Digital Nexus

8 | AI is worse than humans!

Chris Sinclair and Mark Monfort Season 1 Episode 8

Join Chris and Mark in Episode 8 of the Digital Nexus Podcast as they catch up after a month-long hiatus, sharing their recent adventures and diving deep into the latest advancements in AI technology. From Amazon's investment in AI to the ethical implications of "deadbots," this episode covers a wide range of topics. Don't miss out on the fast news updates and shoutouts to the community!

Timestamps:

  • 00:07 - Introduction and Catch-Up
  • 01:34 - Chris's Adventures and New AI Projects
  • 03:06 - Mark's Trip to Vancouver and New York
  • 04:23 - Discussion on Risk-Taking in Innovation
  • 06:52 - Non-Committal Mentality in Sydney
  • 08:13 - Black Friday and Day of the Dead
  • 10:01 - News Updates: AI, Crypto, and More
  • 10:57 - Amazon's Investment in AI and Claude AI
  • 12:39 - Core Anthropic Updates and GitHub Integration
  • 18:08 - Ethical Implications of "Deadbots"
  • 20:38 - AI Regulations and Australia's Proposed AI Guardrails
  • 34:28 - Chain of Thought Processes in AI Models
  • 38:46 - Fast News: Tesla's AI Bot, Facebook's Controversy, Nvidia's Stock Market Loss
  • 48:31 - OpenAI's Financial Struggles and Capital Raise
  • 52:49 - Bing's Deepfake Scrubbing Tools and Google's New Photo Search
  • 55:26 - AI in Financial Transactions and Coinbase's AI Bots
  • 01:02:24 - Shoutouts and Community Engagement

Don't forget to like, comment, and subscribe for more updates on AI and technology!



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>I'm gonna trust it.</b><b>Let's go.</b><b>Let's go.</b><b>I do need a new laptop.</b><b>You've always told me anyway.</b><b>Episode eight, we're here.</b><b>What's going on?</b><b>It's been a while.</b><b>It's been a while.</b><b>Getting back, back again.</b><b>We are back.</b><b>J.D's back.</b><b>Tell a friend.</b><b>Hey.</b><b>You can't copyright.</b><b>If how you sing sounds like copyrighted</b><b>material, can they copyright that?</b><b>I don't think so.</b><b>Right?</b><b>We'll find out.</b><b>Unless your voice is really good.</b><b>Unless we use 11Labs to</b><b>actually sound like some shady.</b><b>Anyway.</b><b>Um.</b><b>Is it been two weeks?</b><b>Three weeks?</b><b>It's been over three, I think.</b><b>Almost four.</b><b>What's a month?</b><b>It's been a month.</b><b>How does that happen?</b><b>Oh, you've had an adventure.</b><b>I've had an adventure.</b><b>You've had an adventure.</b><b>We've both had some adventures over the</b><b>last month and it's...</b><b>But good adventures.</b><b>To adventures and to...</b><b>What is this?</b><b>You've given us some brain power.</b><b>Focus.</b><b>Brain power.</b><b>Ginger shots.</b><b>Cheers.</b><b>Callumancy.</b><b>Is like um, kumquats.</b><b>Which uh...</b><b>Yeah.</b><b>Little limes.</b><b>Little hits of vitamin C.</b><b>Callumancy is Filipino.</b><b>I love you.</b><b>Yeah, that's my Filipino friends.</b><b>And family, etc.</b><b>Thank you very much.</b><b>Oh my god.</b><b>What else to talk about?</b><b>Jesus.</b><b>Uh, where do we start?</b><b>How are you?</b><b>I'm good.</b><b>Okay.</b><b>I'm very good.</b><b>Episode 8.</b><b>I've been busy.</b><b>Work has taken up a lot of time.</b><b>I've been working with businesses on integrating.</b><b>Fantastic.</b><b>And a lot of use of AI in that process</b><b>which has been very exciting.</b><b>A lot of...</b><b>So, it was more than usual.</b><b>More than usual.</b><b>So, it's more of a massive proof case of</b><b>AI in product</b><b>development in general which is</b><b>a lot of stuff I've been talking about</b><b>and working with</b><b>businesses on integrating.</b><b>Well, Chris is the king of</b><b>product development guys.</b><b>Like, you would not have the OzDFI</b><b>Association not centralized</b><b>as it was and just other things</b><b>without this man's input.</b><b>And obviously other companies that you</b><b>would have heard of.</b><b>I hope so.</b><b>Yeah, yeah.</b><b>So yeah, it's been good.</b><b>Just like filling up my time which is</b><b>made up doing a lot of things I love.</b><b>Obviously more difficult but you know,</b><b>focusing on family and</b><b>stuff at the same time.</b><b>Just go back from the Gold Coast</b><b>adventure over there.</b><b>Why is it Brisbane Vegas</b><b>but Gold Coast is more Vegas?</b><b>Or is it more Miami?</b><b>Oh no, definitely it's</b><b>very Vegas-y when you...</b><b>Yeah. There's rides everywhere, rides all over</b><b>the place, idiots</b><b>falling over themselves on the</b><b>road getting complained to police.</b><b>Yeah. You know, it's...</b><b>And what else did your family do?</b><b>That was just me.</b><b>I was going to say that was just you.</b><b>That was my man, that was just me.</b><b>That was...</b><b>We were...</b><b>Yeah, they were back in the hotel.</b><b>They're probably normal compared to you.</b><b>It's my daughter just</b><b>like, "Where is daddy?"</b><b>No.</b><b>Fair enough.</b><b>How about yourself man?</b><b>You've had even more of an</b><b>adventure than I could even...</b><b>It's been an adventure because it feels</b><b>like it was longer than</b><b>two weeks that I was gone.</b><b>But it wasn't really, it was just like</b><b>the two weeks and we</b><b>were in Vancouver to start</b><b>off with, meeting the</b><b>rest of the psych team.</b><b>Yeah.</b><b>Shout outs to Mark,</b><b>Ivan, Tim, Ben over there.</b><b>But yeah, we were over</b><b>there, we had to meet the team.</b><b>We've met Mark because</b><b>he's an expat Aussie.</b><b>And Aussie with more</b><b>of a Canadian twang now.</b><b>Mark will appreciate that.</b><b>The influence.</b><b>Hard Rs.</b><b>Arrr.</b><b>You know, it's not like...</b><b>So he's a pirate, like Canadian.</b><b>No, yeah, yeah, yeah.</b><b>He's a pirate.</b><b>International waters, I always think.</b><b>No, we're meeting a team and it's really</b><b>important in hybrid</b><b>kind of working models.</b><b>You might have superstars all across the</b><b>board, but until you come</b><b>together, you don't really</b><b>get to know the other</b><b>people and stuff there.</b><b>So it was really interesting to do that,</b><b>to do a lot of the</b><b>brainstorming, pushing us</b><b>into areas of focus.</b><b>And we've hit on that now around how we</b><b>tackle workflows and</b><b>something we're calling the</b><b>layer two of generative AI.</b><b>But, you know, meeting the team was</b><b>really important, seeing</b><b>where they live, how they</b><b>do things, what it was like over there.</b><b>And more importantly,</b><b>seeing clients and investors.</b><b>And so on the client side, it was just...</b><b>And sorry, we were in New York as well.</b><b>On the client side, it's just, sorry,</b><b>Australia, we are not...</b><b>And it's for a good reason, right?</b><b>We do not take much risk.</b><b>And because we don't take much risk and</b><b>we don't have this</b><b>culture of like the interest</b><b>in that.</b><b>Like it is...</b><b>Corporate entities in</b><b>particular are very risk averse.</b><b>Which is crazy because we</b><b>end up getting left behind.</b><b>We're protected when bad things happen.</b><b>But for the 20...</b><b>Like we...</b><b>Here's the thing, we're the lucky</b><b>country, 30 years of no</b><b>recession and stuff like that.</b><b>If you've got 30 years of no recession,</b><b>but you're not taking</b><b>much upside risk, you're</b><b>doing yourself a disservice.</b><b>It's like keeping your money in savings</b><b>only and not actually</b><b>investing it while everything</b><b>else inflates and you're</b><b>actually going backwards.</b><b>That's what it feels</b><b>like with innovation here.</b><b>You just have to look at manufacturing,</b><b>mining, government decisions.</b><b>All of that stuff is very behind times.</b><b>We are very good at</b><b>digging stuff out of the ground.</b><b>And I can't remember</b><b>who put this paper out.</b><b>I think it might have been referred to by</b><b>some folks on LinkedIn.</b><b>I'll do the shout out some other time.</b><b>But interesting point was that we're so</b><b>good at digging stuff out of the ground.</b><b>One of the most important things when it</b><b>comes to AI right now is</b><b>the stuff that comes out</b><b>of the ground when it comes to the</b><b>minerals and precious</b><b>metals that are needed to go</b><b>into these chips.</b><b>It's like, "Duh, no brainer here guys.</b><b>What are we doing?"</b><b>So it's interesting.</b><b>But on that point, we</b><b>don't take much risk here.</b><b>They do.</b><b>We protect our downside so we don't get</b><b>hit by bad things when bad things happen.</b><b>We're protected.</b><b>But we just miss out on the years and</b><b>years of glory that we</b><b>could because we don't encourage</b><b>businesses.</b><b>And we don't do much in</b><b>terms of the investment.</b><b>The government will always argue that,</b><b>"Yeah, we're doing all</b><b>this investment and stuff."</b><b>It's like, "Yeah, but</b><b>others do way more."</b><b>And it's not just the government.</b><b>There's a whole ecosystem out there.</b><b>Places like Stone and Chalk and other</b><b>startup hubs are</b><b>definitely trying, but it's still</b><b>not enough.</b><b>Because when you go that...</b><b>Take, for example, we did a</b><b>last minute trip to New York.</b><b>We weren't even looking at it.</b><b>We were just there and we'll see which</b><b>cities we go to, open</b><b>ended ticket kind of thing.</b><b>San Fran got shut down for us because it</b><b>didn't get the kind of</b><b>meetings that we wanted to</b><b>do there.</b><b>We'll go back another time.</b><b>We've got our North</b><b>American partners up there anyway.</b><b>But when you do a last minute trip in</b><b>Australia, if you're</b><b>telling someone two days before,</b><b>"Hey, I'll be in Sydney," and it's like,</b><b>chances are, fat chance</b><b>you're actually going to get</b><b>someone to say yes to anything.</b><b>Over there, so many people</b><b>would be like, "Yeah, let's go.</b><b>Let's meet."</b><b>Even on the weekend, I met</b><b>with people to talk business.</b><b>This is the thing.</b><b>It's just so much</b><b>hustle and growth over there.</b><b>I find it's even...</b><b>There's big things particularly in Sydney</b><b>and this is stepping</b><b>away from obviously the</b><b>big investment corporate thing.