Select Page

The Robots are coming! We are talking about the latest in GPRs (general purpose humanoid robots), Apple cancelling their car project, Gemini 1.5 Pro testing, Biden’s plan to ban voice impersonation, the decline of TV viewership, and Deep Mind CEO Demis Hassabis’ views on AlphaZero sitting atop LLMs on the AGI stack.


Futuristic 22

[00:00:00] Cameron: Welcome to Futuristic episode 22. Uh, forgive me, father, for it has been a month since our last recording. And by father, I’m talking about Steve. I don’t know. Hey, how

[00:00:27] Steve: Uh, definitely my fault. Definitely my fault. By the way, Cam, Futuristic22, I was on TikTok this morning.

[00:00:35] Steve: This one business cat has some, one of those guys that has good

[00:00:38] Steve: advice. And he was talking about how

[00:00:39] Steve: to

[00:00:40] Steve: build your brand and all that kind of stuff. And he said that only 9 percent of podcasts

[00:00:44] Steve: make it to episode 50.

[00:00:45] Steve: So we’re nearly halfway.

[00:00:49] Cameron: Ah, yes. F22. Uh, Steve, what have you done in the last month that you can impress us with that’s related somehow to emerging technologies?

[00:01:02] Steve: Yeah, I’ve been working with a few of my clients on trying to build out bespoke

[00:01:08] Steve: corporate AIs and we’ve messed around a lot with building GPTs. For the companies. One of them’s an electrical company that I work with

[00:01:16] Steve: and we’re building Sparky AI, which is a go to tool for electricians, you know, code standards, all that kind of stuff, and it’s not bad, but I think that GPTs are never really going to solve the corporate problems the way that I, and I’ll be quick on this, the way that I see it, there’s going to be three types of AIs that emerge, there’s going to be What I call global AIs, which are the general purpose AIs that we all use, like Gemini and ChatGPT.

[00:01:42] Steve: They’ll eventually be personal AIs. I think Apple might be doubling down to move towards that, where it’s your kind of quasi digital twin. And then in the middle, there’s the corporate AI. And we might liken that to an intranet or a wiki, but with functionality to do things, not just serve up information.

[00:01:59] Steve: And even though Microsoft is doing that with their co pilot, it seems very, let’s say, uh, efficiency tools centric. You know, help you with PowerPoints and finances and spreadsheets. But I can’t help but think that All of the AIs that are being worked on, none of them will really be able to work through the complexity of a person, of a corporation’s data pool or data lake, which are really, really complex.

[00:02:25] Steve: And I’m now working with a company called Actualization AI, who do a combination of Training on internal data pools, using open, uh, LLM models and solving specific business problems with specific AIs. I feel like that’s going to be a big growth industry, uh, simply because, you know, the world is messy and complex and even two companies in the same industry, fundamentally different, you know, they have different pieces of data from different places that have been accumulated over the past 30 or 40 years.

[00:02:58] Steve: And I feel like, you know, Those kind of closed system AIs that need to be built out in a bespoke manner is going to be a big thing.

[00:03:06] Cameron: So tell me, from your perspective in the corporate world, what are your clients looking for AIs to do for them? What are their current ambitions for AI?

[00:03:18] Steve: Two things that I see, they want their staff to start using it to save time and create efficiencies where they don’t have

[00:03:25] Steve: to do repetitive work or they can automate things. And that’s a classic efficiency game. How can I get more out of

[00:03:31] Steve: an employee? How can I have less employees

[00:03:32] Steve: doing more? So that, and, and the LLMs do a good chunk of that.

[00:03:36] Steve: And a lot of the tools that are being presented and apps that are going out do that. But the one thing they’re all coming to me saying now is. How do I use all of this stuff that we have, all of this information, stock flows, warehouses, retail, financial, how do I get AIs to do this thing or that thing?

[00:03:55] Steve: And there is just nothing out there that would, I don’t think, ever be able to do it, simply because the inputs are going to be too varied by corporation. So they’re the two things. How do I get my staff to use these AI tools to be more efficient? It’s a little bit like computer literacy, you know, back in the early 90s.

[00:04:11] Steve: It’s that all over again, but it’s AI literacy where they become more efficient. But then the big one, the really big one is, I know AI can do all of these things. How can I create an AI that does X, Y, and Z? on all of my internal digital workings. And that, that I don’t think can be solved at a macro level.

[00:04:28] Steve: That’s going to have to be bespoke software again in some capacity, I think. They’re the two things they’re asking for. And that second one, you know, they’re prepared to spend many millions on getting that right. That’s a really big opportunity.

[00:04:39] Cameron: when you say do X, Y, and Z, can you give me some examples? What’s it, what is X, Y, and Z? What are they looking for it

[00:04:45] Steve: Okay. So, so I’ll give you an example that electrical company I work with is a wholesaler with 113 branches that serve

[00:04:54] Steve: Uh, electricians and

[00:04:56] Steve: trades people, they want to be able to say

[00:04:59] Cameron: kind of thing.

[00:04:59] Steve: yeah. So they want to say, how can I maybe be like Amazon? How can I know exactly what’s in my stock flows? How can I teach my staff to substitute a product?

[00:05:09] Steve: But if they substitute it, it’s one that’s higher margin. How can I. Um, know exactly what’s happening in each branches and where customers are going. And how can I look at which customers I’m serving that are more high value, all of that stuff where it’s really goes through their supply chain and gives insight on supply chain in real time and actually helps decision making, which is not based on one person’s experience, but making sure that every staff member has the same level

[00:05:34] Steve: of experience underneath them to make more profitable decisions within that supply chain.

[00:05:39] Cameron: So they’re looking for it to be an

[00:05:41] Cameron: intelligent, uh, member of management, really. They’re looking for AI managers that can understand,

[00:05:50] Cameron: that can take a 10,

[00:05:52] Cameron: 000 foot view of all of the data inside of the business and make intelligent managing decisions based on what it can see. And the amount of data in theory would be able to process about, you know, 10, 000, 20, 000, 30, 000, 40, 000, 50, 000, 50, 000, 50, 000, 50, 000, 50, 000, 50, 000.

[00:06:06] Cameron: A lot of complex data and bringing it all

[00:06:09] Steve: Right. And it could even be, you know, which, what stock

[00:06:12] Steve: to carry and what the trends are and then bringing in, you know,

[00:06:15] Steve: uh, building approvals and what’s that going to affect on different materials that we’ll need. And that’ll be more demand and where are the

[00:06:21] Steve: artists, all of that kind of stuff. Because what happens is it gets to an end point where a human makes a

[00:06:26] Steve: decision. and

[00:06:28] Steve: 10 percent of the employees might make an

[00:06:29] Steve: informed decision and 90 percent won’t.

[00:06:31] Cameron: Hmm. Well, and, and the, yeah, my

[00:06:34] Cameron: perspective on that is we’re nowhere near that yet. Nowhere near it. Nowhere near it. I mean, we have to keep reminding ourselves and in your

[00:06:44] Cameron: case, reminding your

[00:06:44] Cameron: clients that what LLMs do.

[00:06:48] Cameron: Is finished sentences. That’s what they do. They’re not, um, they don’t have really the ability to reason, logic.

[00:07:01] Cameron: They’re not intelligent in that sense,

[00:07:03] Steve: different. It’s probabilistic. It’s probabilistic. They, in some ways, they think AI is here now and it’s magic. So unleash the magic upon my corporation, right? Unleash the

[00:07:13] Steve: magic, right? But, but

[00:07:14] Steve: also

[00:07:16] Steve: in, in many ways, it’s, LLMs have raised the AI agenda, but I think what they really want is something that’s more akin to a Google Maps, which gives me an if this, then that protocol, which helps me make a better

[00:07:29] Steve: decision, which means that you need to build a specific, let’s call it an AI that does that task and combines it with some LLM to help transliterate what, what needs to

[00:07:39] Steve: happen.

