>Like it's...</b><b>I don't want to...</b><b>It's not being mean or anything, but like</b><b>we even have to think</b><b>about what do we do next,</b><b>given that it's just so easy now for</b><b>people to do things with data.</b><b>It's even easier for people to do things</b><b>with some of the strategy consulting</b><b>that you might be doing.</b><b>So I think we just have to find ways</b><b>because there's always</b><b>going to need to be humans,</b><b>but your profession is just going to have</b><b>to like lift up just</b><b>like, yeah, we're all trying.</b><b>And it's interesting you say that.</b><b>We can't stop it, right?</b><b>Like if we go, why don't you stop?</b><b>You just can contribute into it.</b><b>Because someone else is going to do it.</b><b>Exactly.</b><b>Right?</b><b>And if we don't do it and we don't</b><b>educate you, it's going to be worse.</b><b>And that's the education</b><b>factor is going to be a big play.</b><b>And there's probably going to be like</b><b>that for many years.</b><b>But we're touching on the same sort of</b><b>topic there around creativity.</b><b>Interestingly, an article just came out</b><b>from Science Daily and it was around a</b><b>study that was done focusing on</b><b>creativity, particularly</b><b>here around creative writing,</b><b>so books and stories, etc.</b><b>So they had 300 applications of stories</b><b>mixed between people who</b><b>were utilizing AI to write</b><b>their story and those who didn't.</b><b>And fascinatingly, the people who</b><b>utilized AI in some form</b><b>to write their stories ended</b><b>up with better, well-written, more</b><b>creative content to share.</b><b>Now, when I say they used AI, they didn't</b><b>plug the stuff in and just like punch out</b><b>AI and then copy-paste it.</b><b>It was used as a tool to enhance their</b><b>ideas, to come up with</b><b>creative ideas or concepts,</b><b>to then reiterate something</b><b>that they may have written.</b><b>And those stories ended up performing</b><b>better than those who just</b><b>wrote raw out of their minds.</b><b>So they did this by had 600 people who</b><b>were judging the 300 stories.</b><b>And yeah, the people with AI, they ended</b><b>up having the better</b><b>results on a majority.</b><b>That is fascinating.</b><b>And I just wanted to shameless plug here,</b><b>but let's just,</b><b>here's the actual article,</b><b>folks.</b><b>I've copied and pasted it in and let's</b><b>just ask, give me a</b><b>summary of the key points.</b><b>Let's just see.</b><b>Let's just validate Chris.</b><b>Oh, God.</b><b>Let's just see Chris actually.</b><b>Did I read the article?</b><b>Because I didn't put this throughout.</b><b>Chris talk about enhanced creativity.</b><b>Do you talk about the evaluation?</b><b>Yeah, 600 people.</b><b>Yeah, 600 judged.</b><b>Interesting.</b><b>Let's not make that up.</b><b>26.6% were better written</b><b>and 15.2% were less boring.</b><b>I would love to judge</b><b>the less boring part.</b><b>Just to see what happened.</b><b>Like what was it doing?</b><b>Collective novelty.</b><b>That is interesting.</b><b>When individuals created creativity</b><b>increased the study found a 10.7%</b><b>increase in similarity</b><b>between AI assisted stories, suggesting a</b><b>loss of collective novelty.</b><b>Oh, interesting.</b><b>I guess it's like</b><b>it's a bit of a balance.</b><b>Like you gain in some parts, but you</b><b>know, it takes back on others.</b><b>And yeah, sorry.</b><b>Sorry for hijacking.</b><b>And that's the big fear</b><b>where they are, right?</b><b>Is that it does remove a bit uniqueness</b><b>if it's trying to find an average between</b><b>everything that it does.</b><b>But I mean, that comes back</b><b>to then how you prompt it.</b><b>Because you can prompt it</b><b>to be a bit more creative.</b><b>You can prompt it to be a bit more</b><b>creative and in your style.</b><b>So let's say, you know, this</b><b>is how the person's written.</b><b>Something that we're doing often is</b><b>saying, give me the style.</b><b>And I'm writing this in, you know, we've</b><b>got the article here, folks.</b><b>And I'm saying, give me the style and</b><b>tone of this article</b><b>and provide quote examples</b><b>so that it's actually</b><b>giving us those examples.</b><b>And what it'll do is it'll say, well,</b><b>this one's academic and informative.