</b><b>But more around even</b><b>just the people in general.</b><b>No one commits in Sydney.</b><b>I remember working for companies back in</b><b>the day and you'd have</b><b>these special big events</b><b>that whether it's in client invites or</b><b>just general invites,</b><b>doing meetups, webinars,</b><b>all that type of stuff.</b><b>People will say yes to coming and then</b><b>maybe 40, 50% of people</b><b>will show up in Sydney.</b><b>Because every time I ran the same thing</b><b>in Melbourne or</b><b>Brisbane, 80% of people will</b><b>show up.</b><b>There was always in Sydney, there's this</b><b>non-committal mentality</b><b>in corporates and entities.</b><b>Maybe it's because you have a reputation</b><b>and stuff in Sydney.</b><b>People know about you and</b><b>stuff and just like, "Nah."</b><b>No, I'm just joking.</b><b>But you're right with</b><b>free events and stuff.</b><b>It is...</b><b>There weren't many events.</b><b>It's like a 50%...</b><b>We'll say it's your events.</b><b>50% kind of turnout rate when events are</b><b>free and we see that</b><b>with like OzD file, like 50,</b><b>60%.</b><b>You make people pay a little bit and</b><b>stuff like say five bucks</b><b>and then go into an event</b><b>where we help run downstairs</b><b>and you had 90% turnout rate.</b><b>It's interesting like the</b><b>committal that you have here.</b><b>And speaking of which, I mean, we're</b><b>starting off like kind of...</b><b>We're starting off on the negative side</b><b>and I guess it kind of</b><b>makes sense because we'll</b><b>lead off with a positive.</b><b>We'll finish off there.</b><b>But we are in black</b><b>because it is Black Friday.</b><b>It's Friday the 13th and</b><b>I think it's important.</b><b>I'm actually dead.</b><b>This is...</b><b>Exactly.</b><b>This is...</b><b>My pasty skin is a representation of...</b><b>What is the Day of the Dead?</b><b>Dios de la Muerta or</b><b>something in Spanish?</b><b>Yeah, Day of the Dead.</b><b>That's what today is folks.</b><b>And luckily there is going to be a</b><b>resurrection because...</b><b>Well, put it this way, you need people to</b><b>go overseas, come back.</b><b>We need people from</b><b>overseas to come here.</b><b>Australia is a great place to actually</b><b>invest in because others aren't doing it.</b><b>That's the silver lining there.</b><b>Yeah.</b><b>They're all moving overseas.</b><b>They moved to the States.</b><b>They moved to Europe because that's where</b><b>the opportunities are.</b><b>People will take their businesses...</b><b>They'll take their businesses over there</b><b>and we'll get more</b><b>businesses out of there just</b><b>by sheer numbers.</b><b>We only got less than</b><b>30 million people here.</b><b>So sheer numbers, we</b><b>have to go over there.</b><b>But the FOMO will happen when people see</b><b>that, "Oh, so-and-so</b><b>businesses over there, how</b><b>do we do that kind of stuff here?"</b><b>There will be FOMO and this is just the</b><b>formula unfortunately.</b><b>But at the same time, we live here.</b><b>We're still going to be</b><b>doing a lot of stuff here.</b><b>We can take a lot of ideas of what's good</b><b>happening over there.</b><b>Bring it to places here, but we need to</b><b>have that constant</b><b>connection because you get energy</b><b>from that hustle that's over there.</b><b>Hustle and bustle.</b><b>That was good and it meant that I did not</b><b>sleep on the way back</b><b>because the brain was</b><b>just firing on all cylinders and stuff.</b><b>So 24 hours to get back.</b><b>The first day I got back, it</b><b>was like 6am in the morning.</b><b>I'm like, "I'm just going to sleep for</b><b>the rest of the day."</b><b>Then I slept till afternoon, was up for a</b><b>little bit, still tired and then just got</b><b>back into normal routine.</b><b>The jet lag wasn't too bad.</b><b>Nice.</b><b>But yeah, I walked back into just so much</b><b>stuff going on here in local and whatnot.</b><b>We've got a lot to cover in the news.</b><b>Oh my goodness.</b><b>I mean, we've been away for four weeks.</b><b>There is a lot that</b><b>has happened in the news.</b><b>In relation to AI, crypto, all the works.</b><b>It is, shit's happened.</b><b>Some bad things, some good things, some</b><b>interesting, some exciting things.</b><b>And we move forward.</b><b>What do you want to raise first?</b><b>Because I've got a few.</b><b>I'm going to raise something that I</b><b>thought was hilarious</b><b>that I just bumped into when</b><b>I was on scanning my way through.</b><b>Scamming as well.</b><b>I mean scanning.</b><b>My way through the old TikTokaroos.</b><b>It was, I need to find it now.</b><b>Was it eat the dogs, eat the cats?</b><b>Eat the dogs, eat the cats.</b><b>Eat the good old Donald</b><b>Trump and his recent events.</b><b>They're eating the dogs,</b><b>they're eating the cats.</b><b>Yeah.</b><b>I love this.</b><b>That's AI used at its finest.</b><b>Love it.</b><b>I think let's start off</b><b>with kicking off around.</b><b>Talks about Amazon.</b><b>Like.</b><b>Oh yeah.</b><b>Amazon is, you know.</b><b>Doing a lot in the AI space.</b><b>We all have a love-hate</b><b>relationship with them.</b><b>Big corporate ticket over the world.</b><b>I love it if they sponsor.</b><b>Eating, hurting the underdog.</b><b>Love it when they sponsor.</b><b>We all obviously use their products and</b><b>purchase their stuff</b><b>despite, that's the love-hate</b><b>relationship we have with them.</b><b>Exactly.</b><b>We've got Amazon Prime.</b><b>Yep.</b><b>I love the delivery stuff.</b><b>I love the shows they've got.</b><b>They've got arguably better quality stuff</b><b>than some of the</b><b>other free-to-air, oh it's</b><b>free-to-air, pay TV things that I get.</b><b>That's so true.</b><b>But they have a huge investment in AI.</b><b>Yeah.</b><b>And they're doing a lot of work in it.</b><b>And particularly with Claude or Anthropic</b><b>being one of their biggest investments.</b><b>And they're looking to roll that out into</b><b>all of their Alexa products.</b><b>Oh really?</b><b>Which is quite interesting.</b><b>Oh interesting.</b><b>I'll just read about that.</b><b>Yeah, I definitely should.</b><b>There's a lot of great merit to it.</b><b>It gives more interesting results in some</b><b>areas than just like chat GBT.</b><b>Although I still find chat GBT is better</b><b>at it when it comes to other things.</b><b>And Claude hits limits sooner</b><b>than you have with chat GBT.</b><b>At the moment, and maybe that changes,</b><b>but because he hit</b><b>those error limits sooner,</b><b>it's like early days of chat GBT where</b><b>you couldn't just have</b><b>a long discussion with</b><b>it where you needed to</b><b>do discussion type stuff.</b><b>That's right.</b><b>Because Amazon is obviously investing a</b><b>lot in cloud based</b><b>support for AI tools as well.</b><b>Cloud or Claude?</b><b>Cloud.</b><b>Cloud.</b><b>Cloud systems.</b><b>Cloud and they invested in cloud.</b><b>Claude is obviously a big player now, but</b><b>they're with their cloud tech.</b><b>And so this is an</b><b>interesting move for them.</b><b>And given that they have got such a big</b><b>investment in Anthropic,</b><b>it makes a lot of sense.</b><b>But the reason I wanted to start with</b><b>that is actually ease my</b><b>way into Core and Anthropic.</b><b>Because they are rolling out quite a few,</b><b>they've rolled out</b><b>quite a few updates over</b><b>the last four or so weeks in relation to</b><b>enterprise level services</b><b>and support for people doing</b><b>a lot of development and applications.</b><b>Everything from the sort of the cloud for</b><b>enterprise just rolled</b><b>out with over 500,000</b><b>context windows and context windows for</b><b>anyone who doesn't know is</b><b>it's like the textual range</b><b>around the target tokens that a large LLM</b><b>might be using to</b><b>process the time it takes</b><b>to process that</b><b>information and generate responses.</b><b>So they're rolling that out for</b><b>enterprise now, which is</b><b>part of the one of the most</b><b>interesting parts of that though is their</b><b>ability to plug that</b><b>into GitHub directly.</b><b>So let's say you have your website</b><b>applications platforms all properly set</b><b>up and hosted within</b><b>GitHub.</b><b>You can now add Claude as a</b><b>layer in that and utilize it.</b><b>Yeah, please scroll through.</b><b>And utilize it within your own systems to</b><b>generate improve</b><b>optimize code within your</b><b>own platforms.</b><b>Along with a lot of other myriad of stuff</b><b>you can do with the enterprise tools that</b><b>it has available.</b><b>But this I thought was most interesting</b><b>because it has a lot of</b><b>I mean, Claude itself was</b><b>already really good at code generation.</b><b>Yeah.</b><b>You know, kicking things off in</b><b>application formats, websites, etc.</b><b>And now adding that layer into GitHub,</b><b>even though GitHub does</b><b>already have some AR plugins,</b><b>I thought was was really,</b><b>really, it was really cool.</b><b>I'll show you another one just on that</b><b>note of, you know, doing</b><b>stuff here, where I don't</b><b>know if you've heard of a cell before,</b><b>but they do like AI</b><b>driven, oh, they do websites,</b><b>right?</b><b>And you can start off with</b><b>like some of their templates.</b><b>And you can do things like use one of the</b><b>templates to create a</b><b>chat bot and stuff like</b><b>that.</b><b>You can do a whole lot of other like</b><b>hosting things with them.</b><b>And interestingly, they created this</b><b>thing where basically, it's AI driven.</b><b>And what they can do is have a way for</b><b>you to describe what's the</b><b>layout for like a website</b><b>or what's the layout for for something</b><b>that you want to create.</b><b>Now I'm just taking a</b><b>screenshot here on another screen.</b><b>So don't mind me, but I'm taking a</b><b>screenshot of something</b><b>that, you know, you still look</b><b>look at in my finance days.</b><b>And I'll go here, make a similar.</b><b>It's very, it's very</b><b>fast here on my computer.</b><b>This is why I need an update, Chris, to</b><b>this and that I just paste it in.</b><b>Right?</b><b>So you'll see what it is.</b><b>It's basically Google finance.