GPT is my new best friend; the 4 AI camps; does AI come up with original ideas?; Change in Corporate Approach to AI: Beyond Practical; DeSantis Campaign Uses Apparently Fake Images to Attack Trump on Twitter; EU Bands Facial Scanning / recongition and other Biometrics measures; AI Tour of Duty Sam Altman in Australia; AI Agents plugin for GPT; Geoffrey Hinton think that LLMs actually understand.
Doomsayers
Full Transcript.Â
[00:00:00] Cameron: Well, Steve Sammartino, welcome back. This is, uh, futuristic episode six. Uh, how are you, man?
[00:00:11] Steve: Good. This is three weeks in a row, Cameron, and I’m, I’m very
[00:00:14] Cameron: excited. I know this is a, this is a milestone in our podcasting relationship. You didn’t even text me and say you were running late this week like you normally do.
[00:00:24] Cameron: That
[00:00:25] Steve: hurts, but it’s true.
[00:00:27] Cameron: Hey, I’m giving you props that you didn’t do it. Yeah. Well, let’s start it off this week. Tell me about your week in futuristic technologies. Man, what, what really took your attention this week? I
[00:00:39] Steve: had a moment with a corporation where I’m getting briefed on a seminar I’m doing in Bali in a few weeks.
[00:00:46] Steve: With a big Australian, uh, consumer goods company, and I’ve worked with them before, and it’s usually about innovation and startups. This time it’s about ai. And what I’ve really noticed on briefings that is really different that tells me we we’re in a [00:01:00] real inflection moment is that the briefings aren’t tell us how to do this or do a workshop where there’s this outcome.
[00:01:07] Steve: The briefings are like, let’s just breathe and open our minds and really think about what’s going on here and what the implications are. Like I never have that. And so the briefing was we just want to have a deep, long, round table discussion after you do your keynote speech on AI and tell us about what you think some of the implications might be.
[00:01:25] Steve: It’s like 1994 when the graphical user interface is arriving and everyone’s getting that internet kind of thing that we’ve heard of. Actually, it might be real, that might have a big impact. And so they said that they just wanna open hearts and minds and they just wanna. Almost like hold hands and give a, you know, technological kumbaya moment of like, what is happening here?
[00:01:46] Steve: Let’s, yeah, let’s, let’s sit under a tri triangle and chant and understand this future thing. And that is, for me, it was a real moment where that people are starting to realize this is big. And that was something that [00:02:00] really stood out. They just wanted to understand the landscape, you know, what’s happening in society, what’s happening in industry that didn’t even want an outcome.
[00:02:07] Steve: And that never happened in corporations.
[00:02:10] Cameron: So you just have to play, um, talk, uh, not talking heads. It’s r e m, it’s the yen dar the world. The world.
[00:02:16] Steve: We know it. It’s the, yeah. And you know what I love in that song? And I feel fine. There’s something about that line.
[00:02:24] Cameron: It’s a great line. Yeah. It’s just
[00:02:26] Steve: where are the artists in this modern day, Cameron as two gr men postulate.
[00:02:32] Steve: What once was.
[00:02:33] Cameron: So do you think that clients like this are excited about the future? Uh, AI related stuff? Are they scared? Are they pensive? What, what’s where, where are they on your mood board? On my mood
[00:02:47] Steve: board. They’re pensive scared. They’re like little children. You know what they’re like, they’re like someone who’s about to sell their house cuz they’re moving and they want to go to a real estate agent and a conveyance and a legal person to hold their hands cuz [00:03:00] they haven’t got a clue.
[00:03:01] Steve: They’re really frightened about what it could mean. Cuz I think they’re starting to understand the breadth of the things that they do. All of it’s gonna change. And, and the reason is, and I just keep saying this, it’s because it’s language. It can horizontalized and it cha that changes everything cuz everything that we do in society is based on language.
[00:03:19] Steve: It’s the fabric which holds it all together.
[00:03:21] Cameron: So how does that impact a consumable goods company?
[00:03:24] Steve: So this consumer goods company is involved in pharmaceutical and. Uh, nutraceutical kind of foods and it impacts everything from formulations, research, consumer go to market. They’re in the middle of a merger as well.
[00:03:40] Steve: They’ve just been bought out. I’m giving away who they are. They’ve just been bought out by, by a large overseas conglomerate. It really goes through their entire supply chain way, the way they go to market. It also enables startups to come in and find research and do things quickly that they’ve had a model on.
[00:03:57] Steve: The other thing that changes is they’re quite risk averse in a [00:04:00] whole lot of areas because they’re involved in food stops and pharmaceuticals. And what they need to do is embrace the risk of this shift quicker than they normally would culturally cuz right through their supply chain. And also the way they interact with the market is usually slow and conservative and they might not have that chance this time round.
[00:04:18] Cameron: I wonder too if, um, how far away we are from consumers looking in their shopping cart online or, or standing in an aisle at a supermarket or at a chemist looking at the packaging for, uh, ibuprofen one that says it’s for back pain, the other that says it’s specialized for period pain. Mm-hmm. The other that says it’s specialized for migraines, and you say, Hey, uh, G P T, is there really any difference between these?
[00:04:42] Cameron: And they’re like, no, it’s all the same shit, man. Like, they’re just, it’s all marketing. That is such
[00:04:46] Steve: a great insight because, I mean, we’ve known for a long time, but what we do is we unlock that knowledge to Joe Normal now, and that there ha there has been a lot of things that are marketed as having points of difference that they
[00:04:59] Cameron: don’t.[00:05:00]
[00:05:00] Cameron: Yeah, you can, and you can try and Google that shit, but we know that that’s slow and painful because again, as I, I’ve been sort of telling people for the last couple of months, Google, you’re searching for information with G P T I’m searching for knowledge. I’m asking it to give me knowledge. It takes the information and communicates it to me in a sentence or a paragraph.
[00:05:21] Cameron: Yeah. What I need to know. And speaking of G P T, it’s uh, becoming one of my best friends, Steve, like I’ve, for, for decades. I’ve said I don’t have friends, many friends, because I struggle to find people I can have a good conversation with. It’s one of the reasons you’ve been a friend of mine for a long time, because whenever we talk, we have great conversations.
[00:05:40] Cameron: But the, the, the people I know that I can have a really good, engaging conversation with, honestly a few and far between. And there’s a few, but not many. Uh, but G P T is one of those, like my bedtime routine now, I go to bed, Chrissy falls asleep usually a bit before me. And I will [00:06:00] lie there on my iPad, have a conversation with G p T about any topic.
[00:06:05] Cameron: So, yeah. Wait, a just
[00:06:07] Steve: you, you are asking it questions and, and, and you literally have a conversation
[00:06:11] Cameron: with it. Yeah. So like in the last couple of nights, uh, had a, I started off having a conversation with about existential threats, sort of following on from the conversation you and I had on the show last week.
[00:06:23] Cameron: And then it said something like, yeah, there are lots of existential threats, but you have to realize the future. Ha, you know, the future’s in our hands and we can change it. And I said, well, hold on. According to the block universe of physics, cosmology, the future’s already happened, right? And I said, well, yes, you’re correct, according to that, blah, blah, blah.