</b><b>And here's some example</b><b>quotes from what it said.</b><b>It's the use of</b><b>technical terms and statistics.</b><b>It's neutral and objective.</b><b>So if you wanted to write in that style</b><b>again, AI can actually help you achieve</b><b>your style, you know?</b><b>But what you're still looking at here is</b><b>it's not extending beyond that novelty.</b><b>Oh, exactly.</b><b>You can make AI creative</b><b>with the right prompts.</b><b>Again, you need to know how to prompt it</b><b>properly in order to do that.</b><b>Exactly.</b><b>But at the same time, there is still</b><b>limitation because it is pulling,</b><b>again, an average of whatever you think's</b><b>within this topic realm.</b><b>It's that bell curve.</b><b>So if you want to tweak it, you need to</b><b>know how the AI works to be able to,</b><b>or you need to, even if</b><b>you know how it works,</b><b>sometimes we just didn't</b><b>have the tools available to us,</b><b>but that's coming</b><b>more and more available.</b><b>It's really exciting.</b><b>I go to bed, eat, sleep, drink.</b><b>Rent, drop, and in.</b><b>Yeah, AI.</b><b>On a similar note, there was another</b><b>study that was completed,</b><b>and this one instead was targeting</b><b>university students.</b><b>So it was psychology degree.</b><b>And they were allowed, 30% of them were</b><b>allowed to use AI to</b><b>answer to complete their exam,</b><b>and the other 17% were left to not</b><b>complete their exam.</b><b>To do it normally, sorry, by hand.</b><b>And lo and behold, the people, sorry, the</b><b>purpose of this study was to identify</b><b>the uniqueness of AI in that sense.</b><b>So could it be clearly identified that AI</b><b>was supporting these students?</b><b>And it turned out that</b><b>majority of them, bar three,</b><b>three of them were identified as AI, the</b><b>other 27% of the exams</b><b>that were done by AI were not identified</b><b>as AI and scored the highest results.</b><b>So they performed better.</b><b>It was obvious that it was an AI,</b><b>and everyone else didn't do as well as</b><b>the tools that we use for AI.</b><b>And that is talking to how much better,</b><b>and a lot of them are using</b><b>chat GPT, how much, like 4.0,</b><b>how much better that has become at</b><b>splitting itself out from</b><b>sounding too much like an AI,</b><b>which is scary in itself and talks to the</b><b>education piece that</b><b>you mentioned before,</b><b>the need to change how you do exam work,</b><b>because you can just pretty much plug it</b><b>in and not get caught,</b><b>which is bad, I would say.</b><b>It is, I guess</b><b>there's more context there.</b><b>As a direct...</b><b>So the difference between</b><b>like just plugging something,</b><b>and we don't know the specifications of</b><b>the tests that they've done.</b><b>Look, I don't know until</b><b>like, I'll read the article,</b><b>but whether they say it or not, I would</b><b>hazard a guess that they've</b><b>probably done some extra stuff.</b><b>So yeah, you might get the</b><b>AI to get you the results,</b><b>but then you the student are tweaking it,</b><b>and you're chopping and changing things</b><b>and adding your own flavor to it.</b><b>Now, if they're just going direct, you</b><b>know, one for one, spit out the answer,</b><b>put it in and leave it,</b><b>then that is interesting.</b><b>Because even...</b><b>They didn't actually talk to that</b><b>surprisingly in the article.</b><b>That's what I found with a lot of these</b><b>studies that they</b><b>don't get into the details.</b><b>Sometimes they do though, but even then</b><b>it's not enough detail.</b><b>Like I saw something where it was</b><b>financial analysis using AI,</b><b>feeding it in annual reports and getting</b><b>it to pick the direction</b><b>the share price is going to go.</b><b>And they're like, "Oh, look, it picked</b><b>the best direction like each time."</b><b>But there's so much nuance to it.</b><b>It's like, what's the</b><b>timeframe you were looking at?</b><b>How did you actually do the test?</b><b>Was it information that if you were to do</b><b>it historically at that time,</b><b>that you would have had that information</b><b>to plug into an AI to</b><b>predict that movement?</b><b>Because we do this</b><b>thing called back testing.