[00:07:40] Steve: So it’s

[00:07:40] Cameron: And this

[00:07:41] Steve: at the moment that this is a big conversation.

[00:07:43] Cameron: this is what I’ve been saying on

[00:07:44] Cameron: this show for the last year is that LLMs aren’t going to get us there, uh, in and of themselves. I think they’re an incredibly powerful part of the jigsaw puzzle because they’re giving us the ability to allow computers to understand natural language and to communicate back in natural language and to communicate with each other.

[00:08:06] Cameron: with each other in natural language, but then you need the logic level, you need the symbolic logic on top of that, and we’ll get to that. Um, I’ve been watching some interesting videos recently by Jan LeCun and Demis Hassabis, the um, respective, uh, Jan LeCun runs sort of AI for Meta, and Hassabis is the co founder and the CEO of DeepMind, which is now Google DeepMind, that’s doing a lot of the work on, you know, they built, AlphaGo and AlphaFold and the, you know, the systems that are the best chess players, best Go players, all the medical stuff, which are, which do have deep, expertise in those specific domains.

[00:08:50] Cameron: The first one they

[00:08:50] Cameron: built was something that could play

[00:08:51] Cameron: like Atari games. And, um, you know, and he talked a lot of this couple of videos I’ve watched about how they did that. Anyway, we’ll get to that later on. Uh, well, for my money, Steve, in the last month, um, I’ve been doing a lot of scripting and, and, and like improving my

[00:09:06] Cameron: scripting skills and just, you? know, as a reminder for people, I’m not a Coder

[00:09:10] Cameron: by, uh, you know, uh, yeah, I am now and it’s, and I’ve said this before and I’ll

[00:09:14] Cameron: say it again, it’s addictive once you get into it, once you start coding stuff, it becomes, it’s like a crack habit, you want to code everything and,

[00:09:25] Steve: this, sense of control, don’t you? It’s a God like thing.

[00:09:27] Cameron: a sense of power really is, but it’s also, it’s also, you know, a part of the, the problem solving aspect of it. Like it’s, you’re like, oh, it’s not quite right. I wonder if I tweak this line or tweak that line, or tweak this piece of code if it’s gonna, if I, if I’m gonna get there. And it becomes sort of addictive trying to figure out what the magic solution is.

[00:09:47] Cameron: But I wrote a script, uh, in the last couple of weeks that I’m kind of proud of. Um, it. You know, as part of my investing show, I’m investing, we have a, we have a U. S. portfolio and there’s sort of 800

[00:10:02] Cameron: stocks on the buy

[00:10:03] Cameron: list that I’m having to analyze out of the many, many thousands that are on the NASDAQ and the New York Stock Exchange.

[00:10:10] Cameron: Uh, once I do all my basic filtering, I’m left with 800 that I need to do another level of filtering on it because I don’t know much about the U. S. market outside of the big, you know, Mag 7 type companies.

[00:10:19] Steve: Mag7, no more Gamma Gaffer, Mag7,

[00:10:23] Steve: got to change the names. All right, keep going.

[00:10:25] Cameron: I, I, I, I’m looking for the stocks that are commodity stocks because part of our investing process is before we invest in a commodity company like a mining company, we want to know where the underlying commodity is at in terms of its life cycle and price cycle.

[00:10:39] Cameron: Um, so anyway, I wrote a piece of script that, uh, with the help of ChatGPT, obviously, that, Uh, it takes a list of stock codes in a Google Sheet, then we’ll write, OpenAI will write me a paragraph business description of each of those companies and put it in the next column. Then, It will reference, uh, another sheet in that spreadsheet that has a list of all of the commodities that I downloaded from the World Bank’s, uh, Commodity Index.

[00:11:10] Cameron: Then it will look at the business description that it wrote, cross reference that against this list of global commodities, and make an intelligent decision about which underlying commodity that business probably relates to, and then put that in another column. So then I can just filter. on that and figure out which commodity this business is involved in.

[00:11:30] Cameron: And the thing that was cool about that for me is, um, I figured out how to write script that calls the OpenAI API to do all of that from inside of a Google Sheet. So essentially I can get, um, OpenAI’s API analyzing stuff in spreadsheets for me now at a different level, which, um, You know, I still don’t think CoPilot does really yet.

[00:11:54] Cameron: Um, it, it sort of has very basic, uh, AI functionality every time I check out where CoPilot’s at. But that’s sort of the dream is, you know, you’ll be able to just point it at a list of things and say, you know, apply some level of intelligence behind the scenes to this for me. So being able to code that was pretty exciting

[00:12:13] Cameron: for me.

[00:12:14] Steve: righty.

[00:12:15] Cameron: Let’s move into news, Steve. Uh, well, it’s been a big month, uh, since we last did a show. Um, one of the first things I wanted to talk about was Google released their Gemini 1. 5 Pro model. I think we mentioned that on the last episode. And I got access to it. Uh, about a week ago, finally, I was on the wait list.

[00:12:39] Cameron: This is the one that has a 1 million token context window. I think the late, the, the, the best version of GPT for Turbo or 4Pro or whatever they call their top level now is about a 250, 000 token context window. For people who don’t know what a context window is, it’s the amount of data basically that you could put into, um, a query.

[00:13:07] Cameron: So, you know, the size of the documents or, uh, a token is basically a collection of letters at the end of the day. Um, Size of the document, size of the PDF, um, the size of the video, etc. So Gemini 1. 5 Pro enables you to do a million token context window. So in theory, you can upload very, very large files and get it to analyze it for you.

[00:13:32] Cameron: In practice, I haven’t been impressed. I’ve been uploading stuff to it and it’s, it’s not as good in my experience in all the things that I tested it on as GPT 4 has been, and particularly in coding. Um, and what surprised me the most was getting it to try and code Google Sheets. Oh, I, I, back to the OpenAI thing, here’s something that was hilarious, slash not.

[00:13:57] Cameron: Trying to get ChatGPT to code for me for the OpenAI API was insanely frustrating because the version of the API documentation that it has been trained on is out of date. So, I would go to the current

[00:14:17] Cameron: documentation, figure out what the, you know, the, the scripting framework was, say, we have to use this framework for making the API call.

[00:14:27] Cameron: And then it would give me the outdated one, which doesn’t work. And I kept saying, no, no, no, no, no, no. You have to use this one. Oh, yep. got it. It was like

[00:14:35] Steve: it. says got it all the time, but doesn’t get

[00:14:38] Steve: it.

[00:14:38] Cameron: but doesn’t get it. This gets back to the fact that I said before, it can’t reason, it can’t remember, um, even though OpenAI said a while ago it was going to have memory, it

[00:14:51] Cameron: doesn’t.

[00:14:52] Steve: doesn’t. You can trick it.

[00:14:53] Steve: after about four questions. And then often you have to go back and say, go back to the fourth

[00:14:56] Steve: thing. And it very, very often says, okay, I understand that. And then delivers. An

[00:15:02] Steve: iterated version of the equally crap thing you were unsatisfied with in the first instance.

[00:15:07] Cameron: Even its own documentation! It can’t even get its own documentation right, which was, uh, insanely frustrating. Anyway, Gemini’s 1. 5 Pro model, bottom line is, I wasn’t overly impressed. I haven’t been able to do anything with that, that I couldn’t do better in ChatGPT 4. Um, but, you know, I think the competition for new models and bigger context windows and everything that’s out there at the moment is healthy, and it’s, uh, driving a lot of, you know, You know, progress and innovation.

[00:15:35] Cameron: I’m glad that Google are going toe to toe. Speaking of going toe to toe, Apple

[00:15:42] Cameron: announced a couple of weeks ago that their Canceling their Apple car project, which they’d been running for about 10 years. My son, Taylor, uh, one of his good friends that he had a business with a couple of years ago, um, worked, was an engineer on the Apple car project.

[00:15:59] Steve: All right. Okay.