</b><b>And what's interesting here with Vercel's</b><b>dev chat thing is</b><b>that it's like anthropic</b><b>cloud, it'll start to build it on the</b><b>side, as you can see</b><b>coming up here, and then it'll</b><b>get it'll open up a preview.</b><b>So if we go, okay,</b><b>preview mode, here we go.</b><b>Oh, something went wrong.</b><b>It'll, it'll edit it.</b><b>It'll fix it.</b><b>But take a look at this.</b><b>It's building the website, very similar</b><b>to Google finance in real time.</b><b>Oh, something went wrong.</b><b>It'll come back again.</b><b>So look, it'll just fix these things.</b><b>But this is literally off a screenshot.</b><b>You can then get the code, take that and</b><b>actually do something else with it.</b><b>So it's super clever</b><b>how you do things now.</b><b>And it gets to that point where a lot of</b><b>businesses that we talk to</b><b>in, you know, our day to day</b><b>of psych and stuff, they're talking about</b><b>how do they take back</b><b>control of things that</b><b>they've had to just</b><b>outsource for the longest time.</b><b>They've had to outsource that</b><b>they've had to get others in.</b><b>A lot of that you still need to outsource</b><b>it because you still</b><b>need others to be the</b><b>experts there.</b><b>But many things you can bring in house</b><b>and what happens to</b><b>those like consultants and</b><b>the ones that are doing on the outside,</b><b>it's kind of like,</b><b>well, they'll be there, say,</b><b>for the validation because you need their</b><b>stamp of approval that</b><b>you've done the right</b><b>kind of process, but the relationship of</b><b>engagement changes and the</b><b>consultants, and we still</b><b>do some consulting, we have to change how</b><b>we engage in this kind of future.</b><b>We have to be the ones bringing this to</b><b>them before the company</b><b>is doing if the companies</b><b>are you need to be the one</b><b>helping them out with this.</b><b>It's new tool.</b><b>It's just democratized in</b><b>terms of who accesses it.</b><b>But it stems from what you're saying</b><b>there with Claude and</b><b>the usefulness there.</b><b>This was something that was fascinating</b><b>was coming across my desk</b><b>and expanding on Claude.</b><b>What they're also doing is releasing a</b><b>whole bunch of quick star</b><b>repos now within GitHub.</b><b>So you can actually go to the GitHub,</b><b>Anthropics quick start up,</b><b>got a link in the thing, pop</b><b>it up.</b><b>And you can actually they've got a</b><b>collection of projects</b><b>designed to help developers quickly</b><b>get started with building and deployable</b><b>applications using anthropic APIs.</b><b>So just really supporting the startups</b><b>and people with</b><b>interesting ideas getting out</b><b>out there as well.</b><b>Where is that?</b><b>Okay, yeah, I found it.</b><b>I found it.</b><b>Nice one.</b><b>Let's take a look at that.</b><b>I'll just throw it up quickly.</b><b>Yeah, throw it up.</b><b>Throw it up there, please.</b><b>Throw it up there, Jamie.</b><b>I'm still not as good as this.</b><b>And look, guys, it will be faster on a</b><b>better machine if</b><b>anyone's got recommendations as</b><b>to what laptop I should get.</b><b>I'm currently running</b><b>a 2021 HP Spectre x360.</b><b>I bought it because it looked nice.</b><b>I do want something with 32 gig.</b><b>16 gig of RAM is great.</b><b>I do want to avoid the Snapdragon chip,</b><b>because what I've been</b><b>told is that it just doesn't</b><b>spends what you're doing.</b><b>I mean, there's the Snapdragon chip.</b><b>They don't have any dedicated GPUs at</b><b>this particular point.</b><b>So if you're doing a lot of AI work or</b><b>you've been gaming, it just</b><b>doesn't have the performance.</b><b>If you're just doing</b><b>general work, amazing.</b><b>No, but the other thing is</b><b>like video editing as well.</b><b>Yeah, exactly.</b><b>It's going to be a massive</b><b>problem for that kind of stuff.</b><b>Right.</b><b>So just going AMD or Intel</b><b>and stuff for the time being.</b><b>Anyway, so I found the thing</b><b>I wanted to also talk about.</b><b>Have you heard of dead bots?</b><b>Dead bots, I think it's called?</b><b>No.</b><b>What is that?</b><b>There's a company out of, and it's a</b><b>shout out to one of my</b><b>favorite Tiktokers out here,</b><b>Dylan Page with News Daddy, my man, was</b><b>just scrolling through</b><b>some items from today.</b><b>And he mentioned, well, he brought up</b><b>something which is really interesting.</b><b>It's this company out in</b><b>China, which is creating an AI.</b><b>Anyone's heard of character AI?</b><b>You can create those</b><b>bots with characters.</b><b>Yep.</b><b>Imagine you can do</b><b>that for a lost loved one.</b><b>Oh yeah.</b><b>I've heard of stuff like this before.</b><b>Bringing your dead ones back to life.</b><b>That's right.</b><b>Something I see the newer generation</b><b>going in for, but the</b><b>older generation being like</b><b>even just seeing faces move on old photos</b><b>like you could do with some basic AI.</b><b>No.</b><b>So essentially what you can do is you can</b><b>send in a history of</b><b>videos, images of your</b><b>loved ones deceased and they will</b><b>generate a bot from this</b><b>particular AI, essentially</b><b>an AI from this information.</b><b>Oh wow.</b><b>You can put it against your tombstones or</b><b>on the graves or you can just keep it for</b><b>yourself as a photo</b><b>or something like that.</b><b>And it essentially communicates to you as</b><b>if it was that lost one.</b><b>And there was an example that was used</b><b>where this person who had</b><b>done it was engaging with</b><b>the AI tool and the AI tool was like, oh,</b><b>I don't feel any pain.</b><b>Trying to demonstrate this AI level</b><b>emotion to the experiences</b><b>that it was having at the</b><b>current time.</b><b>And I just was like in my head, it's not</b><b>a place I had gone to</b><b>yet with all the different</b><b>things you could do with AI.</b><b>Someone has.</b><b>Bringing my lost loved ones to life.</b><b>I mean, I can imagine some really good</b><b>places for it to be</b><b>applicable like depression.</b><b>Maybe at the funeral.</b><b>People who need more closure.</b><b>Yeah.</b><b>I can imagine that</b><b>would work really nicely.</b><b>But I think that it maybe works after a</b><b>period of grievance for</b><b>certain things like maybe</b><b>historical figures that</b><b>agree to this kind of stuff.</b><b>There were already like talks about this</b><b>and other companies doing it.</b><b>I guess this is like one of them.</b><b>I wonder where they are.</b><b>But it was like places trying to tie</b><b>things into funeral homes</b><b>at the early onset of chat</b><b>GPT.</b><b>So obviously they</b><b>would have gotten better.</b><b>This is interesting too.</b><b>But man, dead bots.</b><b>I actually thought dead</b><b>bots meant something else.</b><b>But D.S. Del Amorte, it is the perfect</b><b>day to be talking</b><b>about that kind of stuff.</b><b>That's right.</b><b>That is that is freaking crazy.</b><b>That's why I wanted to bring it up.</b><b>This is Friday 13th</b><b>live in live and talking.</b><b>We should have had a</b><b>different version of the cover.</b><b>Don't worry folks.</b><b>We'll get a different AI version of these</b><b>guys because we can</b><b>throw this into like chat</b><b>GPT and go reimagine this.</b><b>But with you know,</b><b>yeah, with dead people.</b><b>We'll be skeletons.</b><b>But before we go on to some other stuff I</b><b>want to raise maybe it's at the end.</b><b>But I think it's important now because</b><b>you know people have</b><b>short attention spans.</b><b>This will be something that maybe we even</b><b>cut up into something.</b><b>I know we're going to do we've got an</b><b>automated way that the shorts get cut.</b><b>So so you know</b><b>definitely interesting there.</b><b>But Australia is looking to follow the EU</b><b>as well as California</b><b>in terms of regulating</b><b>AI.</b><b>And so they've got something called the</b><b>AI guardrails, which</b><b>I'll just bring up now as</b><b>you can see here.</b><b>So these are 10 commandments taken down</b><b>from the mountains next</b><b>to a fiery bush I believe.</b><b>Do you hear something folks?</b><b>And the first commandment is establish</b><b>implement and publish an</b><b>accountability process including</b><b>governance, blah, blah, blah, just have</b><b>governance right how you're</b><b>going to use AI in business.</b><b>Now this this is a mix of mandatory plus</b><b>proposed like standards</b><b>or ways that businesses</b><b>and you would think that higher level</b><b>more high risk</b><b>businesses it would be mandatory</b><b>that they have to do this.</b><b>And you think that the smaller ones, you</b><b>know, associations, etc.</b><b>Maybe it's more like guidelines.</b><b>The point is one big point is this too</b><b>early for us to actually</b><b>have legislation around</b><b>this kind of stuff because yes, while we</b><b>hear a lot about the</b><b>negative, there is so</b><b>much good stuff happening when it comes</b><b>to AI and we're still</b><b>just scratching the surface.</b><b>This is almost like can you imagine if</b><b>well, it'd be very hard</b><b>to imagine because this</b><b>is not what happened, but they tried to</b><b>regulate the internet,</b><b>they weren't successful.</b><b>And when I say they I mean</b><b>the powers of governments, etc.</b><b>And because of that, we were able to have</b><b>this really open</b><b>internet like we have today.</b><b>And it was more built around standards.</b><b>Then each country got to apply those</b><b>standards and there's</b><b>laws at those country levels.</b><b>Sure, we've got laws around the internet,</b><b>but they weren't there</b><b>early and we were able to</b><b>develop and yeah, there were bad things,</b><b>but we also had so much</b><b>good that came out of it.</b><b>If we regulate too early, one thought is</b><b>that we are going to</b><b>stifle innovation anyway.</b><b>And that's the biggest complaint about</b><b>the EU's AI Act because</b><b>it tries to regulate the</b><b>technology.</b><b>It has rules in there like if your model</b><b>is over a certain size,</b><b>that's like saying back in</b><b>the 80s and having it</b><b>as a rule to this day.</b><b>Well, if your hard disk, your floppy disk</b><b>is over four megabytes</b><b>and it's like that was</b><b>big back then, but now that's like, what</b><b>are you talking about?