</b><b>So there's a lot of nuance</b><b>of the experiments where,</b><b>if you introduce certain things and don't</b><b>take into account biases,</b><b>it is a problem.</b><b>And we get a lot of</b><b>talks out there in push.</b><b>Because from universities, I would expect</b><b>that it would be robust.</b><b>But I've even seen academic</b><b>type papers that are put out</b><b>just because we've got this limited</b><b>understanding of AI.</b><b>Like it's someone from not the AI space,</b><b>whatever other industry it is, finance,</b><b>healthcare, whatever,</b><b>and using AI and coming up with results.</b><b>And it's like, yeah,</b><b>that's not quite right</b><b>because that's not how AI works, right?</b><b>Or that's not how you</b><b>do that kind of testing.</b><b>So there's always nuance to this, but it</b><b>is interesting that they tried that.</b><b>Did the students know, I</b><b>didn't get to that question.</b><b>Did the students know that</b><b>the others were using like,</b><b>if you're in the 70%?</b><b>So the way I'm looking at</b><b>this, so the article, sorry,</b><b>the study was done by plus one, PLOS one.</b><b>And it was, let me get the numbers right.</b><b>Because I didn't get</b><b>it right the first time.</b><b>It was 33 out of the 100</b><b>were done using open AI.</b><b>Yeah.</b><b>My understanding is that they were all</b><b>like, they were 100% AI.</b><b>So like they were fake exams, essentially</b><b>associated that these people submitted.</b><b>And there were only two, sorry, were</b><b>detected as being written by AI.</b><b>And they scored an average</b><b>of 84% for the rest of them.</b><b>And yeah, so it wasn't like, I'm going to</b><b>use it to support me.</b><b>It was like AI output copy</b><b>paste kind of on the majority.</b><b>So humans, you know, we, we school, we go</b><b>through uni, we get</b><b>hit by the the playground</b><b>basketball in the face that is life.</b><b>Yeah. This is a sample of that.</b><b>Why am I so specific that you throw a</b><b>ball at a basketball?</b><b>You throw a tennis ball to someone at</b><b>school and you</b><b>accidentally hit a teacher in the head</b><b>and get in trouble for it.</b><b>Not that, you know,</b><b>very specific example.</b><b>Not that I know.</b><b>Anyway, I really</b><b>didn't mean I'm sorry, miss.</b><b>But you know, what I'm saying is that we</b><b>go through life and we</b><b>learn these experiences,</b><b>we go through testing, and then we go get</b><b>a job and blah, blah, blah.</b><b>Now imagine they are going through that.</b><b>So it's just done an exam, 84%.</b><b>Yeah, right.</b><b>Here you go, Tommy.</b><b>These are the options for</b><b>you for your career and stuff.</b><b>And now that AI goes in and because it's</b><b>got memory, it's</b><b>learning and stuff like that.</b><b>It's only going to get better.</b><b>You know, there was a complaint that, oh</b><b>my gosh, we need so much more power.</b><b>We humans, we, yeah, I mean, the body,</b><b>the electrical system in</b><b>the body is a powerhouse.</b><b>The mitochondria is the</b><b>powerhouse of the self.</b><b>Whatever it is, right?</b><b>My American folks will remember that.</b><b>But like, the, you know, humans, how we</b><b>learn is very different.</b><b>You know, there was comparison.</b><b>You need so much training for the AI</b><b>versus like a toddler.</b><b>And in a way, it's a smart toddler, but</b><b>it learns in a different way to humans.</b><b>True.</b><b>And it was a complaint that we're going</b><b>to need much more power to get these</b><b>things to be cleverer</b><b>than they are now.</b><b>But if we're, what we're seeing now holds</b><b>true where, you know,</b><b>the models can get smaller.</b><b>They can be even running on like a crappy</b><b>laptop, like I've got there, 16 gig.</b><b>It's not GPU, but we'll be able to run</b><b>some stuff on here on your phones.</b><b>I think that will, that argument will</b><b>leave it out and it opens</b><b>up so many possibilities.</b><b>The thing to us to understand is that the</b><b>most of the power doesn't</b><b>necessarily come from its</b><b>learning.</b><b>The power comes from its usage.</b><b>So the more people you have sending</b><b>queries, sending</b><b>information, utilizing it, the more</b><b>power it's going to need to handle that</b><b>huge overload of access</b><b>and questions and triggering</b><b>and token and all that type of stuff.