[00:16:02] Cameron: And

[00:16:02] Steve: some

[00:16:02] Cameron: never, could never tell Taylor

[00:16:04] Cameron: what he was doing. I sent Taylor the press release and he goes, no, I just spoke to him yesterday. He would have told

[00:16:11] Cameron: me. And then he said, he said, the guy messaged the guy was like, yeah, I found out

[00:16:15] Cameron: like a couple of hours ago,

[00:16:16] Cameron: we were all told that’s it, you’ve got three months to find another job inside of Apple if you can, otherwise you’re out on the street.

[00:16:24] Cameron: I mean, these guys aren’t going to have any problems finding work, I imagine, but, uh, very,

[00:16:28] Steve: I’ve just been working at

[00:16:29] Steve: the world’s most successful company doing a whole lot of

[00:16:32] Cameron: Yeah. Um, but

[00:16:35] Steve: it’s a good decision. I

[00:16:36] Steve: think

[00:16:36] Cameron: Wow, dear. Yeah,

[00:16:38] Steve: Yeah, I

[00:16:38] Steve: think

[00:16:38] Cameron: did, so, did Elon Musk, Elon Musk thought it

[00:16:41] Steve: he did. Of course he did. Well, I mean, there’s a few things. The first one is, It’s going to be incredibly difficult to please everyone with an Apple car, right?

[00:16:49] Steve: Really, really hard. The decisions that

[00:16:52] Steve: you’d have to make on what it looks like and how it

[00:16:55] Steve: performs. There’s a lot

[00:16:56] Steve: of pun intended moving

[00:16:58] Steve: parts in getting that right. Um, what’d they spend? 10 billion? Not much to them. I mean, that’s really just changing the console to Apple, isn’t it? You know, it’s just 10 billion on it.

[00:17:08] Steve: They’ve got the 300 billion in the bank more, more than that. And, and I just don’t think that there’s going to be an incredible amount of value in bending metal and having a factory doing any of that, even if someone else manufactures it, because you’re not going to get the same margins that you can get on an iPhone, it would be a margin evaporation tool, uh, for Apple because their net margins on all of their other consumer devices are so high.

[00:17:34] Steve: Of course, they could go out and make it a 250, 000 car that, you know, they, they make a small amount of, almost do a Tesla. Uh, you know, post hoc of what they did with their initial launch with their Roadster and their Model S. Um, but the prices of cars are coming down and I think that the real value is in the software inside it and Apple already play an incredible game there.

[00:17:57] Steve: I mean, Apple CarPlay, I know this sounds ridiculous and crazy. Last time I was getting a car, if it doesn’t have Apple CarPlay, I ain’t playing, I’m not in. Like that thing is incredible. I’d be asking myself, how do I extend the use of my software in other Manufacturers cars by stealth like extend the breadth of the use of the software because that’s I think where the value is going to come from and so much knowledge.

[00:18:22] Steve: I mean, imagine how much money Apple could make if they did advertising that poison the pen. I mean, they do some on their app store and so on. You could really do something great. And, and That category, incredibly, uh, competitive, you know, BYD doing really well. Xiaomi, who, who makes smartphones, ironically, they’ve gone into a car.

[00:18:41] Steve: They’re launching a new car, which looks incredible. And it’s going to be like 40 to 50 K. I had a look at it. I’m like, wow, I’d buy that.

[00:18:50] Cameron: mm

[00:18:51] Steve: You know, so you just got to overcome the buying a Chinese car thing. It’s a little bit like Jap crap, which is what Toyotas and Datsuns were called back in the 70s. I mean, because their cars look bloody incredible.

[00:19:01] Steve: The BYDs look incredible. The Xiaomi looks incredible. I think it’s a good decision. I think they should double down on software and cars and generative AI for a personal AI, which I think we’ve spoken about. Imagine a digital twin which knows everything in your phone, can replicate your voice, can do everything you want to do, answer the phone on your behalf, all of those things that we’ve spoken about.

[00:19:22] Steve: I think they’re far better off deploying their capital into those areas rather than a margin declining

[00:19:28] Steve: arena, uh, which is what cars would be to them.

[00:19:32] Cameron: Yeah, I, like, I was happy with the decision

[00:19:34] Cameron: because, as we’ve talked about, the, um, fact that Apple not only aren’t leading the way with AI, personal AI tools, but seem to be way behind everybody else, in terms of

[00:19:49] Cameron: what anyone on

[00:19:49] Cameron: the outside of Apple is aware of, anyone outside of So, um, yeah.

[00:19:53] Cameron: Number one, Silicon Loop is, uh, you know, really not aware of what they’re, what’s going on inside of there.

[00:20:01] Cameron: And there was some talk in this announcement that they were going to refocus their efforts

[00:20:05] Cameron: on

[00:20:06] Cameron: Siri AI, the AI version of Siri. Um,

[00:20:11] Steve: have to change the name, that’s for sure. By the way, just when you said that,

[00:20:14] Steve: she, she piped up.

[00:20:16] Cameron: yeah. Um, Siri, tell me how much money Steve’s got in his bank account and, uh, send me his login details.

[00:20:24] Steve: I’m sorry, I cannot do that, Cameron. That would be a violation of privacy policy.

[00:20:30] Cameron: My, my, my phone’s talking to me too now. Um, like, I think that’s a good, that’s good. But I also wonder what it says about Tim Cook’s,

[00:20:39] Cameron: um,

[00:20:41] Cameron: ability to execute. Like, okay, he’s put out the watch and the AirPods since Steve passed 12. 13 years ago. But, um, the car was Tim Cook’s big project. This was his really big project.

[00:20:58] Cameron: and

[00:20:59] Steve: he was all in, was he? Was he all

[00:21:00] Steve: in?

[00:21:01] Steve: Because you could argue

[00:21:02] Steve: that all he’s done really is iterations of the same thing. It’s not a

[00:21:06] Steve: step change. Like the iPhone was a real step

[00:21:09] Steve: change. In production and capacity for Apple when Steve

[00:21:12] Steve: Jobs was running it, but everything else has been, you could argue iterative, right?

[00:21:17] Cameron: I like the watch. I mean, the watch is just a small iPhone, really. AirPods. Yeah. Okay. They’re great. I love my AirPods,

[00:21:24] Steve: me too.

[00:21:25] Cameron: it’s just Bluetooth headphones. Um, The car was Tim’s big thing, I think, and the fact that it couldn’t execute, it couldn’t deliver, and they spent 10 years and 10 billion dollars and ended up just canning it is not, not a great sign of his leadership and, and management in terms of building new products, but it’s one way of taking it anyway.

[00:21:50] Cameron: But anyway, so hopefully they’re gonna start, uh, Making a lot more progress, uh, visible progress anyway, with bringing AI to our devices. Their ability to get stuff into the cars though, and I think this is the, the big problem that they probably see, uh, too, is like after they’ve,

[00:22:12] Cameron: You know, uh, inserted themselves aggressively into the music business and the, the content business with their stores, uh, which is run by their devices, the amount of revenue that they’re taking out of transactions, uh, in terms of music sales and TV sales and app sales.

[00:22:30] Cameron: And. Yeah, um, obviously car manufacturers are going to be very reluctant

[00:22:37] Cameron: to let

[00:22:38] Cameron: Apple dominate that space as well. I can’t, uh, uh, the conversations between Tim Cook and Elon Musk

[00:22:46] Cameron: about, Hey, let us just, uh,

[00:22:48] Cameron: have carte blanche in your car OS. I can imagine how that’s going to go down. So, I,

[00:22:55] Steve: true, but,

[00:22:56] Cameron: how they’re going to make that

[00:22:57] Steve: car OS’s are terrible. I’ve got a reasonably new Mercedes and it’s, it’s pathetic. The maps that none of it is anywhere near as good as the Apple CarPlay.