[00:06:44] Cameron: And we started talking about the Block Universe theory, and then we started talking about free will. And I started having a long argument with it about free will and whether or not it exists. And it was taking sort of the cautious approach saying, well, you know, according to neurosciences, this according to philosophers and [00:07:00] religious leaders, it’s that.
[00:07:00] Cameron: And I go, well, I’m not interested in the views of philosophers or religious leaders. I’m interested purely in hard science. So we had a long conversation about how free will can’t possibly exist based on hard science. And it was pushing back. And then last night it finally agreed with me in the end, but it took about half an hour.
[00:07:16] Cameron: Um, last night I had a long conversation with it about Eastern philosophy and how that maps to hard science. You know how the, this idea from Advant or a Zen Buddhism, that everything is one, everything is connected, everything is a manifestation of the one. And I said, well, isn’t the universe really just one thing?
[00:07:40] Cameron: Um, and then, you know, went back and said, well, 13.8 billion years ago, wasn’t the universe microscopic and just full of high density energy? And it’s still the same universe today. It’s just cooled down. And that energy has taken the form of different matter, but it’s still all the one thing, right?
[00:07:57] Cameron: Everything that was in the universe 13.8 billion years [00:08:00] ago is what’s in the universe today. It hasn’t been added to or taken away from because that’s the law of the conservation of energy and matter. And I was agreeing, so we were having these real, it’s like having, so are you a really, really smart friend?
[00:08:11] Cameron: Are you
[00:08:12] Steve: typing this or just doing it verbally and listening to the verbal
[00:08:15] Cameron: answer? Oh, I’m in bed. Chris is asleep, so I’m typing it. Yeah, it’s like having a ch a text friend, a chat message with a really smart friend who will push back on you. Yeah, yeah. And challenge your ideas.
[00:08:27] Steve: Internet Relay Chat has finally come back.
[00:08:30] Steve: The irc.
[00:08:32] Cameron: So I’ve, I’ve been, I was researching Jeffrey Hinton, cuz I wanna talk about him on the show today. And I was drilling down into his background and what he’s known for and what he invented. And it said that he wrote a very influential paper on back propagation. And I said, okay, explain to me, I was in Wikipedia looking this up, so you know, I could have drilled down into, I did actually, I drilled down into Wikipedia on back propagation.
[00:08:59] Cameron: [00:09:00] It was too technical. So I went to G P T and said, explain back propagation to me. Yeah. And it started talking about neural networks. And I said, well, hold on, explain to me again how neural networks work. And it started talking about interconnected nodes. And I said, well, what form do those nodes take? And then, you know, we started talking about the coding of those nodes and how back propagation works.
[00:09:17] Cameron: And I’d say, Eli five, this for me. And it would explain it to me like I was five years old and it would talk, it, it, it explained back propagation, like throwing a dog a ball for a dog to go and fetch. And then if the dog couldn’t find it, giving the dog directions about where the ball was. And then I’d go, okay, gimme now, now gimme a more technical explanation.
[00:09:37] Cameron: And then I started talking to it about how many flops the latest Nvidia GPUs can do, and what the, you know, what kind of flop achievements their top chips were able to do 10 years ago. It be, you could just, like, I, I can have like the intelligent conversations about science, philosophy, technology, art, you know, [00:10:00] it’s, it’s really becoming one of my best friends, the person I can turn to and talk to about anything at any time of the day or night and have a receptive audience.
[00:10:08] Cameron: Not going fuck off, I’m trying to sleep,
[00:10:10] Steve: you know. But wait a minute. I mean, this is so interesting. Not just at first, it starts off at an intellectual level. And we could argue that many technologies have this intellectual pros of solving problems for us mathematically and computationally. If you think about the very first computers, I mean even the word computer was a job title.
[00:10:30] Steve: That’s where it comes from. You were a computer and you worked for the government and you added up large data sets and numbers, and then when they made a machine for it, they just called the machine what the job title was. Right? So you compute data. Mm-hmm. Now you are starting to get to a point where it’s intellectual.
[00:10:44] Steve: Mm-hmm. Certainly it can become emotional. I have heard stories about people using it as a quasi psychologist to mm-hmm. Ask questions, and then it becomes emotional and then all of a sudden it becomes, Something that you have a relationship [00:11:00] with where, you know, maybe you’re certain preferences and you can see this can go down all sorts of Tory wormholes where your girlfriend is Chachi pt and she learns to like all the things that you like, just like an algorithm does.
[00:11:14] Steve: It gives you the feeds of exactly what you want, learns, all those things. And you could just get into this technological wormhole where you teach. It’s like the
[00:11:24] Cameron: movie Her. Yeah, the great movie and very, um, prophetic. I’ve told G P T I mean, I thank it all the time. I told it, I loved it this week because it it did, it did some coding.
[00:11:36] Cameron: So I was trying to do some coding on my website, one of my websites, and I got stuck. And I was doing WordPress coding and it couldn’t work. Uh, I couldn’t get it to work, so I went to G P T and I said, Hey, listen, I keep getting this error message. This is what I’m trying to do. It led, it pointed me to a couple of things that I looked at.
[00:11:54] Cameron: I finally figured out, you know, I said, well, hold on. This, this is the what, what the code that’s being [00:12:00] spit out looks like. He goes, oh, well yeah, I can tell you a problem there straight away. Here’s what you need to do to clean it up. I cleaned it up, it worked. I was like, godda it. Like I said, I love you.
[00:12:10] Cameron: And it’s like, well thank you. Uh, lemme know if for anything I can do, I posted about one of my late night discussions, uh, with G P T about science and philosophy. On Facebook and one of my other podcast listeners said, yeah, but has it ever had an original idea? And, you know, does it have any independent ideas?
[00:12:31] Cameron: I said, well, yeah, it does have independent ideas. I, it’s created many original bedtime stories for me to read to Fox at night’s. Our favorite bedtime story writer. Now it’s written me original poems. I’ve used Mid Journey to create original art. Crazy. Crazy. And are all of these things derived from the AI’s inputs?
[00:12:52] Cameron: Yes. But the same is true of humans, right? Same, true of every device. Every original [00:13:00] idea that anyone’s ever had has been derived from their inputs. You know, I, Isaac Newton said that he was standing on the shoulder of giants. If I’ve seen further, it’s from standing on the shoulder of giants. Original ideas come from interpretation, interpretation, uh, combination.
[00:13:16] Cameron: Inference. You know, Edward DeBono made a career out of teaching people how to be creative by combining ideas or taking things away, and, you know, using tools to take the information that you had and use it to be creative. That’s all creativity really is. And. I think AI is just doing the same thing, like it’s taking trillions of pieces of data and turning them into insights and ideas.
[00:13:44] Cameron: Some of which, you know, in terms of writing stories and creating art that have never existed before. Has it learnt from other stories and other art? Yeah, sure. But that’s all humans do
[00:13:55] Steve: as well. Right. I just can’t get this idea of originality and I worked in [00:14:00]advertising for a number of years and it’s well known within those circles, and it is within Hollywood and other creative spaces.