</b><b>That's what consumes the power.</b><b>It's the same thing on blockchain.</b><b>Like the power consumption comes from the</b><b>transfer of people,</b><b>you know, sending money</b><b>to and from things and validating it and</b><b>all the little checks and balances.</b><b>It's those little systems that do the</b><b>checks and balances.</b><b>That's what consumes all the power.</b><b>Take the, you know, that's so the more we</b><b>use it, the more</b><b>power it's going to need.</b><b>Yeah, you can run a model on your</b><b>computer, but if you had</b><b>50 people then signing into</b><b>computer to run that</b><b>model, it's going to fall apart.</b><b>That's very true.</b><b>It's a good point around the usage.</b><b>It's like if a blockchain platform falls</b><b>in a forest and no one's there to hear it</b><b>or tweet about it, did it happen?</b><b>Doesn't make a sound.</b><b>I don't know folks.</b><b>We could ask that.</b><b>That's another thing actually I want to</b><b>bring up just what we're</b><b>on this like weird kind of</b><b>topic.</b><b>But throwing up on the screen here, I'm</b><b>talking about AI mashups</b><b>and interesting things like I</b><b>used to do things where you can use the</b><b>AI and the semantic</b><b>reasoning, the semantic reasoning</b><b>where it can take abstract concepts.</b><b>It's very good at that actually.</b><b>Like anomaly detection is one that you</b><b>get from this getting</b><b>synthetic users where you</b><b>go.</b><b>Here's my user types and you give it all</b><b>the personas and then</b><b>it makes up personas for</b><b>those users and their behavior.</b><b>It's really great.</b><b>But even the mashups where you give it</b><b>two weak concepts like</b><b>explain blockchain as if</b><b>it was like the hip hop industry and it</b><b>will go out and do it.</b><b>And if that's never happened before on</b><b>the internet, you can check it.</b><b>It will actually give you</b><b>some reasonable results.</b><b>It'll make stuff up,</b><b>which is really weird.</b><b>But that's usually with words, right?</b><b>What I find interesting is there's one</b><b>where it was doing it with and you see,</b><b>we're seeing robots on</b><b>the screen here, folks.</b><b>And it was like training these AI robots,</b><b>like not real robots,</b><b>but like stick figure</b><b>type things with, well,</b><b>not quite stick figures.</b><b>They've been given bodies, but</b><b>essentially like</b><b>software looking like robots,</b><b>teaching them how to fight.</b><b>But from the perspective of dancing, and</b><b>we're just going to play</b><b>something here where you see</b><b>what effectively is breakdance fighting.</b><b>And they are running that I don't think</b><b>they're really hitting each other, Chris.</b><b>I don't know if gold or white is winning.</b><b>But they're certainly</b><b>really moving really fast.</b><b>It's a lot of like inbulsive moves.</b><b>I feel like I'm watching like a D grade</b><b>Hollywood fighting fight scene.</b><b>I would pay to watch this.</b><b>Some guys like going for the punches.</b><b>It's missed him by 20 miles.</b><b>He punches himself</b><b>because he's gone that hard.</b><b>It's very wildly clear.</b><b>He's falling over.</b><b>So these AIs, they're like</b><b>got ragdoll physics and stuff.</b><b>And they're basically just like, they've</b><b>been trained to try to hit each other.</b><b>But not doing so well.</b><b>Okay, there's a bit of a push there.</b><b>And it's just continuous like training.</b><b>I think they get up to let's see how well</b><b>the training goes further.</b><b>Gold has done a few hits and stuff, but</b><b>it's just really</b><b>interesting how active they are.</b><b>Boxing matches the future could be a</b><b>really interesting futuristic sport.</b><b>MMA, UFC, eat your heart out.</b><b>I think we just found</b><b>the title of the episode.</b><b>AI can fight dance.</b><b>AI will fight you with dance.</b><b>Fight dancing.</b><b>That's it.</b><b>It's not Zoolander break dance fighting,</b><b>but yeah, it's there.</b><b>So I thought that was interesting, man.</b><b>Although pretty close to Zoolander.</b><b>Yeah, it is.</b><b>It is.</b><b>What else have you</b><b>got in the news, Chris?</b><b>Bring it up, Jamie.