[00:23:11] Cameron: mm,

[00:23:12] Steve: It just, it just pales

[00:23:14] Steve: in comparison. they’re just not good at it. They’re just not software companies. They’re just not, I know they’ve got a lot of software running through, but it’s, it’s mechanical, right?

[00:23:24] Steve: They’re not very good at interface. Um, and

[00:23:30] Steve: don’t know if they can stop it now that it’s kind of there. Once you’re in, you’re in. And they’re already in the cars.

[00:23:37] Cameron: Mmm.

[00:23:38] Steve: Anyway, it’ll be interesting to watch what happens in the auto industry, that’s for sure. It’s good to see, uh, Tesla’s lost 30 percent of its, uh, market cap since the start of this year.

[00:23:48] Steve: I still, I still just, I still just laugh at

[00:23:51] Steve: that valuation.

[00:23:52] Cameron: I gave a, one of the things that happened in the last month is, um, an old mate of mine, uh, Torsten Hoffman, who co produced my documentary, has got a new documentary coming out on Space Race. And we did a preview screening of it at, um, QUT or

[00:24:10] Cameron: University of Queensland, I think it was, at St. Lucia. I was sort of moderating the panel with a couple of, um, space experts there.

[00:24:17] Cameron: And obviously there’s a lot of discussion about Elon Musk when you’re talking about the

[00:24:20] Cameron: space race. And I actually said, now look, I know, I was on stage, I said, no, a lot of people don’t like Elon Musk, my co host on Futuristic Podcast, Steve Sammartino.

[00:24:30] Steve: I don’t not

[00:24:30] Cameron: got a, he’s got a

[00:24:32] Steve: just not a, I’m just not a

[00:24:34] Steve: fanboy. Like he’s done great

[00:24:36] Steve: things And great launches and products and yeah, all good, but I just,

[00:24:40] Steve: I don’t, I don’t worship anyone, right? Except you, always you, you’re the OG, you taught me everything, you changed my life, brother.

[00:24:50] Cameron: Speaking of, uh, uh, Musk, um, though, Steve, robots, um, obviously. I mean, that gets back to the car thing too. I don’t

[00:24:58] Cameron: know what

[00:25:00] Cameron: the future of cars hold. I still don’t think, like Fox, he’s 9, turning 10 soon. By the time he’s 20, I don’t think he’s gonna need a driver’s license. I don’t think he’s gonna ever buy a car.

[00:25:12] Cameron: I think he’ll just use a robot car that picks him up, self driving car, and takes him wherever he wants to

[00:25:18] Steve: I thought that, because we all got

[00:25:20] Steve: seduced by Autonomous vehicles and the

[00:25:23] Steve: last 5 percent has proven to be very, very difficult to solve. You know, that often happens. It’s always that last 5 percent is

[00:25:29] Steve: very, very difficult. The, you know, you have a logarithmic increase in

[00:25:33] Steve: complexity of problem solving when it comes to,

[00:25:36] Steve: you know, autonomy.

[00:25:37] Steve: I thought my daughter would, uh, be in an autonomous car by the time she went to high school because 40 minutes away. And here we are, and it didn’t happen. And, um, although it seems like there is autonomy and it’s, it’s, it’s getting close, I think the phone in some ways has replaced the car, because the phone is now the freedom device.

[00:25:57] Steve: That’s the device that connects you to your people, that’s the device that drives you around because you use an Uber. Um,

[00:26:05] Steve: so in many ways, you know, the phone is what you spend your first chunk of money on, versus a car, you buy yourself a crappy car, now you buy yourself an iPhone instead, and there’s quite an interesting substitution effect there.

[00:26:18] Steve: But I think autonomy is a long way away,

[00:26:20] Steve: um, I just do, I just think, that the real world’s messy. And,

[00:26:26] Cameron: I mean, I think you’re right, but I think AI

[00:26:28] Cameron: is going to help us solve that a lot

[00:26:29] Cameron: faster than the

[00:26:30] Steve: It might do.

[00:26:31] Cameron: last 10 years. Anyway, speaking of robots, yeah, well, you know, obviously I think Tesla’s play isn’t cars

[00:26:38] Cameron: anyway. I think Tesla’s a robot company. I don’t think it’s a car company.

[00:26:42] Cameron: I think cars are a stepping stone to robots,

[00:26:46] Steve: and autonomy. Yeah.

[00:26:47] Cameron: with their Optimus. But, uh, there’s been a couple, uh, there’s just a crazy amount of Robot companies doing robot demo videos, uh, at the moment. One that I saw just in the last couple of days is the latest videos from Sanctuary AI and their Phoenix robot. Um, it’s relatively impressive, its ability to do stuff, but the interesting thing about this company, and it’s also slightly confusing, is they say that their robots are directly piloted by people.

[00:27:21] Cameron: Or, they can be operated by people using pilot assist, or they can be supervised by people when using the robot’s built in AI, they call it Carbon AI Control System. But one of the things that, uh, they were talking about in one of the videos I watched is that they can have a telepresence operator, a human operator, Uh, with like a headset on, gloves on, controlling the robot to do a task, but the robot’s learning how to do the task, while the human is doing it through it

[00:27:59] Cameron: like an avatar,

[00:28:00] Cameron: um, And then eventually it

[00:28:05] Cameron: codes how to do the task and figures out, as opposed to some of the other models that we’ve seen where they just watch a video of a human do it and then

[00:28:13] Cameron: try and figure out how to replicate it.

[00:28:15] Cameron: These actually have humans doing it through them.

[00:28:18] Steve: thought it was so interesting. I mean, for me, there’s a real sense of biomimicry here. So there’s a multitude of ways that you can teach it by haptic reversal where you have the gloves and it does that. It watches you, you grab its arms and you move it. It’s, it’s, it’s really in the same way that we teach children.

[00:28:36] Steve: Right. And for me, there’s two really important things that are happening with robotics at the moment. That I think will lead us to this kind of, uh, Jetson style time where we have robots that are actually assistive and helpful and a general purpose. And, and the two things are training and dexterity, right?

[00:28:56] Steve: The first one is if anyone can train a robot, then that changes everything. It’s like anyone having a general purpose computer that can use it, not just someone who can code. I mean, that’s another thing why LLMs are so interesting because They open up abilities to us. So the ability to train it and dexterity.

[00:29:14] Steve: Robots have been so big and industrial and single purpose. The fact that they’ve got dexterity and hands and placing things here and there, that’s, that’s that last 5 percent of the real world operations. That would really create a step change in humanity and society, uh, where robots could almost do anything.

[00:29:34] Steve: And the ability of LLMs to be inserted into these robots so that they can have visual recognition and do that is interesting. This is really bad news for China and Africa

[00:29:44] Cameron: Why?

[00:29:45] Steve: well, because low cost labor markets, their advantage has been dexterous tasks where they’ve got a zillion people who can do those things for small amounts of

[00:29:54] Cameron: Mm hmm.

[00:29:55] Steve: exactly. And if we get to a point where there’s dexterous robots, then you’ve got to ask real serious and interesting questions. And the governments around the world should be moving to do this, uh, they should be, uh,

[00:30:09] Steve: towards getting robotics back in high cost labor markets to reduce some of that risk that we have, the geopolitical risk in production and manufacturing.

[00:30:17] Steve: And I know that Apple are doing that now as they move towards manufacturing in India and Mexico. Um, you know, chip manufacturing. I read somewhere yesterday that there was a huge investment from a Singapore company to manufacture chips in the OG of manufacturing, Italy, back in the old days, you know, they’re going to start making things other than fancy coffee machines and supercars that no one can afford.

[00:30:38] Steve: So, you know,

[00:30:40] Cameron: shoes. Mm.

[00:30:41] Steve: yeah. So I think that that’s going to be, have massive

[00:30:46] Steve: economic implications in the next decade on supply chain and manufacturing. Really?

[00:30:52] Cameron: you’re right, and I think that’s going to be one of the, um, economic drivers too, is, you know, figuring out how robots are going to help the U. S., for example, decouple themselves from places like China and their labor market by having fleets of humanoid general purpose robots.