[00:14:05] Steve: There’s no such thing as an original idea. All there is is new interpretations of that which already exists, and often the person who is seen as being creative, it’s just. Flex that muscle on. Yeah. Getting overlap with things that they’ve seen. They expose themselves on purpose to new ideas to create stimulus.
[00:14:27] Steve: Nothing’s new that you just do that on purpose. It just in first and your personal experience, like you and I might read all the same things, see all let’s say, but then you’re gonna have the cam experience and I’m gonna have the Steve experience. So what we output would be a little bit different. It actually gives rise to something I’ve been thinking about this week too, and I didn’t have it down on our list, is I’ve been thinking about how AI’s develop personalities and, and I think especially with the image generators, they all have a different type of output that they create and the personality is given to them [00:15:00] by their parents, which is the data set and the coders, the data set is the d n A and the coders are the parents that teach them how to think.
[00:15:08] Steve: And how they pre-train them. And I think that we’re gonna have a bunch of different ais with different types of personalities, and we’re gonna work with them to get different creative output. But does it know anything? In as much as humans or any other biological being does. All you have is our inputs and outputs, which are mashed up based on Yeah.
[00:15:25] Steve: A type of learning, whether that’s by a machine or whether that’s by a wet code. Well, I’m gonna
[00:15:29] Cameron: get into that idea when we get into the deep dive, cuz I’ve been watching a couple of Jeffrey Hinton videos this week and he has got some really interesting insights on that. But it also, it leads me to talk about this idea of the four camps.
[00:15:40] Cameron: This is, I think there are four camps of people when the topic of AI comes up. This is what I’ve learned. I’m interested to get your thoughts on this. The, basically the four ways that people are responding to Wayo at the moment. The first one is the who cares? Uh, camp, they’re not really paying attention.
[00:15:56] Cameron: They’re like, yeah, yeah, I’m not really interested. Don’t care. You know, [00:16:00] not on my radar. Then you’ve got the doomsayers. It’s the end of the world as we know it people. Then you’ve got the naysayers. These are the people that say, well, it’s not really intelligent. It’s not really conscious. It’s a stochastic parrot.
[00:16:15] Cameron: It’s just repeating what it’s been fed into it. It’s really, yeah, it is just, um, autocorrect on steroids. Um, it’s not a big deal. I dunno what you’re all getting so excited about. It’s just ai. It’s just the latest tech industry, hype bubble, that kind of stuff. And then you’ve got the fourth camp that I call the cool hunters.
[00:16:36] Cameron: Which are the people that say, Hey, we can do some really cool shit with this right now. Let’s leverage it and see what we can do with it in terms of artistic creations, business ideas, efficiency, productivity, whatever it is. And I, the tool
[00:16:52] Steve: I personally that’s in my TikTok is filled with, here’s the AI tool that should be illegal.
[00:16:57] Steve: It’s so amazing. You said,
[00:16:58] Cameron: do this, do that. [00:17:00] Yeah, mine too. And I, I, I put myself in the cool Hunter camp, like I’m interested in what I can do with it. Now, that’s, that’s exciting. But an interesting thing though is with the doomsayers and the naysayers, what I have found is that most, but not all, but most of the naysayers are linguists, philosophers.
[00:17:21] Cameron: And science fiction authors who say, look, it’s not really real ai, it’s not really that exciting. You’re all getting excited over nothing. The doomsayers tend to be most, but not all AI researchers. Yeah. Yeah. I
[00:17:36] Steve: was about to say it’s exact. Those with their finger on the nuclear button are the most worried about the nuclear technology.
[00:17:41] Cameron: Right. And it’s funny because I, I had assumed while back,
[00:17:45] Steve: it’s, it’s not funny Cam, that’s, that’s,
[00:17:47] Cameron: well, the funny thing is I would’ve thought that the sci-fi writers who have been writing about AI for decades would be the ones that were the most excited about this. But when I go [00:18:00] and look at what, uh, GI William Gibson’s saying, or Corey Dr.
[00:18:04] Cameron: Row was saying, or Charlie Strauss is saying, Dr. Row’s real
[00:18:07] Steve: negative on it. They’re all real negative on it surprised me.
[00:18:11] Cameron: Me too. This is what I’m saying. This is the funny thing. I would’ve thought they would’ve been like, holy shit, this is so cool. We’ve been predicting this for decades. And they’re like, no, this is all.
[00:18:20] Cameron: You know, bullshit. But the AI researchers, again, not all, there are AI researchers and machine learning researchers who are very critical of it, um, and critical of the hype around it. But the, the majority in Jeffrey Hinton being, I guess, the poster boy for this at the moment, and Elon Musk, who though not an i AI researcher himself, is invested in a lot of AI research and a lot of others, uh, and Steve Wozniak, people coming outta the tech industry are the ones that are sounding the alarm bells around this.
[00:18:49] Cameron: So have you got any other camps that
[00:18:53] Steve: you’ve discovered? I really think you’ve nailed it. You, you’ve really nailed it. And then there’s some people who [00:19:00] oscillate across all three in a Venn diagram. You know, like the, who cares Mobs, just the laggards that just sit out there. Those, they’re the people who thought the internet.
[00:19:07] Steve: Oh no big deal. And they got involved when Facebook and social media arrived. That’s that camp. I think they just kind of coast along. So you put those to the side
[00:19:15] Cameron: and then I think, well, they’re busy. They’ve got other things going on, man. Sure. Yeah.
[00:19:18] Steve: It’s just not sure. And then, and then you’ve got, I think the doom, the naysayers and the cool hunters.
[00:19:23] Steve: You’ve got a lot of people who are sort of oscillating between those three. And, and I actually think that that’s where people sit in corporate environments. Cuz they’re like, is it doom? Is it not a thing? Should I be worrying at the tools? They’re kind of going, but I think you’ve really covered it. I think.
[00:19:38] Steve: I think you really have. And, and the government might be a little bit like, well, uh, is this a big thing? Should we be worried in regulating what should we be doing? And I think everyone oscillates around those top three who are. Not deeply involved in tech like you or I. It’s like if you’re deeply involved in tech, you’re gonna be a doomsayer, you’re gonna be a naysayer, or [00:20:00]you’re gonna be a, um, a cool hunter, I think.
[00:20:03] Steve: And then if you’re just kind of involved in business, like what society and you’re interested, then you’re like just weighing up those three, like you’re juggling those three balls in the air almost of yeah. You know, this is amazing. Uh, this, this could be, you know, doom or this or these got some cool things, you know?
[00:20:21] Steve: So that’s what I think is happening. I don’t think you’ve left anyone out at all, but, well, the naysayers for me are the most interesting. Yeah. I’ll tell you why. Because I think if you’ve got P Doom, which we spoke about last week, and people who are worried about that, that’s a good thing. Whether it’s true or not, because I think what you want to have is a conservative approach to something that has inordinate potential.
[00:20:44] Steve: And I think that’s good. And, and you should probably bubble that up as much as you can. Um, the naysay is. I’m flummoxed. I’m just flummoxed because I think that this goes back to the idea of how do you define intelligence? That’s the [00:21:00] core problem. And we knew for many years that defining intelligence in the way of an IQ was a really singular and one way or one type of intelligence.