</b><b>Songs aren't like 20 seconds of a song</b><b>isn't even four megabyte.</b><b>You know, like they're really hard on the</b><b>tech, but four megabytes was quite large.</b><b>Huge.</b><b>That was a computer that was like, we</b><b>said men to Mars floppy</b><b>disks were like in the kilobytes.</b><b>Exactly.</b><b>Do you floppy disk or do</b><b>you mean the save icon guys?</b><b>Right.</b><b>Because it's turned into that.</b><b>It's gone from like when you ask people</b><b>that are younger, like</b><b>how do you hold a phone?</b><b>They do this.</b><b>They do the fingers like that.</b><b>And then like younger people, they go,</b><b>how do you hold a phone?</b><b>They do like that flat hand.</b><b>Do you notice they don't, you really use</b><b>the save icon anymore?</b><b>No.</b><b>Yeah.</b><b>That icon is pretty much</b><b>being removed from a lot of.</b><b>Why would you save stuff?</b><b>Why isn't it just auto saving?</b><b>Why are you using Google Docs?</b><b>But more the people</b><b>don't know what the icon is.</b><b>Exactly.</b><b>So that they're getting rid of it.</b><b>Yeah.</b><b>Yeah.</b><b>The floppy disk icon.</b><b>It's gone.</b><b>So anyway, the point around that the</b><b>guardrails is, it's interesting that</b><b>they're putting it out</b><b>now.</b><b>There is a consultation and various</b><b>groups of individuals are</b><b>responding as well as actual</b><b>communities themselves.</b><b>So for example, the build club is I'm</b><b>putting it to Michael and</b><b>the guys at DSAI to put out</b><b>a response.</b><b>And we know there's going to be various</b><b>industry bodies,</b><b>lawyers, et cetera, et cetera,</b><b>consultancies and whatnot.</b><b>So it'd be interesting to see what people</b><b>have to say about it.</b><b>I feel that we do need to have some</b><b>guardrails, but they need to be a bit</b><b>more looser and maybe</b><b>even like a moratorium.</b><b>It will still be years before if they go</b><b>through the process and</b><b>it's even fast tracked, it</b><b>will still be two or three years.</b><b>Why do you think they need to be looser?</b><b>What's so strict about them right now?</b><b>I'm thinking about what's happening.</b><b>The biggest strict, the biggest</b><b>strictness is if they do it</b><b>and depends on the application.</b><b>This is the first draft.</b><b>They'll always like fine tune it after</b><b>like they get feedback from people.</b><b>But I think the thing that they need to</b><b>think about is biggest thing.</b><b>It should be guardrails and regulations</b><b>around the applications,</b><b>not the technology itself.</b><b>And they need to be careful around</b><b>pushing onus on developers of the tools,</b><b>unless the tools are</b><b>actually purposely doing deepfake porn</b><b>and stuff like that.</b><b>You know, things we should guard against.</b><b>If it's just a general tool and depends</b><b>on how someone uses it, right?</b><b>So it's like regulating math just because</b><b>hackers use math in the</b><b>Python code that they write or</b><b>regulating Python, the app, because</b><b>hackers use it to create</b><b>viruses and stuff like that.</b><b>We don't regulate the app.</b><b>So we don't regulate the actual tool.</b><b>We regulate the outcome,</b><b>which is the applications.</b><b>That's what should be regulated.</b><b>I think math is the</b><b>wrong thing to regulate.</b><b>You'd be regulating Python because that's</b><b>the thing that creates code.</b><b>But it's almost the same kind</b><b>of argument and stuff, right?</b><b>Like by arguing that, hey, we should</b><b>regulate the AI models</b><b>and stuff and what they do.</b><b>But that's almost like</b><b>that's just the math.</b><b>It's how it's applied.</b><b>It's how you use it, folks.</b><b>Like it's not the side, blah, blah, blah.</b><b>But you need to--</b><b>Sure.</b><b>I mean, the same argument goes against</b><b>the social media platforms.</b><b>They create these platforms for people to</b><b>access and then fake news,</b><b>terrible content, sharing of really dark,</b><b>awful stuff that is</b><b>mostly only relevant for</b><b>Friday 13th material.</b><b>Yep.</b><b>Yeah.</b><b>It's end up putting out there.</b><b>But it's the same thing</b><b>from an AI perspective, right?</b><b>You have these tools that have the</b><b>capabilities to do</b><b>some really dark stuff.</b><b>So they should be regulations around how</b><b>you monitor and protect people from that.</b><b>And bad use of it, but don't</b><b>stop it from going to market.</b><b>So the thing that I'm also saying is that</b><b>it's not necessarily like a free for all</b><b>that we just should have no regulation.</b><b>It's somewhere in the middle.</b><b>And part of that middle, when you look at</b><b>the other AI guardrails,</b><b>so one thing that they talk about here</b><b>and need for humans in the loop.</b><b>So they talk about human control and</b><b>intervention in number five.</b><b>And then at the same time, they talk</b><b>about the need to keep records of this.</b><b>Why these guardrails have really been</b><b>interesting is because</b><b>since late last year,</b><b>when we developed Psyche and we developed</b><b>it with our mindset</b><b>of blockchain involved,</b><b>we saw that the we guessed that the</b><b>future would be that we</b><b>will need at some point</b><b>to prove that humans are in the loop and</b><b>not just for business,</b><b>workflow purposes, but for insurance, for</b><b>compliance, for audit.</b><b>And this is what's actually going to</b><b>probably happen for high-risk businesses.</b><b>You're going to need to prove for</b><b>insurance compliance or audit</b><b>that humans are in the loop.</b><b>The problem that I have with this though,</b><b>and where I think it</b><b>could be improved is that</b><b>they're putting the onus on businesses to</b><b>keep records of all this kind of stuff.</b><b>Businesses keep records and I have to</b><b>prove to you as an auditor</b><b>by sending you my</b><b>data to show compliance.</b><b>A, how do you know it</b><b>in tamper with that?</b><b>Because it's just log</b><b>files out of my system.</b><b>B, how do you know that</b><b>if I'm moving files across,</b><b>that's a cybersecurity risk</b><b>depending on what's being sent.</b><b>So what I'm saying here is that there are</b><b>ways to prove humans in the loop like</b><b>we've got with Psyche.</b><b>I was going to say, no,</b><b>no, no, I was going to say,</b><b>like, isn't there a</b><b>company under that's doing this?</b><b>Yeah.</b><b>I'm pretty sure.</b><b>We'll talk about them.</b><b>Yeah.</b><b>Oh, right.</b><b>Cool.</b><b>Yeah, we'll talk about it.</b><b>No, you're right.</b><b>Sorry.</b><b>Sorry.</b><b>I thought you were going to</b><b>go and find another tension.</b><b>I'm like, no, let me get</b><b>my argument out, folks.</b><b>No, Jonathan.</b><b>No, I absolutely agree with you.</b><b>There are better ways to do this.</b><b>And one way is that if</b><b>you record things onto,</b><b>whether it's a private</b><b>blockchain or a public blockchain,</b><b>that's a way that's tamper proof.</b><b>Those log files and recordings of the</b><b>process that's being used,</b><b>not all of the AI itself, because that's</b><b>too much for a blockchain,</b><b>but just recordings of the</b><b>logs so it can be tamper proof.</b><b>And then secondly, when it's on this</b><b>immutable kind of blockchain,</b><b>you want to have selective disclosure,</b><b>which is where zero</b><b>knowledge proofs come in.</b><b>And by doing something like that,</b><b>I can prove to Chris the order to a</b><b>regulator that I have complied</b><b>based on the questions that he's asked,</b><b>but Chris never needs to see the data.</b><b>You better watch out.</b><b>I'm coming for it.</b><b>Yeah.</b><b>Oh, exactly.</b><b>This is lucky that there's this desk in</b><b>the middle thing here.</b><b>So anyway, the</b><b>guardrails are interesting.</b><b>People might be interested in giving us</b><b>their feedback responses</b><b>and we can put that</b><b>into the appropriate areas</b><b>or they're doing a response themselves,</b><b>in which case I'd encourage them to have</b><b>a lot of discussions about it.</b><b>But this is interesting.</b><b>It's still going to be years before the</b><b>government comes out</b><b>with this kind of stuff just because of</b><b>how bills are brought into Senate.</b><b>I was going to say,</b><b>that sounds too quick.</b><b>Oh, yeah.</b><b>We talked about this at</b><b>the start, their ability.</b><b>2030, folks.</b><b>No, no, no.</b><b>But if something's</b><b>important like AI FP is to be,</b><b>and it's such a hot topic, they are</b><b>moving faster with this,</b><b>I feel than when they</b><b>did with blockchain.</b><b>I was going to say to the contrary,</b><b>blockchain is a perfect</b><b>example where speed has proven</b><b>that they just can't move fast.</b><b>Exactly.</b><b>So, yeah.</b><b>It won't be.</b><b>I mean, it's good that</b><b>they've written some stuff up and,</b><b>I think their ability to</b><b>put it in in a meaningful way</b><b>is going to be very, very slow.</b><b>Super optimistic and or</b><b>super fast and just unrealistic</b><b>would be that they would get this out</b><b>within like a couple of years,</b><b>three years because like a</b><b>year for the consultation,</b><b>then they've got to write it in.</b><b>Then there's like another year.</b><b>Three years is internal.</b><b>And it's exactly, but</b><b>exactly this is the thing.</b><b>So another thing, they</b><b>should have standards,</b><b>maybe guardrails or guidelines, but like,</b><b>rules is going to be hard.</b><b>But the other part is that even if it's</b><b>two years to actually get it out</b><b>and it's in law, it's not like everyone</b><b>just gets hit straight away.</b><b>Typically, there's months or</b><b>maybe even a year for businesses</b><b>to start complying</b><b>that weren't complying.</b><b>There's a moratorium.</b><b>So these things need to</b><b>be taken into account.</b><b>What it means is just continue developing</b><b>all that type of stuff.</b><b>Grandfather.</b><b>Yeah, you have to</b><b>accommodate for all that.</b><b>But like people just need to develop, but</b><b>also respond to this.