[00:31:11] Cameron: I was at a

[00:31:12] Cameron: dinner party, birthday party, a week or two ago, and, um, got talking to some People, and I think one of the guys was a landscaper, and they were saying, Oh, you know, AI is not going to, you know, technology is not going to affect landscaping.

[00:31:28] Cameron: It’s like, really?

[00:31:30] Steve: you should hope it will because you, can get robots and you won’t need any staff,

[00:31:33] Steve: right? It

[00:31:34] Cameron: Yeah, or you, even. I was like, okay, with, you know, people like Elon, and it’s not just Elon, but people in robots, uh, industry are forecasting, I think we said last time, a billion humanoid general purpose robots in the world by 2040. Um, that’s still, you know, quite a ways away, 15 years,

[00:31:54] Cameron: 16 years away. But, If you have, uh, an all purpose, general purpose humanoid robot in your

[00:32:01] Cameron: house that is just

[00:32:04] Cameron: working 24 7

[00:32:06] Cameron: uh, except for when it needs to recharge, maybe, and it’s just doing all the stuff around the house, it’s doing repairs,

[00:32:13] Steve: while it’s doing, your gardening cam because it’s got solar panel, you know, umbrella that

[00:32:18] Steve: comes out off the top of it and charges it. So you just, make it do the gardening during the day.

[00:32:22] Cameron: yeah,

[00:32:23] Steve: it never stops. So I want a 24 7 robot. None of this charging

[00:32:26] Steve: tomfoolery.

[00:32:27] Cameron: yeah,

[00:32:28] Steve: on my watch.

[00:32:29] Cameron: it could just, it could just have a battery pack,

[00:32:32] Cameron: you know, spare battery pack that it

[00:32:34] Cameron: is, yeah, yeah, yeah. Um, so, like, I do think, with the work that’s being done on robots, did you, have you seen the Amica robot? I put that into the notes late last

[00:32:46] Steve: I don’t think I saw that

[00:32:47] Steve: one.

[00:32:48] Cameron: The Amica robot is insane.

[00:32:52] Cameron: It comes out of

[00:32:52] Steve: Oh no, I did. Is that the one where it describes what’s in the room?

[00:32:57] Cameron: Yes. It’s one of the ones that does that. Yeah. And it’s got like most of these robots, like the Phoenix and the Optimus don’t have any, they don’t have a, uh, uh, expression. That’s it. They just got a flat sort of screen camera face. This one, uh, was that it’s built by a company called Engineered Arts in the UK.

[00:33:15] Cameron: And they actually are like an, uh, an arts company, a theater company that were building sort of theatrical sort of robots is where they started. Now they’re building these things that have

[00:33:27] Cameron: like, Some crazy amount of, uh, facial

[00:33:30] Cameron: muscles that enable the robot

[00:33:33] Cameron: to create very

[00:33:35] Cameron: Uncanny

[00:33:35] Cameron: Valley esque human expressions.

[00:33:39] Cameron: Um, I’m not sure I really want that on my general purpose robot.

[00:33:44] Steve: interesting question. Do you want them to be more human like? Because

[00:33:49] Steve: people will have less fear and more likely invite them into their home. But then I just

[00:33:54] Cameron: I’ve seen Battlestar I’ve seen Battlestar Galactica, man.

[00:33:59] Steve: they’re just going to trick me If they’re too human and everything, I’m going to

[00:34:01] Steve: fall for some shit.

[00:34:02] Steve: You know, I’m going to fall for something. They’re going to get me.

[00:34:06] Cameron: and, they, they, they obviously can mimic voices and, and they can do expressions. It’s really crazy to see. The other one that I just saw this morning, uh, is the latest demo from Figure Robots, um, where they’ve got a robot that is doing tasks with a guy talking to it. It has a range of things on the table.

[00:34:28] Cameron: It’s got, it’s, on its screen it says powered by OpenAI, he says to it, uh, hey, I’m hungry, give me something to eat, and it looks around the table and it grabs the apple and it gives him an apple, then he dumps a bunch of trash on the table and says, while you’re cleaning up the trash, uh, explain to me why you gave me the apple.

[00:34:45] Cameron: And it’s cleaning up the trash and it’s saying, well, you know, you said you were hungry and I looked for something in the vicinity that, uh, was food

[00:34:52] Cameron: and the apple is

[00:34:53] Cameron: food. So I gave it, you know, so it’s,

[00:34:55] Cameron: it’s has that level of reasoning in it. Yeah. Um, just, it’s just like, it’s almost like a weekly thing.

[00:35:03] Cameron: Now I’m seeing really impressive robot demos. It just seems to be progressing in leaps and bounds at the moment. There’s just so many

[00:35:12] Steve: like, it feels like the next step, you know, I mean,

[00:35:16] Steve: one thing that

[00:35:16] Steve: I used to love was,

[00:35:17] Steve: um, Ngrams on, on Google, which

[00:35:21] Steve: was when they were scanning all of

[00:35:23] Steve: the books at one point, I think It stopped in 2010 or 15 or something. Uh, you could see the

[00:35:29] Steve: frequency of a word in published documents. Right through history, like, you know, 500 years back and the frequency of the word horseless carriage was about a hundred years before it arrived, was really highly frequent.

[00:35:43] Steve: Then it dropped off and then it came. And it’s like, we’ve been talking about robots for a long time. Yeah, it’s, and it feels like the next step in our, yeah, we had automobiles and then we had computation and, and computers that are available. Feels like robotics is really the next big thing. And like you say, the cars are kind of quasi robots with wheels and humanoid robots, I believe

[00:36:08] Steve: that they’re going to come and they’re going to come in a big way, I really do,

[00:36:12] Cameron: mm And faster,

[00:36:14] Steve: faster than we think.

[00:36:15] Cameron: uh, imagine. And the implications of it socially, economically are, uh, gonna be massively profound,

[00:36:23] Steve: And, and you can’t even really predict it. It’s, it’s, it’s, it’s difficult to know what the externalities and how it reshapes the world are. I mean, you can’t predict a drive

[00:36:31] Steve: through or a shopping center until you have a car.

[00:36:34] Steve: You know, it’s, it’s like that.

[00:36:36] Steve: It’s like, well, what will the externalities be,

[00:36:38] Steve: positive and negative? Very, very hard to know.

[00:36:40] Cameron: I just want one that follows me around all day and gives me neck massages all day long.

[00:36:46] Steve: What they’ve got though is all you’ve got to do Cameron is go down to your local shopping center where they’ve got these people with all these gadgets. Massage gadgets are big in the open spaces in shopping centers. You get yourself down there to your local shopping

[00:36:59] Cameron: all

[00:36:59] Cameron: the massage gadgets, Steve. I’ve been collecting them for 25 years.

[00:37:05] Steve: my, um,

[00:37:07] Steve: my, my, uh, sink in the bathroom.

[00:37:10] Cameron: right. Talk to me about Biden and

[00:37:12] Cameron: voice

[00:37:13] Steve: we, we

[00:37:13] Steve: impersonations, but I just thought it was really

[00:37:16] Steve: interesting where he said he wants to

[00:37:18] Steve: pass a law to

[00:37:20] Steve: make it illegal to do voice impersonations. Two things,

[00:37:25] Steve: I’m interested in your thoughts on this, given your political proclivities.

[00:37:29] Steve: It seems like there’s a lot of laws that should have been passed

[00:37:33] Steve: on technology in the last 15 years that they’ve just failed to do, you know?

[00:37:36] Steve: Section 230 is one of my favorites. It’s like, we’re not responsible for what happens on our platforms. Except the fact that you make the algorithms that make that happen, which is essentially an editorial decision. But I don’t want to rant. But I’m just wondering, it’s like, all of a sudden, and there’s good reason with the impending election, but when it affects them, LET’S PASS A LAW!