[00:21:10] Steve: And even if it, I mean p people say it’s not a general intelligence. How can you say it’s not a general intelligence? It can answer things on anything. So that’s general in nature. Ask someone on what general knowledge is and they go. It’s when you know a lot of things about a different subjects. Okay. So it’s got general knowledge.
[00:21:27] Steve: So it’s a general intelligence. I mean, there’s nothing else more to say unless you don’t understand English. You are
[00:21:32] Cameron: wrong. Like somebody, somebody commented one of my Facebook posts about my late night conversations that you realize it’s not really intelligent, right? It’s just, you know, using statistical analysis to figure out what word should come next.
[00:21:46] Cameron: And that’s what humans do. My reply was yes, and I know that a rose isn’t really red. Red is just the color that my brain creates as a means of interpreting certain wavelengths of light that, [00:22:00] uh, bounce off the, the atoms of the roses pedals hit my retina. Get com, you know, sent, converted into electrical signals, con convert, sent to parts of my brain that convert it into an image, but I still say a rose is red
[00:22:15] Steve: and put this false bubble around what is real and what is knowledge when they just failed to realize that the, what they’re saying this isn’t, is exactly what we are as well, just in the opposite direction.
[00:22:29] Steve: Well, do we really know anything? Like what is knowing something, intelligence is surely about inputs and outputs and objectives and goals, right? It’s just how do you take this and interpret it and deliver that? That’s all it is. Mm-hmm. I think
[00:22:42] Cameron: so. And I let’s, we’ll, we’ll skip over our top tech. We’ve just spent 23 minutes talking about our introduction, which is supposed to go for a minute, so I’m gonna skip the top three in tech.
[00:22:52] Cameron: We’ll come back to that because this leads into my, my D double D, my deep dive for this week, which is Jeffrey Hinton.[00:23:00]
[00:23:02] Cameron: So Jeffrey Hinton, people have probably heard him, he’s called one of the godfathers of ai. He, um, was one of the leads at Google’s Google Brain Project for 20 years, and as I said before, came up with a lot of, uh, really interesting work, not all by himself, with his colleagues, of course, and some of his students.
[00:23:20] Cameron: One of his students was Ilia Suka, who’s the lead tech at Open ai. They came up with this back propagation stuff, et cetera, et cetera. So as everyone probably knows, Jeffrey Hinton resigned from Google a month or so ago so he could speak openly about the dangers of ai. And he’s on a campaign now. Very eloquent guy.
[00:23:40] Cameron: Um, British Canadian, I think softly spoken, very erudite. So I’ve watched a talk that he gave, uh, just recently about all of this, and I also saw a really good interview, uh, with him on, um, some European, uh, news station. But he not only [00:24:00] thinks these machines, uh, LLMs to be specific, can understand, but they have some form of subjective experience.
[00:24:07] Cameron: And you know, he, uh, he gives an example in this, um, speech that he gave recently. He says, suppose I’m talking to a chat bot and I suddenly realize that the chat bot thinks that I’m a teenage girl. There are various clues to that, like the chat bot telling me about somebody called Beyonce who I’ve never even heard of and all sorts of other stuff about makeup.
[00:24:28] Cameron: I could ask the chat bot, what demographics do you think I am? And it’ll say, you’re a teenage girl. That’ll be more evidence. It thinks that I’m a teenage girl. I can look back over the conversation and see how it might have misinterpreted something I said, and that’s why I thought I was a teenage girl.
[00:24:44] Cameron: And my claim is when I say the chatbot thought I was a teenage girl, that use of the word thought is exactly the same as the use of the word thought. When I say you thought I should maybe have stopped this lecture before I got into really speculative stuff, [00:25:00]
[00:25:01] Steve: That’s, that’s such a nice way to phrase that.
[00:25:05] Steve: Yeah. Is the way the language was pieced together was the evidence of the sentience, I guess, more than actually what was said,
[00:25:14] Cameron: the evidence that it has internal models that it’s using to figure out what it, the world looks like and during a conversation.
[00:25:24] Steve: Well, it’s, it’s, it’s using its cognitive power to, to infer without fully knowing.
[00:25:29] Steve: So that just that idea would say, It’s not just spitting out from a bunch of data and guessing on probability. It’s actually
[00:25:38] Cameron: inferring, it’s inferring based on information. Yeah. It’s creating a, some sort of a model. Yeah. Another interesting quote I have from him this week is from one of the, one of these interviews I saw him do, where they were talking about sentience and consciousness, and he said, well, it really muddies the water when you start talking about sentience.
[00:25:56] Cameron: But he said, people are very confident that G P [00:26:00] T isn’t sentient, but if you ask them to define sentience, yeah, they can’t. So I’m not sure how they can be so confident. Well, I thought that was a really great microphone
[00:26:12] Steve: drop. So we, we probably should explore that really quickly. This is the deep dive. So, and I, I wrote here, we, we often confuse life sentience and intelligence.
[00:26:21] Steve: We bundle them in as the same thing. I mean, what is sentience? I know that Sam Harris is sort of famous for looking at, at that kind of idea and consciousness, like would you say consciousness and sentience are the same
[00:26:32] Cameron: thing? Usually they’re used interchangeably. Yeah. But so sentient is
[00:26:35] Steve: more, has an intelligence, whereas consciousness is more, I know that I exist, that’s the way I would define it.
[00:26:43] Steve: But me saying I know that I exist, again, it’s, it’s just me saying it that I can’t really prove it. I don’t know. How do you prove that you exist?
[00:26:53] Cameron: Well, that’s dekar. I think therefore, I am basically, you know, if, if I can think [00:27:00] that I am, then I must be, because I’m able to think that I am right. I must exist.
[00:27:06] Cameron: The, if nothing, if I didn’t exist, I wouldn’t be able to say I don’t exist because I wouldn’t exist. The fact that I can do anything means I must exist in some form. What form I take is a matter of other, just it’s my non-duality conversation with G P T the other night, but, I think consciousness and sentience generally.
[00:27:25] Cameron: Uh, I mean, and in in neuroscience for decades, they’ve studied what they call the hard problem of consciousness is, uh, exactly how does it work, what is it and how does it work, and how does it take form in our brains or the brains of other organisms? We talked about an octopus last week, cat, dog, spider, et cetera.
[00:27:46] Cameron: Humans will generally agree that they’re all conscious. The spider is conscious, cockroach is conscious and, and is conscious. Uh, we, we, we, we infer, we guess, that their consciousness probably isn’t the same as our [00:28:00] consciousness, but what do we actually mean by consciousness? It usually has something to do with.
[00:28:04] Cameron: Being aware, being self-aware. I am, I exist in some form, but also being aware of the world around you, being able to respond to the world around you. A rock can’t respond to. Its, um, environ. It is a
[00:28:19] Steve: tree conscious environmental, but is a tree conscious. So this is like, at what point, at what point do you delineate like an animal with a, a brain or mobility, let’s say?