</b><b>Because if we don't</b><b>influence the response,</b><b>it'll be people that have no idea of the</b><b>technology, all right,</b><b>or the nuance that will do things here.</b><b>And yeah, do we think that's going to</b><b>hinder innovation in the space?</b><b>Yes.</b><b>Yeah.</b><b>And in a positive or negative way?</b><b>Yeah.</b><b>Depending on how restrictive</b><b>it is, I see the EU AI act,</b><b>if something like that was here, it would</b><b>be very restrictive.</b><b>Well, here's the thing.</b><b>Like, it's restrictive on those that are</b><b>actually creating the base models.</b><b>We don't really have folks</b><b>doing that sort of thing here.</b><b>And I know folks will try to develop that</b><b>sort of thing, but it is hard.</b><b>And so I think it'll affect certain</b><b>people more than others.</b><b>Hopefully, what they do here for</b><b>Australia is conducive</b><b>to what we have here.</b><b>And we've got a culture, we've got a</b><b>society of builders and developers</b><b>and businesses that are risk averse.</b><b>We've got very smart developers.</b><b>Some of the best things to come out and</b><b>being used by the world</b><b>have come out of Australia, like Canva,</b><b>for example, or the Atlassian stuff.</b><b>And in the AI space,</b><b>we've got great companies,</b><b>like the ones that do ratings that are</b><b>used a lot to rate different systems.</b><b>They call it artificial analysis.</b><b>Shout out to Micah and the team there.</b><b>But we've got great communities, like the</b><b>Build Club and DSAI.</b><b>We've got a lot of</b><b>these great things out here.</b><b>We need to have stuff that, given the</b><b>society that we've got,</b><b>very community driven, very smart, clever</b><b>tech that gets POC'd and tested here</b><b>and then taken overseas.</b><b>We need to foster that because that's</b><b>just the culture that we're in.</b><b>I think the government can do a lot more</b><b>to actually help out</b><b>with that sort of thing.</b><b>It could even have a</b><b>community of experts.</b><b>Unfortunately, right now, we have the</b><b>industry expert bodies and stuff</b><b>that I don't feel tie</b><b>in as well as they could.</b><b>They're starting to do some stuff like</b><b>NAIC, National AI Center,</b><b>does a lot of stuff with the community.</b><b>Could be more.</b><b>Others out there could be doing more</b><b>stuff with the community</b><b>because that's where the</b><b>real kind of stuff is happening.</b><b>In blockchain, whilst OzdeFi and Digital</b><b>Economy Council of Australia</b><b>and these other groups do their thing,</b><b>I feel like government</b><b>really taps into that.</b><b>Or they were when</b><b>blockchain was a hot thing.</b><b>Now it's by the back burner, but there's</b><b>still stuff going on</b><b>in the blockchain space.</b><b>I digress.</b><b>That's a big piece of</b><b>news that came out last week.</b><b>So it's just fresh, hot off the press.</b><b>And another big one as well</b><b>is that came out this morning.</b><b>And you probably saw this as well, was</b><b>the whole open AI now has more reasoning.</b><b>More reasoning.</b><b>More reasoning.</b><b>We talked about this a month ago, right?</b><b>Can I have some more?</b><b>Reasoning.</b><b>Good old strawberry here to take us into</b><b>the new world of reasoning.</b><b>Reasoning.</b><b>So what we've got here,</b><b>something like Moss from the IT crowd,</b><b>which is appropriate</b><b>for this show, right?</b><b>They've got a whole</b><b>lot more reasoning stuff</b><b>and strawberry is one of the things that</b><b>they highlight as well.</b><b>As well as it's completing logic puzzles.</b><b>It's doing stuff from, I'm reading from</b><b>others that have been trying it out.</b><b>It's doing puzzles</b><b>faster than other models.</b><b>This is not to be confused with like some</b><b>of the recent press that they did around.</b><b>And they did an opposed to talking about</b><b>next, GPT next, which</b><b>is not what this is.</b><b>And GPT next is also not a thing.</b><b>They were just using that term as the</b><b>next iteration of AI coming out.</b><b>So not to be confused with</b><b>that, but this is pretty exciting.</b><b>Just because of the quantum physics or</b><b>more about the next</b><b>set reasoning that is.</b><b>I thought you could talk about the</b><b>ponytail and they do.</b><b>But no, it is interesting.</b><b>So it's when you read into it,</b><b>and I don't know</b><b>exactly where it is here.</b><b>It's hard to navigate</b><b>on the screen just here.</b><b>But what I did read was that they're</b><b>using more chain of thought.</b><b>So they're giving it like system prompts</b><b>so that when you ask a question,</b><b>it takes your query, it</b><b>goes through its system prompt,</b><b>which is running it through</b><b>a chain of thought process.</b><b>Plus it's a better model.</b><b>Meaning that because you can run chain of</b><b>thought on just normal.</b><b>I was going to say this is kind of the</b><b>things I would use to do is</b><b>actually tell it to question</b><b>itself, run and can question yourself as</b><b>if you were talking to yourself.</b><b>Create four versions of</b><b>you and have a discussion.</b><b>And many people do that kind of stuff and</b><b>they talk to the different experts</b><b>sitting in the train,</b><b>you know, trying to tell which one of you</b><b>is the AI and stuff like that.</b><b>There's things like that out there, but</b><b>this is what it's</b><b>doing as its system prompt.</b><b>But it's not only doing that,</b><b>it's doing on a better model.</b><b>So it can do strawberry, it can count the</b><b>letter R's in there,</b><b>it can do other things that other models.</b><b>Here's the thing, you can do it with</b><b>other models, but doing it in here</b><b>has been proven so</b><b>far, so far to be faster.</b><b>My biggest take-</b><b>What's the caveat?</b><b>Yeah, what's the caveat?</b><b>It's a pretty big one.</b><b>The caveat is something I'll put out on</b><b>LinkedIn and I'll share it because I just</b><b>shared it this morning.</b><b>But it was the immediate thought that I</b><b>had and you come to the</b><b>thoughts like this when</b><b>you basically are deeper in the weeds</b><b>than the traditional kind</b><b>of just men on the street,</b><b>right?</b><b>And there's a lot of people like me that</b><b>are deep in the weeds</b><b>and we've got a unique way</b><b>that we look at it because for us, black</b><b>box models like this, normal chad GPT,</b><b>whether it's O1 or whatever, at the</b><b>moment, they haven't</b><b>shown this to not be true,</b><b>but they're a black box model.</b><b>And what I mean by that is that you don't</b><b>know how they're</b><b>coming up with the answer.</b><b>It is magic, it is more right</b><b>and we can evaluate and test.</b><b>But the evaluations, even if they got to,</b><b>they're at 89% for certain tests,</b><b>even if they got to 99%, it's that 1%</b><b>where it gets it wrong that is really</b><b>a potential problem.</b><b>And not only that, the guardrails that</b><b>Australia talks about, the</b><b>submission that we're going to</b><b>put in there is that for certain</b><b>specialized high-risk</b><b>things, GPs, financial advisors,</b><b>lawyers, it's not enough to just rely on</b><b>a black box model</b><b>where it just gives you the</b><b>answer even if it's 99% right.</b><b>And yet, there's a human in the loop.</b><b>What we need sometimes is a way to prove</b><b>that there are the sources,</b><b>the sources of information</b><b>you can see it in click on.</b><b>When you go to perplexity, you can see</b><b>the sources because they</b><b>run rag on website search.</b><b>This is where tools like perplexity and</b><b>psych come into play.</b><b>Exactly.</b><b>Much more of a rag model approach.</b><b>So it does hinder slightly creativity,</b><b>but it means that you have more trust</b><b>in the data that is being researched.</b><b>For example, like I don't trust GPT when</b><b>it comes to doing my</b><b>research work because</b><b>I can't trust that the information is</b><b>poorly, even if I asked it is true.</b><b>Yeah.</b><b>So you might use perplexity for that or</b><b>you've got rag running on your own</b><b>documents, whether it's psych or</b><b>something else, you're running rag</b><b>because you know that</b><b>you can see and you</b><b>can click on the sources.</b><b>And it was interesting that they're</b><b>showing a clinical use case</b><b>here because I would never,</b><b>even if the model is 99.9%</b><b>right, solely rely on this.</b><b>And I'm not saying that you have rag only</b><b>or you have this model</b><b>only because like you said,</b><b>there are tasks that you need where you</b><b>need to be more creative.</b><b>So you use the raw model.</b><b>There are tasks where</b><b>you need to be specific.</b><b>So you use the rag kind of model.</b><b>And there's going to</b><b>be even a hybrid rag.</b><b>And what I mean by that, you can have the</b><b>creativity of the answer you can get,</b><b>because what rag does rag does that first</b><b>search of what are the chunks of text?</b><b>What are the websites where I'm actually</b><b>going to find the answer?</b><b>That's a basic search.</b><b>And then it runs 01 GPT for Omni.</b><b>It runs whatever it is to get</b><b>the answers out of that text.</b><b>So the model actually in</b><b>rag, it's not one or the other.</b><b>It is a bit of a hybrid there, but I can</b><b>see ones where you</b><b>can have the creativity,</b><b>but have a rag process run on</b><b>top, just to prove that yeah,</b><b>the answer actually</b><b>came from these things.</b><b>It's interesting.</b><b>But yeah, I would not have used like a</b><b>clinical model from the first part.</b><b>Anyway, it's certainly worth watching and</b><b>people can try it out.</b><b>So I find it fascinating that they've</b><b>finally they've done one unlike Sora,</b><b>where it's like, hey, look</b><b>at this and you can't try it.</b><b>You can actually try it folks.</b><b>So worth a look.</b><b>Some fast news stuff that's going on.</b><b>I mean, we can't not talk</b><b>about Tesla and its AI bot.</b><b>Have you seen all the latest media around</b><b>the new AI bot that they'll be making?</b><b>What's it called?</b><b>Optimus.</b><b>Oh, Prime.</b><b>Yeah, they've put a tag against it.