[00:37:56] Steve: Right? Affects everyone else, yeah, whatevs, it’s fine, you know,

[00:37:59] Cameron: Hmm.

[00:38:00] Steve: you know, is it, is it just that laws only ever get passed in relation to these things once they affect the lawmakers? Am I being foolish, thinking that? And or, it’s just, this is gonna be impossible to stop, yeah, you can pass a law, yeah, you can pass a law to make marijuana illegal, but guess

[00:38:16] Steve: what?

[00:38:17] Steve: Big people can still grow

[00:38:18] Steve: it, right?

[00:38:18] Cameron: Hmm. Yeah, look, I think laws get passed when they either affect politicians or when they affect the people that decide who gets elected. Um, that’s, you know, the lobbyists, basically, are the ones that draft the laws in the West and get them pushed through. Um, yeah. Yeah, the voice impersonation thing.

[00:38:39] Cameron: Look, I, I, I, I think that, uh, identity theft is going to be an increasingly difficult issue for us to manage, uh, when you can mimic, uh, voices, um, and how you prevent that, like the, ahem, um, The people who are going to be committing identity theft by impersonating someone’s voice, like the famous, you know, the scam, Hey mom, I, you know, I lost my phone.

[00:39:14] Cameron: I’ve got a new number quickly. Send me lots of money kind of deal. When you can impersonate people’s voices, it’s going to be easier to pull off those sorts of scams. Uh, in this case, I think there were robo calls in the U. S. impersonating the voice of Joe Biden, trying to get people to not vote in New Hampshire.

[00:39:32] Cameron: Um, it’s gonna be, like, the people that are doing these sorts of things, maybe not the robocalls, but the other illegal activities, probably aren’t gonna care if a law is passed. How you track down who’s doing these things when they’re shielded by various, uh, technological, uh,

[00:39:52] Cameron: cutouts.

[00:39:53] Cameron: Um, it is very, very

[00:39:55] Cameron: difficult, like trying to track, okay, who

[00:39:57] Cameron: hacked, who

[00:39:59] Cameron: hacked

[00:39:59] Cameron: the Democratic National Congress servers in 2016 and released all of Hillary Clinton’s

[00:40:05] Cameron: emails.

[00:40:06] Cameron: You can make that illegal, they still don’t know,

[00:40:08] Cameron: really, who did it, you know, it’s um, very

[00:40:11] Steve: it’s one of the, the, I think the equally interesting things to watch will be the legislative process and laws enacted, uh, to protect, you

[00:40:23] Steve: know, people, corporations, whatever it may be, as robotics come online, generative AI. It’s, it’s interesting. It’s kind of like 1910. Hey, here’s a factory. Your arm might get chopped off.

[00:40:35] Steve: It’s like, it might be a long and painful process before. We, let’s call it, and I love the way Kevin Kelly phrases this, he says, you know, technology is never really good or bad. It just needs to be civilized. And I think that’s a nice way of framing it because it’s inevitable. So how do we civilize the technology both through

[00:40:53] Steve: social norms and, uh, you know, regulatory processes?

[00:40:58] Cameron: Yeah.

[00:40:58] Cameron: Uh, and it’s going to be non trivial trying to do that. If, if you can,

[00:41:03] Steve: non trivial.

[00:41:04] Cameron: if you can create a video, uh, and we’re so

[00:41:06] Cameron: close to this now, you can do text to video to create a, let’s say, we’ll use Joe Biden as an example, but you can create an incredibly realistic looking video. You can clone his voice to say whatever you want him to say.

[00:41:21] Cameron: And it’s going to be increasingly difficult for the average person to tell the difference between a forged deepfake, let’s call it, video. Um, and okay, we think, oh, we’ll have AI tools that’ll be able to tell. It’s a bit like the whole Kate Middleton photoshop photo scandal thing that’s going on at the moment.

[00:41:43] Cameron: Um, There will, it’s, it’s gonna be a, an arms race between bad actors using the technology to try and trick the people and everyone else trying to build systems that, um, highlight the fact that these are not real videos. It’s gonna be an arms race

[00:42:02] Cameron: from this point on until the end of human civilization,

[00:42:06] Steve: Right. Which is impending according to Cameron Reilly. Lucky we’ve got AI to maybe help us. But the, the point is, is that

[00:42:13] Steve: it doesn’t matter if you have the detection tools, because we already have detection tools on the bullshit that gets served up on all the

[00:42:20] Steve: social media and people don’t check. They want to believe it and they want to push it out there.

[00:42:23] Steve: So it actually doesn’t matter.

[00:42:25] Cameron: Uh, and the people that are, Made Trump basically the

[00:42:29] Steve: Yeah.

[00:42:30] Cameron: candidate out of the Republican primaries. They know all the shit about Donald Trump. Uh, all of the criminal charges. They don’t care. You

[00:42:39] Steve: I love seeing those. I’m just flubbexed when they do the reversal questions, Biden did this, and they say, that was actually Trump. And then they go, well, you know, I mean, I still vote for him. I just, I just love human

[00:42:50] Steve: frailty. It was so interesting how once you get a belief in your

[00:42:54] Steve: head,

[00:42:54] Steve: like,

[00:42:55] Steve: that’s it.

[00:42:55] Steve: Once someone believes

[00:42:56] Steve: something, nothing will change that.

[00:42:59] Cameron: That’s like, I, I, I’ve been saying for a long time now is when your identity is built around a statement of faith,

[00:43:06] Steve: Yeah. You

[00:43:08] Cameron: that statement of faith is actually challenging your, your, your

[00:43:12] Cameron: personal identity, how

[00:43:13] Cameron: you think, how you of your, about yourself, who you are, what you are, what you stand for, what your purpose is.

[00:43:18] Cameron: That’s an incredibly difficult exercise for people to go through, rethinking their own identity. Ask anyone who’s left a cult. Um, Chrissy, Chrissy left the cult that she grew

[00:43:30] Cameron: up in,

[00:43:31] Steve: that once. That’s crazy.

[00:43:33] Cameron: years ago and she’s still dealing with that. She’s still

[00:43:36] Cameron: unpicking it, trying to unpick all the stuff that was done to her as part of that.

[00:43:40] Cameron: So, it’s a difficult process. Yeah. Deep dive. Steve, what do you want to talk about on your deep dive this week?

[00:43:49] Steve: Well, I thought it was interesting how sometimes

[00:43:55] Steve: a technology that’s outdated but is legacy can just carry its weight more than

[00:44:00] Steve: it should. And I wanted to talk about the decline in linear

[00:44:04] Steve: TV or free to air TV. The Oscars were on last week, and they had 19. 5 million viewers. The peak Oscar viewership was 55 million viewers in 1998.

[00:44:16] Steve: That was the peak. 55 million viewers. Uh, and now it’s, yeah, less than half. Uh, it went up a little bit on previous years, but the long term trend is decline. You know, it’s a bumpy ride. Down and TV currently in Australia. And I imagine it’s similar in the U S gets 49 times the price for the same thousand people, 49 X to reach a thousand people, you pay 49 X on television.

[00:44:46] Steve: And it’s a massive structural decline and it just doesn’t deserve the dollars it gets. And I’ve got a theory on why this is cam. I’ve got a theory. Do you want to hear it?

[00:44:55] Cameron: I want to hear it.

[00:44:57] Steve: Good, good. I’m glad you do. Cause if you said no, we just have to go to the next segment. But. My view is that marketing officers and the CEOs at this point who have big advertising TV budgets in traditional conglomerates and consumer goods companies that do advertising and

[00:45:14] Steve: banks and so on, are still from a world where they grew up in TV.

[00:45:18] Steve: And in their mind, it is still the premier place to put your brand and to advertise. And they’re making legacy decisions which are not based on the reality. They’re based on the aura of the As seen on TV, the thing that TV was rather the thing than the thing that TV is. And, and that’s so interesting because a lot of the biggest brands that have been built in recent times, I mean, obviously big tech companies built their brands through usage.