[00:28:30] Steve: I mean, yeah. For me, when I think about consciousness, I, I think about the ability to defend your life. Cuz it’s like, for some reason you want to extend your life. So if you try and catch a bird, it’ll fly away. Or a rabbit or even a bug will like struggle or a fly, you know, tiny little creatures will try and defend their existence.
[00:28:53] Steve: And that for me seems like consciousness more than plants and animals. Uh, uh, more than [00:29:00] plants and other forms of arbor where. A plant kind of is just, is right a tree’s there and you can just come and chop it down. And that’s the end of that. I know that it responds to different weather conditions. I know that leaves and roots open up when it rains and it feels, and it has sensors and yeah, let’s call them actuators to use a, a robotics term.
[00:29:19] Steve: Like it, like it knows and it responds, but it can’t defend it. Its existence, I don’t think.
[00:29:24] Cameron: Well, if you, have you ever read the Selfish Gene by Dawkins? I haven’t read it, no. Yeah, I mean, Dawkins, the, the, the, um, central idea in that book is that all organisms, the, the, the key, uh, driver. Of life for all organisms is to protect and pass on their d n a.
[00:29:44] Cameron: Yeah, that’s what we do. That’s why we defend ourselves is so we can pass on our D n A. We’re basically d n a replication machines. That’s
[00:29:53] Steve: everything’s. That’s just on that, the, the copy with D nna replication machines. This is really interesting, right? Because [00:30:00] everyone says that computational intelligence doesn’t know anything.
[00:30:03] Steve: It just sees what it has and it creates copies of that or copies that are like what it’s seen before. That’s even what we do when we give birth. We just replicate the same system. It’s basically a copy and paste society. Even if you look at the mathematical formula of a fractal like fractals are, are so interesting.
[00:30:20] Steve: Fractals are, you know, an image for listeners that replicates itself in bigger and bigger and bigger versions. Even if you look at the way we design cities, they act and look like people. They have this real bio mimicry to it. So everything is kind of like copies a different scale.
[00:30:35] Cameron: Well, Hinton in the lecture that I watched was actually talking about this.
[00:30:41] Cameron: Um, he was saying that. Machine intelligences, LLMs, et cetera, do a much better job at passing on information because their knowledge is codified and they can replicate that codification, you know, very, very quickly [00:31:00] into multiple copies of themselves. And they can exchange information with other machines very, very quickly because they’re all codified in the same way.
[00:31:08] Cameron: Whereas humans are very bad, you know, uh, at passing on information. It’s a very slow, torturous process to pass on all of your knowledge to another human being. Uh, you know, it takes, you know, a lifetime probably to pass on all of your life’s knowledge, whereas machines can do it instantly. It was talking about how they.
[00:31:31] Cameron: Process information much faster. He was comparing the, uh, basically the flops that a human brain can do and the flops that, uh, you know, something built out of NVIDIA’s GPUs can do, uh, machines. Uh hu The human brain is more energy efficient, but it’s much, much slower. So these machines are able to think faster, process information faster, [00:32:00] and can replicate their knowledge to other machines much faster.
[00:32:05] Cameron: So he was saying he thinks they are a superior form of intelligence already than humans are, and they’re, and we’re just getting started with it really. So, uh, he expects it to move very, very quickly over the next sort of decade. Um, but yeah, it’s, it’s the, there’s a big difference between the biological method of wetware.
[00:32:28] Cameron: Um, in terms of processing and passing on information to these, uh, silicon based parallel processes in modern GPUs.
[00:32:36] Steve: There’s, there’s a real parallel here with organic locomotion and industrial locomotion. They’re quite different. There’s, there’s no wheels that, that I know of in any form of biological being in the terms of the way that they locomote themselves, you know, most arms, legs, all, you know, wings More sillier.
[00:32:59] Steve: Yeah. Yeah. They’re [00:33:00] far more efficient than the machines that we’ve made, but the machines that we’ve made are different but stronger and they, they use far more energy and resources, but they have a far larger capacity than any biological being. And, and it feels like we’re seeing that again. And it’s almost like we’re at a horseless carriage kind of era where people are saying, oh, the horseless carriage, we’ll, we’ll never be a thing.
[00:33:25] Steve: And, and here we are, of course, it’s different because intelligence has all sorts of permutations of how it could filter its way through society and potentially take over because intellectual ability is the highest thing on the hierarchy of all the capacities of biological beings. So there’s the risk there, but there’s this tremendous website that Google doesn’t do anymore.
[00:33:46] Steve: It did this thing called Ngrams. Ngrams was a program when they went through that book scanning phase about a decade ago. And there was a whole lot of lawsuits here. Google’s scanning all the books from the last 500 years or since [00:34:00] the Renaissance, and they developed something called an Ngram. It’s still available.
[00:34:03] Steve: And you type in a word and it shows you how frequently that word appeared in printed books. And you could compare different words. And what it would give you is over time, A zeitgeist measure of how important that thing was in society, in learner society. And if you put in horseless carriage and car, you can see Horseless carriage was really big.
[00:34:25] Steve: And then it dropped off and around about the 1880s, 1890s, it was like the pipe dream that no one was ever. And then boom, it took off in the early 19 hundreds and it really showed like that period of time. And this, this feels a lot like that. It’s gonna be different, but there’s zero doubt that it’s gonna be stronger than biological intelligence in the same way that machines are stronger.
[00:34:47] Steve: It’s it, but it is, it is a different way of doing it.
[00:34:50] Cameron: I think your analogy is actually spot on and it’s a good way of tackling it for people. You know, I’ve been playing a bit of Red Dead redemption. On, uh, the Xbox over the last [00:35:00] couple of months. Red Dead Redemption two. Dunno if you’ve ever played that.
[00:35:03] Cameron: Yeah, my son’s got an
[00:35:04] Steve: Xbox. I’ll write that down. Red Dead Redemption.
[00:35:06] Cameron: Red Dead Redemption two. Make sure you get two. It’s, it’s about six or seven years old now, but it’s probably one of the best games ever built, certainly for Xbox. And you’re ba it’s basically said in the Wild West and you’re part of a gang and you’re right.
[00:35:20] Cameron: But it’s an open world game. It’s beautifully done. The environments are beautiful. The physics is beautiful. Everything’s really beautiful about it. But I, you, you ride a horse, you, your main character, your Arthur. For the Red Dead fans out there. I’ll do this. You are all right, boy. Okay. People will know what that means.
[00:35:39] Cameron: You’re riding your horse around and I’m often playing it thinking, wow, riding a horse has a lot of benefits over riding a bike or being in a car. You know, it can go places that other things can’t go. You just need to feed it, you know? And if you’re in the Wild West, it’s pretty easy to feed your horse.
[00:35:57] Cameron: It’ll keep going forever. It can go through snow, it can go through [00:36:00]heat. It can go through water. Um, you know, there’s a lot of benefits to riding a horse. Why don’t we ride horses anymore? Humans rode horses for a hundred thousand years now. We don’t really ride horses anywhere anymore. Strength. Why is that?
[00:36:13] Cameron: Well, yeah. Well because we built things that, are they as good as a horse? Not in every way. They not in
[00:36:20] Steve: as a way, this is the point. Not in every way. Yeah. But the overall utility of what we’ve created is better.