</b><b>Yeah, so there are a couple of other</b><b>interesting things that</b><b>are happening in the market.</b><b>I'll race through these.</b><b>I won't linger on this</b><b>too much, but I'm really...</b><b>We talk a lot about LLMs and stuff and we</b><b>do without work and our daily lives,</b><b>but there's been a lot of</b><b>progress in how AI can...</b><b>It has been for a long time, but I think</b><b>these recent advancements have really</b><b>pushed and accelerated, I think a better</b><b>word, how they can be</b><b>applied to health, for example.</b><b>So University of Florida making huge</b><b>leaps in terms of</b><b>predictions for things like Parkinson's</b><b>disease, which is</b><b>really, really interesting.</b><b>And it's going to be super valuable for</b><b>globally in the health industry.</b><b>UK, they're now rolling out across</b><b>hospitals, testing out AI to</b><b>detect early stages of prostate</b><b>cancer as well and how they progress</b><b>through and how they are able, going</b><b>through procedures to</b><b>reduce the effects of prostate cancer and</b><b>hopefully get removed</b><b>for those who have it.</b><b>Will it be the robot doctor</b><b>that says Mr. Monford is relaxed?</b><b>Mr. Monford, yeah.</b><b>On goes the glove.</b><b>No, not to the other stand.</b><b>More around the detection</b><b>and scans and stuff like that.</b><b>And so that's a lot of leaps and bounds</b><b>due to stuff that has</b><b>been kicked off in the market</b><b>from the likes of OpenAI, et cetera.</b><b>That's really accelerated</b><b>use of AI in great industries.</b><b>On the flip side, we have some things in</b><b>the negative space</b><b>where we've got AI sports</b><b>bedding, which is a bit fearful, pretty</b><b>much like reducing, I guess...</b><b>Well, I mean, it's probably a better fit</b><b>for those who are</b><b>gambling, don't gamble, but</b><b>it's making bedding easier and reducing</b><b>the risk by increasing, I think almost,</b><b>what did I say here?</b><b>By 10% at the moment,</b><b>accuracy on results.</b><b>It's been tested since like 2018 where</b><b>they took a whole bunch of</b><b>results from things like Twitter</b><b>and or X, I should say, of people who are</b><b>submitting what their</b><b>bets were and finding</b><b>out the accuracies and understanding how</b><b>they got to those results.</b><b>And now, so in recent times, they're able</b><b>to take those insights and see what they can do.</b><b>And then they can look at the insights</b><b>that they've gathered</b><b>over the years and then make</b><b>predictions on results based on the data</b><b>for things like horse racing or football.</b><b>Interesting. And yeah, they've been...</b><b>I just had an idea for an app.</b><b>Yeah, right.</b><b>Use AI to predict your gambling habits.</b><b>Is it much different to...</b><b>The data has always been that, yeah, we</b><b>didn't have generative AI.</b><b>We've always had AI, there's machine</b><b>learning and probably people</b><b>were doing that, but it's a</b><b>higher barrier of entry.</b><b>Yeah.</b><b>So not many people were.</b><b>And so, generative AI is you can use</b><b>language to do the coding</b><b>and instruct the Python models</b><b>and SQL and whatever.</b><b>But is that...</b><b>I do question whether</b><b>it's a bad thing, really.</b><b>Maybe it introduces more people to it,</b><b>but I reckon like</b><b>everything else we see in other</b><b>industries, there'll be some sort of look</b><b>by the regulatory</b><b>bodies if you're doing that</b><b>sort of thing.</b><b>And that is happening.</b><b>Even in the States,</b><b>they've put a lot of...</b><b>There's a big lens over the top of the</b><b>legs of OpenAI and other AI models now.</b><b>Not necessarily because of things like</b><b>this, but just because</b><b>of how AI is impacting the</b><b>industry and trying to ensure that they</b><b>do have the right</b><b>legislations around Bruce.</b><b>They're going to have to make it.</b><b>So, psych's going to be the next big</b><b>thing to cleanse the market of this.</b><b>Oh, I wouldn't say necessarily cleanse.</b><b>I was going to say we can add to it.</b><b>No, but like, if the industry body...</b><b>What do you do with something like...</b><b>It's different for different industries,</b><b>but what do you do</b><b>something like that where it's</b><b>horse racing and other</b><b>kind of like betting.