</b><b>They've had some examples where they're</b><b>showing it working in the</b><b>in the kitchen environment</b><b>and in the in your living room and</b><b>picking up your cups</b><b>and serving and cleaning.</b><b>It's these real life bots are now, you</b><b>know, they're becoming a reality.</b><b>They put a price tag on it.</b><b>It's like 10 to 20 thousand, I think,</b><b>depending on what market you're in.</b><b>Okay.</b><b>And they're predicting, I think it to be</b><b>available in the next like year or so.</b><b>Wow.</b><b>That's that's pretty interesting.</b><b>So thank you.</b><b>Good old good old Tesla.</b><b>Good old Elon coming coming to town with</b><b>our new or is Donald Trump calls in Leon.</b><b>Good old Leon.</b><b>Good old Leon.</b><b>On the topic of quick news, we've got our</b><b>Facebook getting a little bit in trouble.</b><b>What are they doing?</b><b>They've been they've admitted essentially</b><b>where it relates to all of their AI tools</b><b>scraping the images of their Facebook,</b><b>all the public photos to</b><b>support the data that they're</b><b>doing.</b><b>Okay.</b><b>And that is without anyone's consent.</b><b>And while I'm not massive in places like</b><b>the US, not not an issue</b><b>in places like the US in</b><b>the European Union, where it is obviously</b><b>really important that you're adhering to,</b><b>you know, consent and, you know,</b><b>Like all the GDP.</b><b>Yeah, the GDP are kind</b><b>of stuff exactly right.</b><b>They were straight and they are so</b><b>they're going to be in trouble for that.</b><b>So that's probably going to have some</b><b>backlash for them over the coming months.</b><b>So about that on the news.</b><b>Yeah, yeah.</b><b>I lost my mouse and video.</b><b>Hang on.</b><b>Hang on.</b><b>Let's just scroll through this.</b><b>And video.</b><b>Yep.</b><b>Yep.</b><b>Okay.</b><b>What's the video on?</b><b>Nvidia stock market.</b><b>They've been sort of the reigning champs</b><b>in the AI space in relation to the stock,</b><b>but they've taken a huge</b><b>dive over the last week or so.</b><b>Dropping 10% almost 12</b><b>and a half or 12, 12, 212.</b><b>I mean, I can't read 212 billion pounds,</b><b>which is half a billion dollars.</b><b>That's a heavy.</b><b>Raped off the top.</b><b>That's in heavy stock right there.</b><b>Largest single day loss in US history.</b><b>So when was that?</b><b>So we're looking at they took a nosedive</b><b>due to some stuff that</b><b>was released to the press</b><b>around practices that are hindering</b><b>client flexibility at switching to</b><b>alternative semiconductor</b><b>suppliers.</b><b>Monopoly based, shitty practices,</b><b>monopoly based, market ownership.</b><b>And it is, yeah, that's a big hit.</b><b>I wonder if they use GPT or other tools</b><b>to figure out what's a</b><b>good marketing model to</b><b>keep up.</b><b>No, they need a rag approach.</b><b>So this is the AI telling</b><b>us the application folks.</b><b>It's not the AI.</b><b>It's the application</b><b>that's the how you use it.</b><b>But that is not good.</b><b>I mean, it's talking to a lot of the</b><b>stuff that's happening,</b><b>in particular in Europe,</b><b>where they're putting a lot more eyes</b><b>restrictions, as you pointed</b><b>out before, across the stuff</b><b>that's happening in the AI space, not</b><b>just from a software,</b><b>but also from a hardware</b><b>perspective.</b><b>And competition needs to be available.</b><b>And you can't have businesses like</b><b>Nvidia, you know,</b><b>blocking the ability for other</b><b>companies to get out there in the market.</b><b>So it's kind of a good thing.</b><b>I'm happy that it's happening.</b><b>I want the competition to be there.</b><b>Unfortunately, what it will lead to is</b><b>just like, see, we were</b><b>right, we needed to have</b><b>these really</b><b>restrictive, AI like laws and stuff.</b><b>It'll be that kind of canary in the coal</b><b>mine that well, maybe</b><b>that's the wrong analogy.</b><b>But there's a Malcolm Gladwell analogy</b><b>that something something</b><b>along the lines of when</b><b>things have been restrictive, and you let</b><b>one of the people</b><b>from the restrictive area</b><b>through, it's an argument to just go see,</b><b>we told you like, you know, I don't know,</b><b>I'm probably confusing that analogy.</b><b>But my point is, when I get to the point</b><b>that I could see this</b><b>as being like positive</b><b>that yeah, we make Facebook like pay, not</b><b>Facebook, but who was Facebook, Facebook,</b><b>yeah, meta, meta, sorry, I'm thinking I</b><b>keep thinking that the</b><b>double names and stuff</b><b>there.</b><b>But like, we are</b><b>restricted on the big company.</b><b>And because of that, it uses an argument</b><b>to restrict the smaller companies to it's</b><b>dangerous.</b><b>And like, we still got a better chance</b><b>here, because we don't</b><b>have regulation kind of</b><b>laws here in Australia.</b><b>Maybe it turns out like I can see an</b><b>irony that we become the</b><b>ones that dig stuff out</b><b>of the ground that all the chip makers</b><b>need, and we make money</b><b>off that and have a boon</b><b>to the economy.</b><b>And then we also become the only place</b><b>that isn't as restrictive</b><b>in terms of its AI laws.</b><b>And all of a sudden, all these businesses</b><b>come here to develop this, I don't know,</b><b>corrupt machine run country and look, one</b><b>can only hope and</b><b>dream on a Black Friday.</b><b>But yeah, I don't think it's right.</b><b>Speaking of is that the</b><b>you don't have the meta one?</b><b>Oh, yeah, that's not for the meta one.</b><b>Like I've got some more</b><b>stuff like continue you go go.</b><b>Let me let me jump in there.</b><b>Yeah, because I think</b><b>this one is also important.</b><b>The screen there folks that are listening</b><b>online, I can never get a handle on like,</b><b>I'm snapping things to screen here.</b><b>And this is an article</b><b>from crikey of all places.</b><b>And why it's interesting is because I</b><b>highlighted something I wasn't aware of.</b><b>Look at that like article title AI worse</b><b>than humans in every way.</b><b>It's summarizing</b><b>information government trial finds.</b><b>I mean, tell us how you really feel</b><b>author article person that's right.</b><b>But also show me the data that supports</b><b>this because every test</b><b>and that I've ever seen</b><b>were there really otherwise there is an</b><b>article that you can download and stuff.</b><b>And I will great this is</b><b>yeah, I will find it here.</b><b>We're relying on a technology discussion</b><b>from a company that makes</b><b>you download a PDF to read</b><b>the article is that what we're well,</b><b>yeah, there is a little</b><b>bit of that and stuff.</b><b>But I'm going to open up the</b><b>article on like another side.</b><b>But the point was was that what happened</b><b>was that there was this</b><b>acid trial earlier this year.</b><b>And it was all about they</b><b>were doing stuff with Amazon.</b><b>So I kind of wanted to mention it there,</b><b>but they were doing</b><b>stuff to test how would it</b><b>answer a question versus how human would</b><b>answer that same question.</b><b>And there is a Senate paper, or it was</b><b>spoken about in Senate, the Select</b><b>Committee on Adopting</b><b>Artificial Intelligence. So we had</b><b>ministers talk about it. We had the</b><b>chairman of ASIC talk</b><b>about it being, yeah, it was pretty bland</b><b>the answers and stuff.</b><b>And overall, the messaging</b><b>was is that the report and this is with</b><b>Amazon, and they go</b><b>through the findings and stuff.</b><b>Unfortunately, they don't release the</b><b>full results, which I'm pretty miffed</b><b>about, like I would have</b><b>loved to have seen side by side. Here is</b><b>human answer one</b><b>obfuscate stuff. Don't tell us who</b><b>the people are, but at least go, this was</b><b>human answer one. This is</b><b>how the AI answered it. So we</b><b>can at least evaluate and stuff</b><b>ourselves. It's a frickin black box,</b><b>guys, you know how I feel</b><b>about black boxes. But also, this stuff</b><b>is can be trained very,</b><b>very easily. Like it can be and</b><b>you just tell it to go, can you summarize</b><b>this? Of course, it's going</b><b>to give you a shit response.</b><b>And then you go, can you summarize it</b><b>looking for these points?</b><b>Now it's hard to get better.</b><b>That's what they were doing. They were</b><b>getting better and better at the</b><b>responses. So there's a</b><b>long paper here, folks, definitely read</b><b>it or use AI to help you</b><b>read it and verify that you're</b><b>getting the right answers. But the point</b><b>is, my takeaway, they got better at</b><b>actually the answers</b><b>because they it was over a month. And</b><b>they did prompt</b><b>engineering to iterate and get better</b><b>responses. Now they should. That's just</b><b>prompt engineering, you can</b><b>get better results when you</b><b>have control over the whole kind of AI</b><b>workflows. And you say, the first thing</b><b>is to do this, you're</b><b>looking at this document, the second</b><b>thing, and as part of that,</b><b>take these steps, and you can</b><b>break it down into sub tasks with agents,</b><b>psych. But secondly, even</b><b>if they weren't using psych,</b><b>they showed a improvement. And B, they</b><b>didn't speak about it. But</b><b>this thing will do like AI,</b><b>as we know, we'll do things in a 10th of</b><b>the time, or like 1% of the time that it</b><b>takes a human to do.</b><b>So they didn't speak about that being the</b><b>positive thing that even if this thing</b><b>to come in afterwards, and then make it</b><b>more human. And this is like a classic</b><b>kind of argument that</b><b>you see in businesses that even though</b><b>the Yeah, sure, it wasn't as</b><b>good. And even if it was 50%</b><b>as good, or it gets to 80% as good with</b><b>time and training, that's still way</b><b>better as position that</b><b>you're in in terms of time. So because</b><b>now the human yet, it takes</b><b>five minutes for this thing</b><b>to do an hour long task, and then it</b><b>takes the human another five minutes,</b><b>there's 40 minutes or</b><b>sorry, 50 minutes, if it was like an hour</b><b>versus 10 minutes, that you</b><b>have now to do something extra,</b><b>I don't know why people keep on not</b><b>talking about this kind of stuff. It's</b><b>offensive, really. So I</b><b>feel like, you know, I'm offended. I'm</b><b>offended. This kind of stuff</b><b>was good. But it could have</b><b>been like way better. What we should be</b><b>doing and I put an article</b><b>like on this, like last week,</b><b>is that we should be finding ways to</b><b>improve these models,</b><b>because it is improving instead</b><b>of nitpicking going, Oh, look, it's not</b><b>as good as like a human.