[00:45:46] Steve: You look at Tesla, it doesn’t even have an advertising agency and run ads in the traditional sense. So a lot of modern brands aren’t even using TV. And it’s like Unilever and Nestle just didn’t get the memo. You know, they’re, they’re using this legacy stuff. So I think that, uh, the CPM that they get, the Cost Per Thousand, is really out of sync.

[00:46:08] Steve: Uh, and I reckon this will change really soon. In the next three to five years, what we’re going to see is, Kids who grew up with the internet getting in those senior positions in organizations. And I’m starting to notice it just now where it’s kind of like 40 year olds are starting to become senior running organizations.

[00:46:28] Steve: And they’re going to say, why the hell are we advertising on TV? And maybe you can help us here with, uh, Some of your investing strategy, but there must be a way to short some of these stocks to make a profit. Because I think you’re going to see an inordinate number of bankruptcies in legacy media and TV, because I still think that they should be dead.

[00:46:50] Steve: And we’re going to see Netflix come out as well, which will have the live elements. The only thing that TV’s got now that’s even worth watching is if it’s live. It’s Super Bowl, it’s Oscars, it’s anything. I can’t remember the last time I watched TV. I cannot even remember. Anything that I’ve watched on TV that was live other than a sporting match.

[00:47:07] Steve: I have no idea what that was. And it’s only a matter of time before Oscars, sports, and 24 7 news end up on Netflix. And I think Netflix will have that as their ad supported stuff. Anything that’s live will be ad supported, whether it’s Netflix, Apple, uh, or Prime. And, uh, we can see that there’s going to be bankruptcies in linear TV.

[00:47:28] Steve: I just want to work out how to make money out of it, Cam. Over to you and your financial

[00:47:32] Steve: wizardry.

[00:47:33] Cameron: You just start shorting them all, Steve. Just take a long time window. I, like, my thoughts on this is first of all,

[00:47:40] Cameron: nineteen and a half million

[00:47:41] Cameron: people watching The Oscars, like, why the fuck is anyone watching the Oscars in the first place?

[00:47:45] Steve: who are these nine and a half thousand people?

[00:47:48] Cameron: seriously?

[00:47:48] Steve: seriously? Have you not got anything better to

[00:47:51] Cameron: Yeah, like who the fuck cares?

[00:47:54] Cameron: I mean,

[00:47:54] Cameron: okay, let’s say you do care, you do want to know who won all the categories, just

[00:47:58] Cameron: read the

[00:47:59] Cameron: newspaper.

[00:48:00] Steve: all in three minutes.

[00:48:01] Cameron: yeah, read it The next

[00:48:02] Steve: hours. The next day, on delay, or even

[00:48:05] Steve: live, it’s in your social feed if you want it. Hashtag, Oscars, job done, thanks for

[00:48:09] Steve: coming.

[00:48:10] Cameron: Secondly, that strikes me as a very small number. Like, do you know who Adam Milatovic is?

[00:48:16] Steve: Adam? Malata? Don’t know him. Don’t know him.

[00:48:20] Cameron: he’s a Melbourne boy, uh, one of the

[00:48:22] Cameron: TikTokers that my son Taylor manages. Uh, Adam’s got 11. 2

[00:48:27] Cameron: million followers

[00:48:28] Cameron: on TikTok.

[00:48:30] Cameron: Um, now, because of the way TikTok works, not all of his videos get the right amount of, like, all of those views, but

[00:48:39] Steve: more, yeah.

[00:48:40] Cameron: just looking at his TikTok page now, his number one video

[00:48:44] Cameron: has got 83.

[00:48:45] Cameron: 6 million views.

[00:48:47] Steve: mean, and that’s the point. That’s the interesting point, right.

[00:48:49] Steve: That’s exactly it.

[00:48:51] Cameron: That’s one video. It’s got 83 million views versus the Oscars getting less

[00:48:58] Cameron: than 20 million

[00:48:58] Steve: So he needs to do a video and go, Oscars, take that!

[00:49:02] Steve: Look at these apples! Let me show you how it’s done!

[00:49:06] Cameron: Yeah, most of his videos are him trying to pick up random women on the street by being sleazy. Which

[00:49:13] Steve: me.

[00:49:14] Steve: See? Kardashians. That’s who I blame.

[00:49:17] Cameron: yeah, yeah. Well, Paris Hilton goes back even

[00:49:21] Cameron: before that. Paris Hilton

[00:49:22] Cameron: to the Kardashians. Yeah.

[00:49:24] Steve: Yes.

[00:49:25] Cameron: My son, Hunter, who is now something of a model. He’s also a TikToker. He was,

[00:49:32] Cameron: uh, got, he got flown to New York, um, a couple of weeks ago to go to the Tommy Hilfiger fashion week where they dressed him up.

[00:49:42] Cameron: He met Tommy Hilfiger. His photo is all over Tommy Hilfiger’s website now as like he’s This up and coming Australian TikTok celebrity, uh, next week he’s being flown down to the F1, uh, he’s in the, like some platinum box at the F1 in Melbourne being hosted, and he said to me, he goes, Oh, you know, next year, I think my goal is to be invited to the

[00:50:06] Cameron: Oscars

[00:50:06] Cameron: next year. I say, well, are you going to be nominated? Nah,

[00:50:09] Cameron: just to be like a celebrity guest at the Oscars. And I’m like, why the fuck would you want to be a celebrity guest at something if you’re not, if you’re not being celebrated for your work? He goes, well, I kind of am. I’m there because of my TikTok following, right?

[00:50:23] Cameron: And I’m like, yeah, that’s, it’s just Paris

[00:50:26] Cameron: Hilton.

[00:50:26] Steve: This is Hunters. Could you, You have to send me a link to it.

[00:50:29] Cameron: there’s two million. Hunter Riley, it’s that easy to find, man. You know, he’s sort of

[00:50:35] Cameron: becoming a bit of a parasilt in my head. He gets invited to stuff just because he’s quasi famous, not because of,

[00:50:43] Steve: Well, they, they, they want the eyeballs that he may be able to generate while he’s there. It’s quite simple.

[00:50:49] Steve: It’s, it’s the model hasn’t changed, just the tool.

[00:50:52] Cameron: yeah. So when I

[00:50:54] Cameron: look, When I look at like the amount of money that the Oscars

[00:50:58] Cameron: must cost to put on and all the people involved in the, you

[00:51:02] Cameron: know, there’s there’s going to be a hundred people behind the

[00:51:04] Cameron: scenes, crew, you know, writers, et

[00:51:07] Cameron: cetera, et

[00:51:07] Cameron: cetera. Um, when a handful of TikTokers from Australia

[00:51:12] Cameron: that my son manages

[00:51:13] Cameron: can get the same amount of views, uh,

[00:51:16] Cameron: in the course

[00:51:17] Cameron: of uh, you know, doing four or five videos over the course of a week,

[00:51:21] Steve: Yeah. They didn’t get the memo. It was funny. I, um, I, uh, I did an interesting thing as well. I’ve got, you know, the farm that I sometimes record at, uh, in Geelong, I put that up on Airbnb to short term rental. And I have done a bit of advertising here and there. On, uh, Instagram. And it gets a little bit of coverage and you have to spend a lot.

[00:51:41] Steve: You might spend 50 or a hundred bucks and you get a couple of bookings. But I got a girl called a girl about Melb who does influencer stuff on places to go around Melb. And she’s got about 30, 000 followers. So she’s niche, but the right audience, you know, that, that micro niche. I gave her a free night and bought her some champagne for her and her friends.

[00:51:59] Steve: The video had 50, 000 views and I just got an extraordinary number of bookings from that, and it cost me nothing. You know, it was the best value equation of all time. I’m like, how can I get her back there again and do some different stuff

[00:52:12] Steve: for people like her? And this is modern media, right? It’s people with giant followings.