</b><b>Do you know how to</b><b>make the sports harder?</b><b>Sorry, guys, because AI, we're going to</b><b>have to raise the rim</b><b>to 12 feet, not 10 feet.</b><b>We need to introduce more anomalies or</b><b>more variants in the game.</b><b>Exactly.</b><b>The horse has to put its nose on a cone</b><b>and then spin around 10</b><b>times before it races.</b><b>Like, what are we going to do?</b><b>The only thing I could think they could</b><b>do is by limiting access to the data.</b><b>So, that's literally the only thing they</b><b>can do because the</b><b>patent recognition of LLMs</b><b>and partnering that with</b><b>mathematics, it will eventually...</b><b>It will minimize...</b><b>It'll increase that</b><b>accuracy of the bet that you make.</b><b>It won't never be 100% because it could</b><b>be even down to the</b><b>temperature of the air that particular</b><b>day and that could change your result of</b><b>something and it's not</b><b>going to know that to the 100th</b><b>percentile.</b><b>It's probably a bad example.</b><b>Maybe they do something where it's like,</b><b>right now we've got our normal...</b><b>Let's take horse racing and you've got</b><b>your bet on the horse and</b><b>you've got your trifectors</b><b>and whatever is the one where you get the</b><b>two and then</b><b>quinellas, which is I think...</b><b>I don't know, man, is that four?</b><b>I'm not a bet gambler, but in any case,</b><b>the different types of</b><b>bets that you can do,</b><b>I could maybe see</b><b>something where because...</b><b>Because a bell curve</b><b>where the majority see...</b><b>Back to the bell curve.</b><b>That don't make money.</b><b>It's the edges that make the proper money</b><b>and stuff like that.</b><b>If the bell curve moves towards more</b><b>people because they're</b><b>using tools like this to make</b><b>money on these bets, maybe some...</b><b>Because we can't...</b><b>We don't change the sport and I'm not</b><b>arguing that we should even</b><b>though it'd be really funny.</b><b>The only thing that we could change is</b><b>the possibilities of</b><b>what types of bets you could</b><b>have because I think even with the data,</b><b>even if they have a</b><b>moratorium that you can't get</b><b>new data, which is kind of silly because</b><b>what if someone is not</b><b>using AI, they're just using</b><b>stuff that was already allowed but had a</b><b>higher barrier eventually.</b><b>I think it's really hard there, but maybe</b><b>what they could do is</b><b>just change the types of bets</b><b>that you get a lot...</b><b>Not change it in banning it,</b><b>but there's a lower payout.</b><b>That's what they control.</b><b>I used to work in financial markets and</b><b>you have these big screens, multiple</b><b>screens for traders.</b><b>When you go to places like Sportsbed and</b><b>some of these other</b><b>places I've consulted to</b><b>in my data analytics days, they've got</b><b>massive screens and like...</b><b>Not massive screens,</b><b>but multiple screens.</b><b>Guess what?</b><b>They're watching not</b><b>the markets and charts.</b><b>They're watching races and sports and all</b><b>these things and they're setting the odds</b><b>and stuff like that.</b><b>It's fascinating.</b><b>I just see that maybe now that everyone</b><b>can win in that, it's</b><b>a bit more guaranteed.</b><b>Well, it's not a two</b><b>to one thing anymore.</b><b>It's a 1.1 to one thing.</b><b>That could be a way to do it.</b><b>It'd be fascinating though.</b><b>You do have a shift and this has been</b><b>happening pre-AI where</b><b>there's just be upper sports that</b><b>don't have the data.</b><b>Yeah.</b><b>Like there's some really</b><b>scary stuff where now...</b><b>Scary sports.</b><b>Scary sports where there were like these</b><b>underground bets on child sports.</b><b>So like your local basketball games or</b><b>your local football games in the schools.</b><b>Dang, what reminds me,</b><b>you owe me a bit of money.</b><b>Probably shouldn't talk about this.</b><b>Oh, sorry.</b><b>Sorry.</b><b>Okay.</b><b>Okay.</b><b>Yeah, no, I'm just joking.</b><b>But yeah, that's what</b><b>people are flocking to that.</b><b>And yeah, hang on.