</b><b>Yes, you idiots just like cars,</b><b>sorry to the idiots out there. But it's</b><b>just like cars. No, they're</b><b>not idiots. It's just that we</b><b>need to sorry, I'm being really mean</b><b>here. It is Black Friday folks. I don't</b><b>normally am like this,</b><b>but it's sense the passion, the rage, the</b><b>rage, right. But it's just</b><b>like when cars like I want</b><b>everyone to do better. I don't want to</b><b>keep people out or anything</b><b>like that. So I'm actually trying</b><b>to be encouraging you. When cars first</b><b>came out, the biggest complaint was that</b><b>one of the complaints,</b><b>they don't do well on these like roads,</b><b>these cobblestone roads</b><b>and stuff. And then what</b><b>happened when they paved the streets? Not</b><b>only did cars do really</b><b>well, but they had new types</b><b>of transportation buses and motorbikes</b><b>and other things opened up</b><b>that you could never think of.</b><b>Think of the paving of the roads like</b><b>we're seeing with AI here, what more we</b><b>could do. So this is</b><b>actually pretty important, more important</b><b>that people think that</b><b>stuff like this is happening,</b><b>because it's happening in the realms of</b><b>people that actually make</b><b>the decisions in government.</b><b>Right. So we need better examples is what</b><b>I'm saying. Yep. No, the</b><b>answer good call. All right.</b><b>Hit me up with some quick news talking</b><b>about, you know, the</b><b>people who sort of triggered the</b><b>whole this new AI generation, we're going</b><b>back to GPT and open our</b><b>friends 200 million users.</b><b>They've like doubled since since last the</b><b>last fin year, I think it was 1 million</b><b>enterprise users, apparently, which is a</b><b>big 600,000 enterprise</b><b>licenses in April. Yep.</b><b>To a million recently. Are there a</b><b>million businesses out</b><b>there? No, no, it's not. Yeah,</b><b>exactly. A million people or licenses</b><b>them on and even be people.</b><b>There's definitely a million businesses</b><b>out there. So I mean, that's which is</b><b>incredible. Yeah. But</b><b>what's still is perplexing is how much in</b><b>a negative revenue</b><b>driver this business is at the</b><b>moment. And so they're not financially</b><b>doing well, but and they're</b><b>back in the open AI and back in</b><b>the market for capital race. So they're</b><b>looking for $6.5 billion. Is</b><b>that all going for a market</b><b>sort of valuation of over 150 billion</b><b>company. So there's</b><b>various reasons why that could be</b><b>happening. And I haven't looked at</b><b>financials. Like it's well, you're</b><b>overstating. So Amazon</b><b>Amazon didn't make money for a long time.</b><b>Yeah. You know, so they're</b><b>not worried about revenue,</b><b>they're worried about growth. And that's</b><b>fine. Because if you think</b><b>about it, like when you see</b><b>and obviously they don't apply the same</b><b>pricing that they apply to</b><b>folks that use their API's</b><b>when you see the pricing of the API's,</b><b>whether you're using it</b><b>through Azure, in a private kind</b><b>of way where it stays in Australia or</b><b>other countries or using</b><b>it directly with open AI,</b><b>there's a cost that you pay for input and</b><b>output tokens. Even if</b><b>they don't apply that model to</b><b>themselves, I mean, they should there's</b><b>there's an internal cost to</b><b>doing it. There is a cost of</b><b>these things. And if they're only</b><b>charging you 20 US dollars or whatever</b><b>that is in Australia,</b><b>and 30 bucks, there are many people</b><b>probably like you and I that would use</b><b>way more than that, you</b><b>know, in our monthly usage. And they're</b><b>relying on, you know, many people that</b><b>wouldn't. But because</b><b>of that and other things they spend on</b><b>research, no wonder they're</b><b>losing money, but they're getting</b><b>growth. And that's the most important</b><b>thing. Exactly right. I'm</b><b>sticking with the AI, open</b><b>AI guys, one of the co founders, Mr. I'm</b><b>never gonna be able to</b><b>say his name, right? Say it.</b><b>Ilya. Sutskova. Sutskova. Sutskova.</b><b>Sutskova. What a name. Obviously, he, he</b><b>left or a while back. Yeah,</b><b>he did. Started something new. And</b><b>started his own company, the</b><b>Safe Super Intelligence SSI.</b><b>Did you do you remember when he was</b><b>leaving when there was a</b><b>whole like they fired Sam and then</b><b>they brought him back and he was like</b><b>part of that group and Sam</b><b>puts out that like, I mean,</b><b>he's the master of like, for strawberry</b><b>model and stuff, he put out</b><b>a picture of a garden full of</b><b>strawberries and stuff, right. So you</b><b>know, fascinating. And</b><b>then for the whole Ilya thing,</b><b>he put out a tweet where it started off</b><b>with I love you all. I L Y A Ilya. So</b><b>he's a master of like,</b><b>ooh, deception. The master. But did I</b><b>mean that? But yeah,</b><b>what's what's Ilya doing?</b><b>So obviously with his business, I mean,</b><b>he doesn't actually have a product in</b><b>Markier. So this is,</b><b>I mean, this just talks to how much money</b><b>particular states</b><b>getting thrown around at AI</b><b>ideas. He's raised a billion dollars for</b><b>his idea, which is, I</b><b>mean, the product is like,</b><b>is really interesting. And it talks to a</b><b>lot of the points you</b><b>were raising before around,</b><b>you know, safety and alignment, you know,</b><b>so AI supporting companies</b><b>and being risk more risk averse,</b><b>which is which is great. So the product</b><b>idea really, really cool.</b><b>But to be able to raise a</b><b>billion dollars, and this talks to the</b><b>power of the person, right?</b><b>Yeah, such a big influencer in</b><b>the AI space, be able to raise that much</b><b>capital. Oh, it's crazy.</b><b>It's in doing it off an idea,</b><b>but it's not like I mean, he's a founder</b><b>with pedigree, he's helped</b><b>grow that business of AI,</b><b>and he was seen as the the mind of what</b><b>grew that space there. But</b><b>I find it fascinating that</b><b>there's these people that are saying, and</b><b>maybe it's because of the</b><b>restrictive clauses that</b><b>supposedly were there, that were there</b><b>before, supposedly not</b><b>there now, with leaving open AI,</b><b>but the whole no, no, everything's all</b><b>fine here. Like I was doing</b><b>safety research and stuff,</b><b>but I'm just going to go do it outside.</b><b>Yeah, you know, it's kind of</b><b>fascinating. Like I love my</b><b>girlfriend, but I'm going to go love</b><b>another girlfriend,</b><b>because I just have to be outside</b><b>here and stuff like nothing wrong with</b><b>that girlfriend. And</b><b>like, what are you? Anyway,</b><b>people can read between the lines. It's</b><b>going to be interesting what he creates.</b><b>Hopefully, it's not a $1 billion thing</b><b>wasted. Let's hit up the</b><b>trillion dollar companies out</b><b>there. So we've got our Bing has really</b><b>some interesting deep</b><b>fake scrubbing tools, which I</b><b>think also a little bit of the other</b><b>platforms like Google, etc.</b><b>I've already released this.</b><b>So that's in collaboration with Stop NC</b><b>II or NC2. And they're allowing people</b><b>to, I guess, identify,</b><b>to flag and even for the system to</b><b>automatically identify, you know, fake</b><b>content that's going out</b><b>there. And there's a lot of stuff that's</b><b>been happening with</b><b>pornography, etc. has been going</b><b>around quite bad. Very good. Google's</b><b>released a new photo search in currently</b><b>in beta. So for like</b><b>select people out in the US region, so</b><b>essentially in your photos</b><b>app, you'll be able to click</b><b>into the application and do language</b><b>language based search. So</b><b>you could type in, you know,</b><b>what did I eat for when I was in the Gold</b><b>Coast last weekend and</b><b>boom, it would be able to find</b><b>images based on that and show you food,</b><b>etc. So contextual image</b><b>search, which is which is</b><b>quite interesting, no longer you can just</b><b>type in, you know, oh, yeah.</b><b>And painting in the backyard.</b><b>Yeah, I do it currently now because I've</b><b>got a Google phone folks.</b><b>And I can type in things like</b><b>I had a friend of mine, shout out to Rob</b><b>Morris. He rocked up to a</b><b>party of mine, like after I</b><b>requested it because I was like, Hey, can</b><b>you do it? Because like,</b><b>you've got it there. Mr. Rob,</b><b>because he's got armor he had like I've</b><b>been in his place. He has a</b><b>cool thing called the whiskey</b><b>dungeon. And in the whiskey dungeon is a</b><b>night in is it's a night</b><b>suit, like a literal heavy armor</b><b>night suit. Right. And he's telling me</b><b>about chain mail and</b><b>everything. Yeah. Oh, he wore it to</b><b>your house. Remember, he wore it. And I</b><b>went into Google because I was talking</b><b>with a common friend</b><b>of ours and trying to find it and then</b><b>just type in night and it</b><b>comes up. This is even more than</b><b>that. You can actually get into certain</b><b>like phrases and it</b><b>will exactly look it up. So</b><b>fascinating. That's cool. So that's in</b><b>beta at the moment and</b><b>early access for some people who</b><b>have applied for it. And on the crypto</b><b>space coinbase, oh, yeah, conducted a</b><b>transaction managed by</b><b>AI bots. If you saw this, this was that</b><b>this was very interesting.