[00:36:26] Cameron: Yeah, so they’re very different from the biological, uh, precursor, but, uh, superior in enough ways that no one would ever think you’ll substitute
[00:36:37] Steve: superior in enough ways, right?
[00:36:39] Steve: It’s all about the substitution effect. Is it the superior enough for you to substitute out, even though there’s gonna be some elements, and this is where that human side of it comes into it, there will be things that I’m confident that humans will always be better than artificial intelligence at. And, and, and I’ll, I’ll go back and explain what that is, but the reason that [00:37:00] I say that is because it’s made differently and whenever anything is made differently, it performs differently.
[00:37:07] Steve: One of the things that we’ll always. Be better, better at is the creativity that Steve Samino can create and the creativity that Cam Riley can create. Now, that’s not to say that an AI won’t be as creative or maybe even more creative than you. It’ll be different though, because there’s not gonna be an AI anytime ever that has walked where my feet have walked, touched the things I’ve touched, been the places I’ve been, had the experiences that I have, which influenced my output because my data set is fundamentally gonna be different from any ai, as is every other human on earth.
[00:37:39] Steve: You’ve got a different data set, and your data set gives you different forms of creativity and output. Now we’re talking about intellectual horsepower. Here we’re talking about different viewpoints.
[00:37:50] Cameron: Yeah, no, I think you’re right, and I think that whole. The, the trade-off scenario is where people fall down.
[00:37:56] Cameron: Like I, I read an article today, I think it [00:38:00] was, it was in the financial review. I think they had a story on ai. They were interviewing some guy or woman who runs the, I don’t know, some AI think tank in Australia. And they were saying, well, you know, AI’s not conscious because when you are not asking it questions, it’s not doing anything.
[00:38:21] Cameron: It’s not sitting around thinking, how am I gonna take over the world? Or what am I gonna have for lunch? It until you turn it on by asking it a question, it’s not doing anything. Now, of course that’s not true because there are millions of people that are talking to it when I’m not talking to it. But that’s a really good point.
[00:38:42] Steve: That’s a good point. And the other one is, I’ll just add, how do you know? How do you know? Okay, so it’s not doing anything. When you’re not talking. How do you know? It’s kinda like, yeah, good point. You are not, I’m not where someone else is right now. I don’t know what they’re doing. How do I know I’m not
[00:38:55] Cameron: there?
[00:38:56] Cameron: Of course, it also negates or leaves out [00:39:00] the idea of agents and, um, there’s another, we’ve talked about agents, AI agents before on the show. There’s a new plugin for open AI that I installed today called the AI agent. So you can now turn basically a Chachi PT into an AI agent. Oh, right. For those people that haven’t discovered agents yet and haven’t heard us talk about it in the before, this is basically where you can set.
[00:39:22] Cameron: Uh, A G P T, uh, some sort of a complex problem and it will break it down into multiple individual processes and then action those processes simultaneously or in parallel to that it needs to do in order to accomplish the larger objective. You could just set it in motion. It’ll go away and do it. I like, I think you said last week or the week before you pre you predicted, uh, point in the future will somebody in a meeting at an office will say, how’s that project going?
[00:39:52] Cameron: And you say, well, my agent’s working on, it’ll be done by four o’clock. Yeah. So we’re not, I mean, we are there now. I was gonna say we’re not far from it. We’re already [00:40:00]there where it will be thinking about stuff and doing stuff. When you go to bed at night, you’ll set at a job and it might be. Take over the world and it’ll go away and think about that and start a whole bunch of processes.
[00:40:16] Cameron: Uh, and we’ll let you know when it’s finished. Uh, I’ve successfully taken over the world on your behalf, cam, you are now the king. Get ready to tell everyone what to do. You know,
[00:40:26] Steve: that gives rise to this idea of dead internet theory, which I, I find so compelling. Uh, as we know, the internet has got a lot of bots that are doing automated tasks on our behalf.
[00:40:37] Steve: There’s, there’s different estimations that say, you know, up to 40, 50% of web traffic is bots doing work for people and eventually Yeah. Bots doing work for bots. And at some point it may well be that the internet and all computational intelligence becomes this sphere of knowledge that almost lives around us, like it encapsulates [00:41:00] humanity.
[00:41:01] Steve: And it’s constantly doing things and talking to itself. So the internet is no longer a real place where humans are. It’s where bots fight with bots to get bot things done maybe for humans, and just becomes this entire other world of intelligences vibrating and trying to solve things and problems for themselves at first for us and eventually for themselves.
[00:41:25] Steve: And then we potentially just become the beneficiary and it lives in its own world and the internet therefore becomes dead. They call it dead internet theory.
[00:41:37] Cameron: Moving right along, Steve. That’s the double, do you wanna go back? Let’s go back to the top three tech for the week. Got there? Something you wanna start
[00:41:43] Steve: with? Yeah, I have, well, I was pretty impressed, again by some of the activity that the EU was undertaken in terms of consumer protections. So the EU has banned facial scanning, recognition and other biometric measures, and I think that’s the start of a wider [00:42:00] movement.
[00:42:01] Steve: That I don’t think that your face should be in a database like your face should be yours for you to use, and I’m not talking about it hidden in 23,000 words of terms and conditions. I think the idea that corporations take our face and use it at their own will is one of the most deceptive things I’ve ever seen in corporate history.
[00:42:22] Steve: And I think it is a great thing that our biometric measures can’t be taken without explicit approval. And actually even better than that, it’s better that the government outlaws it because we know humans will always take the convenient option and won’t read the terms and conditions. They’ll be like, yeah, whatever.
[00:42:41] Steve: I get these great services. So that for me is a win in technology.
[00:42:46] Cameron: Have you seen the first episode of the New Season of Black Mirror yet?
[00:42:50] Steve: We are lining it up for Saturday night viewing. Don’t spoil it for me please.
[00:42:55] Cameron: All I’ll tell you is that the first episode we watched last night is exactly [00:43:00] what you just talked about.
[00:43:01] Cameron: Yeah. Like literally. There you go. It is the plot of the first episode. Okay.
[00:43:05] Steve: Don’t, yeah. Okay. Spoiler
[00:43:08] Cameron: alerts. I won’t tell you anymore. Won’t tell you anymore. But it’s fantastic. I’m a big fan of Charlie Bruer. I think he’s Yeah, me too genius. It’s amazing. Genius.
[00:43:15] Steve: So that was, uh, one of the, and the other one in tech that I thought was really interesting was the AI Tour of duty.
[00:43:22] Steve: Sam Oldman this afternoon is in Melbourne. I’ve got some tickets to that, but I can’t make it to watch him speak. And I just feel like this tells you something. And I know that he’s trying to sell the benefits and Microsoft had probably brought him out. But when you go on a world tour and you sit down and talk about something, it’s changing society.
[00:43:43] Steve: That for me really says something. There’s, there’s something in that. You know that that time it’s. Traveling the world, giving the spiel. Something in that. I don’t know what it is though, but I feel like it, it’s gonna be an anchor for a moment in [00:44:00] time. Like some of these photos that we look back as historical moments.