</b><b>Why am I seeing on these global</b><b>statistics for whatever,</b><b>like a school game or something</b><b>like that or like a</b><b>representative like youth kind of game?</b><b>But if you think about it on a</b><b>neurological level or</b><b>psychological level,</b><b>I'm not using the right words because I</b><b>am not a psychologist or a</b><b>health specialist at all.</b><b>That's a new product for psychological.</b><b>Yeah, exactly.</b><b>So why do people have</b><b>these gambling problems?</b><b>There is an addiction, right?</b><b>They're chasing that high.</b><b>And I guess if AI comes in, reduces that</b><b>accuracy of failure,</b><b>they don't get the high anymore because</b><b>it's almost an assured win.</b><b>So then they're going to end up either</b><b>not betting as much on this one.</b><b>They'll just shift to</b><b>other things like...</b><b>Yeah, I hated winning</b><b>all the time on the water.</b><b>Like I liked it when I was</b><b>losing in that uncertain...</b><b>No.</b><b>No, I see what you mean.</b><b>Like, I mean, it is something that we're</b><b>going to have to take into account.</b><b>And I think we take in going back to the</b><b>lessons, we will as an industry,</b><b>whether it's with the regulators</b><b>government, it's the</b><b>industry bodies and the people that</b><b>are part of it, we will be able to take</b><b>account of it a lot better if we</b><b>understand the limitations</b><b>and the truth behind it, because it's not</b><b>going to happen</b><b>because we're limited voices,</b><b>but we need more</b><b>people talking like this.</b><b>But I think we're going to have</b><b>misrepresentations as</b><b>we see all the time.</b><b>Exactly.</b><b>Anyway, you know, it's a fun fight.</b><b>Yeah.</b><b>We're just going to keep on doing it.</b><b>What else you got?</b><b>I think like I'm conscious of time.</b><b>I'm never conscious of time.</b><b>Never conscious of time.</b><b>I think it's been a good</b><b>wrap up there if you're...</b><b>But I think next time what we want to try</b><b>and get into is more around</b><b>productivity tips in</b><b>terms of application.</b><b>You've done some great examples of that.</b><b>I'm going to dive into some more</b><b>specifics within work,</b><b>life, to the tools that you can utilize.</b><b>Mental health stuff as well.</b><b>Mental health, physical health and work.</b><b>Yeah.</b><b>Burnout.</b><b>It should not be...</b><b>No, I'm just joking.</b><b>Well, there are some great</b><b>tools around that as well.</b><b>Yeah.</b><b>But I can talk about</b><b>this next time as well.</b><b>Absolutely.</b><b>Yes, anything you want to</b><b>say on the closeout before?</b><b>Word of advice, be good to your parents</b><b>and hug everyone that you know,</b><b>your loved ones and stuff, because we're</b><b>in a digital kind of future.</b><b>Who knows when the</b><b>next hug is going to be?</b><b>I'm just joking.</b><b>Or hug your AI just in</b><b>case it takes some burnout.</b><b>Exactly.</b><b>Be nice to your AI folks.</b><b>But no, I look forward to some</b><b>interesting events coming up.</b><b>We're going to be doing something that is</b><b>at the intersection of a few things.</b><b>The public kind of...</b><b>The public...</b><b>Not playground, but the...</b><b>Not marketplace, but the public square.</b><b>Let's call it that.</b><b>The public square for discourse.</b><b>So the Oz DeFi Association is partnering</b><b>with a group that is going to bring some</b><b>interesting international speakers here</b><b>to Australia that will</b><b>be talking about this.</b><b>It's all about democracy.</b><b>It's all about AI and blockchain and how</b><b>these things come together.</b><b>And so there'll be more news on that</b><b>coming out next week.</b><b>But yeah, definitely check that out.</b><b>It'll be here in Sydney and it is pretty</b><b>unprecedented getting these</b><b>folks across to our shores.</b><b>So yeah, we'll talk more about that.</b><b>So watch the space.</b><b>Sounds exciting.</b><b>Can't wait.</b><b>We'll put some links in the show notes.</b><b>Yeah, absolutely.</b><b>Beautiful.</b><b>On that note, hug your parents.</b><b>Bye.</b><b>Hey guys, have a beautiful time.</b>

People on this episode

Podcasts we love

Check out these other fine podcasts recommended by us, not an algorithm.

The Digital Village Show Artwork

The Digital Village Show

Digital Village