</b><b>Yeah. So they ran using using</b><b>doing the exchange token. So Brian</b><b>Armstrong, the founder of Coinbase ran it</b><b>put a post out of this</b><b>on x good old x God, I hate x but engines</b><b>cannot get bank accounts,</b><b>they can get crypto wallets,</b><b>they can now use usdc on base and</b><b>transact with humans, merchants and other</b><b>ais, which is kind of</b><b>interesting. So these transactions are</b><b>instant global and free. And he ran a</b><b>test and shared this</b><b>out with everyone on the on x as well for</b><b>people to see so</b><b>fascinating, fascinating stuff for you</b><b>know, AI taken over our financial</b><b>decision making processes. But very cool</b><b>to see that in trade.</b><b>I think it's interesting as well. It's</b><b>not just like the</b><b>transaction people typically think of</b><b>like, crypto is only or blockchain tech</b><b>or web three or whatever is only being</b><b>about the transaction</b><b>side of things. And whilst it is in many</b><b>ways, there are other things that</b><b>blockchain is really</b><b>useful for such as being able to record</b><b>for supply chain</b><b>purposes. And this is what one of</b><b>the base use cases of blockchain tech has</b><b>been for supply chains,</b><b>being like beer manufacturers</b><b>recording the quality of items that are</b><b>being used and making</b><b>sure that the process has the</b><b>high quality they need. Having a record</b><b>onto an immutable</b><b>blockchain where it can't be tampered</b><b>with is really key, as I mentioned</b><b>earlier, as well as like how it's going</b><b>to help AI in terms of</b><b>that security side of things. I think</b><b>it's important. But</b><b>transactions is really good too.</b><b>It could be interesting that there will</b><b>definitely be negative</b><b>things that happen as a result of</b><b>this. But we need those negative things</b><b>so we can see those issues</b><b>before they become too big. So</b><b>this is really fascinating stuff. What's</b><b>next? And lastly, I just</b><b>want to talk about video.</b><b>I love that there's a lot of movement in</b><b>the video space, a lot of</b><b>tools that are releasing</b><b>out of the market. But Looma AI,</b><b>definitely one of my favorite ones out</b><b>there. They had 1.5,</b><b>which launched back in August, which was</b><b>better prompting around video generation,</b><b>editing images, etc. But they just</b><b>launched version 1.6, I</b><b>think about a week ago. Okay.</b><b>Which came with a whole bunch of really</b><b>cool camera control</b><b>tools. So imagine now you can</b><b>articulate motions or selections within</b><b>it for the camera. So let's</b><b>say you don't like the video,</b><b>what the heck single frame, you can be</b><b>like, I want you to pan up or</b><b>pan around pan left zoom in.</b><b>So you can actually start interacting</b><b>with the video content that you're</b><b>generating to create</b><b>your final product. Unbelievable. I'd</b><b>love to see very, very cool. Because you</b><b>and I have done this,</b><b>right? Like I want to have like, if I</b><b>take a screenshot of</b><b>something, say to the AI, insert</b><b>this into this image, or maybe you don't,</b><b>you just use an image generator to</b><b>actually like not image</b><b>generator, you just insert it like normal</b><b>ways, like Photoshop or</b><b>Canva. Yeah, you insert that</b><b>into an image generator like this, and</b><b>then you can move around the</b><b>screen, but it's still showing</b><b>your product or something like that.</b><b>Crazy stuff that really,</b><b>and it's not just like,</b><b>saying pan around or pan into that you</b><b>can tell it to like, you know,</b><b>do a slow pan into a slow pan</b><b>out. It's actually controlling it</b><b>meaningfully. Because if you</b><b>just said like, you know, pan</b><b>around to the right, if the camera just</b><b>moves, there's, you know,</b><b>it would be horrible, but you</b><b>can actually control how pansy and how it</b><b>controls the depth of field in that</b><b>motion. It's very, very,</b><b>very, very clever. Fascinating.</b><b>Fascinating. What's next?</b><b>That's it for my side of things.</b><b>Anything you want to call?</b><b>Yeah, I mean, if people are interested in</b><b>the AI guardrails and</b><b>what's going on there, definitely</b><b>join the build club if you haven't</b><b>already, because I'll be</b><b>speaking about it and we'll be</b><b>sharing away for people to give their</b><b>feedback. Speaking of the video stuff</b><b>that you showed there,</b><b>have a look out for what's been going on</b><b>in the gaming space when it</b><b>comes to things like Doom,</b><b>where they're generating as the player is</b><b>moving around, they're</b><b>generating the senior, and it</b><b>gets a bit choppy, but it's improving.</b><b>But literally done a lot</b><b>of demos around even just</b><b>conversation. Yeah, communication with AI</b><b>tools. Oh, yeah, they're no</b><b>longer the NPCs, but NPCs 2.0,</b><b>because you can actually have interesting</b><b>conversations and some</b><b>esoteric like conversation</b><b>with them where they're like, am I real</b><b>or not kind of thing. And</b><b>then Minecraft was a really</b><b>interesting one that you mentioned</b><b>before, what's what's going on there. So</b><b>they someone created an</b><b>environment within Minecraft and had a</b><b>whole bunch of AI bots</b><b>running around. And actually,</b><b>they had to create their own little</b><b>society within the game. And so when he</b><b>had to had got to a point</b><b>where someone went missing, and they</b><b>rallied together to try and</b><b>find the person because they</b><b>felt like they were missing out. So there</b><b>was actual real life sort</b><b>of emotions civilization.</b><b>Yeah, like a civilizations being built up</b><b>in this AI scenario. And then</b><b>they had, they had what was,</b><b>it was like, good versus bad, they had</b><b>some like, had some negative things</b><b>happening to them. And</b><b>how would they react and respond? Yeah,</b><b>it was it was, it was</b><b>scarily realistic and human like in</b><b>terms of actions. Obviously, there was no</b><b>real true emotions or still</b><b>based, it's still an AI tool</b><b>doing its thing. But it was very, very</b><b>cool to watch. I get it talks to the</b><b>potential that within</b><b>video games, which you just touched on,</b><b>like imagine being able to</b><b>jump into a game. I don't</b><b>know, like you think about the cyberpunk</b><b>27 seven or any of those other games</b><b>where the AI tool, the AI</b><b>NPCs in there are actually interacting</b><b>and living their lives. And you're just</b><b>sort of going around.</b><b>It's like free guy. Yeah, like free guy,</b><b>but like real life free guy happening.</b><b>If you haven't seen it, Ryan Reynolds,</b><b>he's an NPC, but he comes to</b><b>realization that he needs to</b><b>change how he does things. And anyway,</b><b>like, you get to that and it gets to</b><b>things like, could you</b><b>have, based on how you play a</b><b>representation of yourself that plays a</b><b>game offline when you're</b><b>at work and stuff and it plays mimics</b><b>like how you play. So</b><b>that's one another would be that</b><b>you have these categories right now. Why</b><b>is he dead? You have you can</b><b>have suicide from depression.</b><b>Yeah, you could have. Are you okay was a</b><b>day that we had yesterday and</b><b>stuff folks or the day before.</b><b>So definitely worth looking at that. But</b><b>you could have right now</b><b>we've got versus others versus the</b><b>MP versus just like one player, which one</b><b>play typically in games is</b><b>just like you just playing</b><b>as a computer. But you could have one</b><b>player now where it could be</b><b>smarter. Like it's like the</b><b>quasi the hybrid that versus AI kind of</b><b>thing. So you can play</b><b>against normal NPCs, you can play</b><b>against like the hybrid AI ones, or just</b><b>versus others. And you'll</b><b>never know like what imagine</b><b>like the lobby of like Call of Duty when</b><b>it's like 11 labs voice</b><b>of someone you think you're</b><b>playing against, I don't know the rock or</b><b>something like that.</b><b>And it's not like it's,</b><b>it's fascinating stuff. We're talking</b><b>about in gaming. What about</b><b>in work? What happens when</b><b>there's like future bots that remote</b><b>workers you think you're</b><b>talking to them and they're doing</b><b>this work and it's not it's really good</b><b>quality work. You don't</b><b>think that it's not human. But</b><b>the person sound that you're talking to</b><b>an emailing and talking to</b><b>online is even like an AI agent</b><b>taking it to the extreme. There couldn't</b><b>be full automation of</b><b>someone's life there while they're</b><b>sitting on a beach in Kosamui. That's</b><b>exactly it. Anyway, we'll be building</b><b>that next week, folks.</b><b>You've just like you just given away all</b><b>IP. So just shout out to</b><b>this looking up the brands here.</b><b>So project said was the name of the</b><b>project that I ran here within Minecraft.</b><b>Was an ex MIT professor, neuro,</b><b>neuroscientist with Dr. Robert Yang. So</b><b>his team at Altair AI</b><b>started focus on building state of the</b><b>art autonomous agents.</b><b>So essentially, there were</b><b>a bunch of agents that were running</b><b>around within Minecraft, creating this</b><b>little civilization from</b><b>scratch. So it was very, very cool to</b><b>watch and recommend jumping on to the</b><b>links will share below</b><b>if you want to check out how AI can</b><b>create its own little civilization within</b><b>Minecraft. Fantastic.</b><b>And with that, I think we should probably</b><b>wrap up given the room is</b><b>going to be occupied pretty</b><b>soon. Yeah, we've got to wrap up. But</b><b>folks, shout out to psych AI, shout out</b><b>to stone and chalk for</b><b>the space shout out to digital village.</b><b>Yeah, and everyone, if you subscribe,</b><b>they really help us out.</b><b>If you have any questions for us, feel</b><b>free to reach out as well.</b><b>If you want to know any more</b><b>anything more about AI, psych, digital</b><b>village, other other things that are</b><b>going on in the space,</b><b>any questions we have, particularly</b><b>around the legislation</b><b>discussions, relation to AI.</b><b>Yeah, feel free to hit us up. It is up,</b><b>folks. We'll see you next</b><b>week. Thanks for watching.</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