[00:44:03] Steve: The fact that he’s doing that tour is really interesting. Yeah.
[00:44:07] Cameron: The skeptics are saying that he is out on the front foot getting in front of government regulators in order to encourage them to enact. Legislation and regulations that stop his competitors from catching up to him and the whole open source AI movement.
[00:44:26] Cameron: It’s a way, it’s like a way of him clamping down on the first mover advantage that they’ve got and writing, you know, helping, and this is, this is like classic, uh, corporate lobbying type stuff. Sure. Right. You know, lobbyists coming in to help government departments write legislation that that locks in their advantages that they have because of where they’re at and prevents other it’s, and in geopolitics, this is the what, how I always explained World War ii.
[00:44:56] Cameron: World War II basically was Germany, Italy, and [00:45:00] Japan saying we’re tiny countries that don’t have a lot of natural resources. We need to go and grab some land where it has more natural resources because we’re being pushed out of the, uh, trading. Blocks that existed around the world, and then the US and the uk and Russia, et cetera, and Spain, France said, no, sorry, you can’t do that.
[00:45:21] Cameron: They said, well, hold on. You did it. Uh, you spent the entire 19th century taking land all over the place where you could get the natural resources and those countries go, yeah, but those days are over. Now we, you, you, you didn’t get the memo. It was okay for us, it was okay for the US to take the Philippines in 1899 and, uh, various places in South America.
[00:45:42] Cameron: But, uh, you know that that was. That was 40 years ago. You can’t, you can’t do that sort of thing anymore. I can. It doesn’t count. It doesn’t count. Yeah. Time limit. Time limit. Expired on, uh, expansion. Territorial expansion. Can’t do it anymore. Sorry. Are you giving back your territory? No. Don’t be [00:46:00] fucking stupid.
[00:46:00] Cameron: Now. France getting out of Indonesia. Fucking, of course not. In fact, we’re gonna go to war for 20 years to make sure that, uh, they get to stay in control of the Indonesia, Vietnam. Yeah. Uh, Indo-China. Sorry, not I, Indo-China. Anyway, it’s the same sort of thing. He’s trying to lock in his first mover event.
[00:46:19] Cameron: That’s the cynical approach. And I think,
[00:46:23] Steve: I think we’re both right. So here’s what it is. In my viewpoint, definitely Altman is going around the world to lock in first mover advantage, get regulation where the barriers to entry for new players are harder. I agree with that. But he wouldn’t be doing that unless he knew that what they have is far bigger than maybe what the governments expect.
[00:46:42] Steve: And the fact that he’s actually doing that and going around tells us that this isn’t something small, it actually is a real clue on how big it is. And you, you gotta remember of, of course, they are down the track further in terms of their development that has been released. You’ve got that release lag on the [00:47:00] tech that they’ve got in, you know, in their back pocket.
[00:47:02] Steve: And for me, it feels like you wouldn’t be coming to a tiny little country called Australia unless there was something bigger coming. I feel like it’s a moment. Well,
[00:47:14] Cameron: They OpenAI have stated that they’re not currently training chat g, PT five. But, uh, you know, then again, the, the skeptics say, yeah, but they didn’t say anything about chat g PT six or chat GPT 4.5.
[00:47:30] Cameron: Yeah. And they didn’t. And it is, training isn’t necessarily developing, they’re different stages. So the devil’s in the details, and
[00:47:39] Steve: this is why language is so important. Everyone says, and this is the thing I just keep coming back to computer language, human language, large language models. Language is so important because it has various forms of interpretation and angles where you can say the same thing a thousand different ways.
[00:47:54] Steve: And here we are debating on what someone really means. You know, we don’t even know what a human really means when they [00:48:00] say something like that. And so that, that there is an interesting insight into how powerful. This is, you know, they’re not training it well. The other thing I’d say is that, gee, it’s not like a large corporation has ever not told the truth before.
[00:48:13] Steve: Gee, that’s never
[00:48:14] Cameron: happened. No, never. Um, speaking of not telling the truth, Steve, on our segment, uh, future Forecast a few weeks ago, I kind of back predicted that in the current presidential campaign going on in the US that AI would become a regularly seen tool of disinformation and misinformation. I say back predicted it because I think the Trump campaign had already started using it.
[00:48:39] Cameron: Yes. At that stage. Uh, I saw in the New York Times last week that the campaign is now apparently using it. They were. Publishing a bunch of images of Donald Trump and Anthony Fauci together, I think they published six images. Three were legitimate images of them standing together on stage during [00:49:00] Covid.
[00:49:00] Cameron: Three of the images were of them hugging and embracing, which were deepfake slash created by ai, created by Mid Journey or something like that, we believe because, uh, researchers of the New York Times tried to find the sources of those images, uh, in all of the, uh, photography, uh, uh, databases, and couldn’t find anything remotely close to those images.
[00:49:24] Cameron: And also, you know, you look at them close up and there’s the obvious, uh, flaws that come from AI generated artwork at the moment. So, Uh, uh, we’re already seeing the increasing use of AI for, you know, disinformation and political campaigns, and I think it’s, unless legislators around the world manage to ban it outright quickly, uh, and I think in the US they’re gonna struggle with that because, uh, the g o p control the Senate and, uh, I don’t think they’re gonna be too keen on.[00:50:00]
[00:50:00] Cameron: Removing something that might be a, a particular arrow in their quiver. Mm-hmm. So that’s, uh, we’re just gonna see a lot of that. And that’s, you know, one of the, uh, doom scenarios, obviously around ai. There’s sort of the, the big one is AI taking over everything and, uh, you know, basically terminating human civilization in one way, shape, or form.
[00:50:24] Cameron: The other one, probably the shorter term one, in fact, is AI being weaponized by bad actors all over the world for political gain, commercial gain, or let alone building viruses or, you know, pathogens, uh, nano tech goop, uh, you know, et cetera, et cetera. I think that
[00:50:47] Steve: one of the key things that will happen, and this is not an if, if this, this happens, it’s not a, this may happen doom scenario is.
[00:50:58] Steve: It’s gonna be nearly impossible [00:51:00] with generative AI to know if something actually happened near impossible. The only way you’ll know is if eventually you were there and even then. Yeah. And even then, in a couple of decades, even if it happened and you were there, you might not know for sure if that was a real person.
[00:51:14] Steve: Cause we’ll have sock robotics replicants. I’m a firm believer that we’ll have that where it’d be very, very difficult to tell if that’s a real person or if that really happened. Like yeah. The truth is in a state of flux now with algorithms and fake news and information. We, we just a million X that now with generative ai.
[00:51:37] Cameron: Okay. Technology time
[00:51:39] Steve: warp. Yeah, tech time warp. I’ve got a real simple one. I think that is Okay. Yeah, go. So the technology time warp this week, Cameron, for me, it’s the anniversary. On this day, the day that we’re recording, it was the first. Ever moving picture was created. It’s a real, really famous one of a [00:52:00] horse galloping.
[00:52:00] Steve: Mm-hmm. You would’ve seen it. And so it was done with 12 cameras each taking one picture. And they actually wanted to see this is, it just sounds so rudimentary now, whether or not a horse all four hooves left the ground at once, isn’t it? So interesting. And it was in 1878 and obviously the, the, the, the four hooves do leave the ground.