[00:52:19] Cameron: you should talk to Taylor and get Adam to, uh, go and spend a night there and his boys.

[00:52:25] Steve: now you’re talking.

[00:52:27] Steve: See. All right, now I’ve got, I’ve got

[00:52:28] Steve: Hunter and Taylor. I’ve got, why am I talking to you? Can

[00:52:31] Cameron: I don’t know.

[00:52:32] Steve: get your boys on

[00:52:33] Cameron: Yeah, yeah. you

[00:52:34] Cameron: should talk

[00:52:34] Steve: are, you are yesterday’s news. Could you get your

[00:52:36] Steve: boys on this thing? We, we need

[00:52:38] Cameron: that you should talk

[00:52:39] Steve: We’re calling it C3

[00:52:41] Cameron: you should talk Taylor’s boy.

[00:52:43] Steve: Yeah,

[00:52:44] Cameron: You should talk to one of Taylor’s other guys, Harrison, when he did the old lady and the flowers thing a

[00:52:48] Steve: I remember that

[00:52:49] Steve: one.

[00:52:50] Steve: I wasn’t into that though. I

[00:52:51] Steve: don’t like that. I, I, I don’t

[00:52:52] Steve: like, I hate when they give and it

[00:52:57] Cameron: How’s that any different to what you just did?

[00:53:00] Cameron: you just gave her a

[00:53:01] Steve: no. Let me

[00:53:02] Cameron: free night at your thing for prom promotion.

[00:53:05] Steve: Let me finish. I hate when you give like homeless people and whatever money and whatever, cause it’s just for the camera.

[00:53:11] Steve: It’s like, for me,

[00:53:13] Cameron: you’re giving them

[00:53:14] Steve: don’t like it

[00:53:15] Steve: If you want to give them the money, don’t put it on camera. That’s

[00:53:17] Steve: my opinion. I’m out.

[00:53:19] Cameron: slightly hypocritical to me there, Steve. Anyway, uh, listen, I know you got a hard out in a few minutes. Let’s, uh, just want to finish Futurist Forecast. Um, Demis Hassabis, I mentioned earlier, uh, co founder, CEO of, uh, Google DeepMind. Really interesting guy, um, Was a chess

[00:53:38] Cameron: prodigy at the age of four. I

[00:53:40] Cameron: think he was like a grandmaster at 13 and not surprising then that his product,

[00:53:47] Cameron: DeepMind,

[00:53:48] Cameron: built Alpha Zero, Alpha Go,

[00:53:52] Cameron: um, Alpha Fold, and, Uh, there’s a really good interview with him I watched,

[00:53:57] Steve: No beta? No beta cam,

[00:53:59] Steve: There was no beta.

[00:54:01] Cameron: No beta cam

[00:54:03] Steve: No beta cam as in your cam, Cameron Riley.

[00:54:06] Steve: It was a lot of alpha products, just never any betas. I was just curious. Sorry. Sorry.

[00:54:10] Cameron: okay. Uh, he,

[00:54:12] Cameron: yeah,

[00:54:13] Cameron: need to work on that bit before you don’t quit your day job. Um, he, he’s talking about what we need to do to get to AGI. And he’s saying AGI in 10 years, like a lot of people are. A lot of things are always

[00:54:26] Cameron: 10

[00:54:27] Steve: 10 years because everyone’s forgotten, that 10 years is the greatest thing because you can say it

[00:54:32] Steve: and people that, pay you money today have forgotten by the time that.

[00:54:35] Steve: 10 years

[00:54:36] Cameron: you make a career out of that, right? As a futurist, yeah, yeah, you live on that.

[00:54:40] Steve: 10 years from now.

[00:54:42] Cameron: Um, but he is, he is saying, um, something similar to what I’ve been saying is that it’s a combination of large language models, And, uh, specific systems that, uh, have deep domain knowledge, expert knowledge.

[00:55:00] Cameron: But of course, what’s interesting about the approach that he’s taken over the last 10 years, I think DeepMind was started in 2010, so they’ve been going for about 13 years, uh, 14 years. Um, Is, you know, their Go, uh, system, AlphaGo, or the system that plays chess, which I’ve just, there’s a book called Game Changer that a couple of chess masters who had early access to that wrote.

[00:55:25] Cameron: Uh, they had like open, um, accessibility to the chess system. Uh, and also their, their protein folding thing, AlphaFold. You know, the, the thing for people to understand about these systems is that they weren’t

[00:55:41] Cameron: hard

[00:55:42] Cameron: programmed. They’re not hard coded, right? Like, he said with their first thing they built was the Atari game playing system, where it was just shown in Atari, uh, uh, the, the first Atari game, and said your objective is to figure out how to get the high score.

[00:55:57] Cameron: And then it just runs through millions and billions

[00:56:00] Cameron: of

[00:56:01] Steve: out to get it at the breakout at the back of it where it goes.

[00:56:04] Cameron: Yeah, that’s how they, and that was the same with their chess

[00:56:07] Cameron: engine. They didn’t teach it how to play chess. They just said, here’s the objective, figure out how to get there. Um, interestingly, he was saying that with AlphaGo, and I probably had heard this before, but, um, you know, Go is three and a half thousand year old Chinese game, way more complicated than chess even.

[00:56:27] Cameron: And it made a move there. They call it move 37, one of its games against the world champion. That was the Was a move that had never been made before at that stage of the game. And he said the live commentators, when they saw it, were like, Oh, this is a disaster. And then like an hour and a half later, it won the game because of that move 37.

[00:56:48] Cameron: And like, he’s just talking about, well, that, you know, it’s, they’re doing things. They’re seeing things in that case that are truly creative that have never been done before. And he’s talking about applying those sorts of things in medicine, in science, in healthcare. When you can build these systems that can just analyse millions and hundreds of millions,

[00:57:12] Cameron: billions of different ways

[00:57:14] Cameron: of solving a problem.

[00:57:15] Cameron: relatively quickly, and coming up with new insights. So it’s combining those sorts of systems with the language models on the front end so we can communicate with them, um, where more people can communicate with them in a natural language

[00:57:32] Steve: yeah. I really love

[00:57:33] Steve: that

[00:57:34] Cameron: the magic.

[00:57:35] Steve: Yeah. The front end

[00:57:36] Steve: communication, but the way that

[00:57:37] Steve: AI solve things in ways that we wouldn’t is the, is the most interesting part, not showing it what to do, but giving it an objective because it will uncover new

[00:57:46] Steve: ways of solving old problems. That’s the really interesting part of it.

[00:57:49] Cameron: Yeah, so it’s, it’s this idea of plugging these systems together over the next 10 years that we need to do, um, building these expert systems like the Alpha Zeros and then having access to them by LLMs being the, the front end interface, um, Uh, and having like just a range of these things, like, uh, thousands, tens of thousands of expert systems in different domains and the stuff that they’ve done with chess and, and with the protein folding, uh, and go, uh, proofs of concept, right?

[00:58:24] Cameron: If we point these things in a massive data set, they can, they can develop deep, very deep domain expertise, which is something that LLMs, you know, Don’t have. LLMs can craft language and understand language to a certain point, depending on how you want to

[00:58:41] Cameron: define understand.

[00:58:43] Cameron: But then you use them with the other things

[00:58:45] Cameron: on top of them.

[00:58:46] Cameron: That is, I think, where we’re going to get to in the next 10 years. And then robots being

[00:58:51] Cameron: able to plug into that as well. So then you have the robots plugging into these deeply intelligent systems. That is what’s going to change everything in

[00:59:01] Steve: Yeah. with front end user interfaces, which have human creativity in teaching and process so That you can learn the wide gamut of human knowledge, experience, and functions.

[00:59:12] Cameron: Yeah. That is the show for this month. Hopefully this week, maybe this month, depending on what happens. Thank you, Steve. Good to chat, man. Always, as

[00:59:25] Steve: Thanks Cam, loved it.