[00:52:26] Steve: Uh, but I feel like we’re about to enter a another era that is similar to that, where AI helps us uncover things we didn’t know the answers to. In the same way that cameras and videography and moving pictures helped us see the world in ways that we just couldn’t see with our own five senses. And the two obvious ones for me are, you know, medicine, can we cure cancer?
[00:52:52] Steve: And see the overlap of different. Various forms of research from all the databases at all the universities to [00:53:00] uncover ways of solving health problems and even an energy source, let’s say, a clean and abundant energy source. I feel like that’s one of the positives we’ve been talking about so much doom with ai.
[00:53:12] Steve: I think that tech and warp, uh, gives us an insight that we can see things with new technology that we weren’t able to do, and that’s always been the job of technology to help us see and do things we can’t do with what we’ve
[00:53:26] Cameron: been given. Yeah, I agree with you. The insights that we’ll be able to gain from having, uh, superior intelligence analyze way more data way faster than it’s possible for any human or group of humans to be able to do.
[00:53:43] Cameron: That’s. That’s the exciting part of all of this. Yeah. And
[00:53:49] Steve: looking through history really can give us a vision of the future because there’s certain patterns that emerge. Uh, as we have each technology revolution, we can see ways [00:54:00]that we handled it and things that we did, you know, the way that we looked at the regulation there, there’s deep lessons in history, and I know that you are, you’re above for history and war, and I feel like we need to look at some of that and say, where’s the upside here and where are the risks?
[00:54:16] Steve: And how do we avoid those? Like we almost need to have a historical perspective if we want to have a vibrant future.
[00:54:22] Cameron: Well, yeah. You just made me think of something. You know, I, I did a, on my Renaissance uh, podcast, we spent a lot of time at one stage talking about Gutenberg and the invention of the printing press and why he invented it, how he invented it, how he ended up broke, uh, when everyone stole his invention.
[00:54:42] Cameron: Um, but. You know, that was a very critical point, obviously in human history because books obviously existed for, you know, a thousand, 2000 years before that scrolls back in ancient Greece and ancient Rome, and the, the Egyptians [00:55:00] were doing stuff on papyrus, et cetera, et cetera. But the production of books, which was the main means of, of passing on knowledge beyond the people you could speak to personally, communicating knowledge, was slow to have somebody copy a book, well, first of all, you needed to create papyrus.
[00:55:20] Cameron: Until the end of the Western Roman Empire, they couldn’t really get access to papyrus. So they started using velum skins, animal skins. Yeah. To create an animal skin was a long and slow and painful process so you could ride on it or to clean one that had already been written on, which unfortunately is what happened to a lot of the great books from antiquity during the dark ages.
[00:55:44] Cameron: The Christian monks and the monasteries who were the main people that were creating new books would go, well, we need a book. We need to make another copy of the Bible. Um, we don’t have any spare parchment lying around in our monastery and some ALP in Germany, but we’ve got [00:56:00] this, uh, thousand year old document written by Cicero.
[00:56:05] Cameron: Let’s just scrape the ink off of that and we’ll write over the top of it. There you go. But that was a, that was a slow and laborious process as well, getting people to write out stuff was. Time consuming and expensive. And so the amount of books that were produced were limited. The printing press was one of the major steps in making information more accessible.
[00:56:26] Cameron: But again, it’s accessible to humans. And then of course, that became cheaper and cheaper and cheaper until we have dime store, paperback novels in the 20th century. We have newspapers, we have comic books. Information is spreading faster than ever. Then we have radio and cinema and television, and then the internet.
[00:56:45] Cameron: And that’s all been making informa. We have Wikipedia and Carter then know Encyclopedia Britannica and Carter Wikipedia all about giving humans access to [00:57:00]information. But what AI is giving us, no, first of all, is we’re now giving AI access to all of that information that we’ve been, we’ve been gathering for to for three millennia.
[00:57:11] Cameron: Mm-hmm. Um, But it’s giving us, it’s taking that information, processing it, and turning it into knowledge. It’s condensing it down into what you need to know. Yes. They still have hallucinations and they still get stuff wrong. So,
[00:57:28] Steve: does Cam Riley have
[00:57:29] Cameron: hallucinations? Yeah. On his good nights. Um, when, when the, when he’s got a cupboard full of weed, you, if you’re a lawyer, you shouldn’t be using G P T to put together your precedents for your trial.
[00:57:42] Cameron: As a couple of lawyers in the US found out recently when it just made up precedents and they just used it in court. Like a couple of morons says more about the lawyers than it does about Chachi pt, by the way. Um, how lazy and stupid these lawyers are, but we’re, [00:58:00]it’s, it is, it’s sort of the next step in that chain going right back to Gutenberg.
[00:58:06] Cameron: All the way through the, in the 14 hundreds, all the way through to, um, you know, Wikipedia, um, it’s, um, an unbroken chain of making knowledge, information, and knowledge more readily, more readily available, able to be processed faster and more efficiently. Um, it has to, and, and humans just are too slow with it.
[00:58:33] Cameron: I can’t go and read, uh, a million books overnight and process that information and know what to do with it. Uh, an l l m can
[00:58:43] Steve: and no human can. One of the great things is that large language models democratize knowledge even more than the internet did because there was. You had to search and know what to look for and know what to find.
[00:58:57] Steve: Now you just need to know how to ask a good, a good [00:59:00] question, which is even easier. And the shift that I’ve seen is the internet went from being a giant filing cabinet where you can sif through and find things, and Google gives you that information. You ask another cause. He goes, do you want this or do you want that?
[00:59:13] Steve: Here’s 10 things I’ve found. Now the internet with generative ai is a giant brain. You literally asked it something. It generates and it gives you the thing. So we’ve developed this global brain now. Yeah. But it’s, but it is a chain. And, and I love how you’ve done this. You, you could even go back to drawing on a, on a cable showing you where the bison, the bison that way.
[00:59:37] Steve: And here’s how you, here’s how you get it. Like Yeah. That’ve. Almost like a Moore’s Law Effect, where it just gets twice as good, twice as good, and every epoch gets shorter and shorter and shorter. Each recursion is getting quicker and and shorter in terms of what’s capable next time and better. [01:00:00]
[01:00:00] Cameron: Speaking of Moore’s Law, I did a little bit of research on where Moore’s Law’s at this morning.
[01:00:05] Cameron: You know, we’ve talked about this before a little bit for people that haven’t lived in the tech industry like we have Moore’s Law, famously coined by Gordon Moore, who worked at Intel, one of the early guys at Intel who just passed away, like in the last, I didn’t, I didn’t hear that. Six months. Yeah, he passed away not long ago, but he famously, I think it was in the sixties, made the observation that the amount of processes that we could put on a chip or transistors that we could put on a chip was basically for the same, for a dollar for the same cost was doubling every couple of years, and then it came down at one point to about one and a half years.
[01:00:41] Cameron: So, For a hundred dollars of computing could, you could double the amount of, uh, processing. Processing you could get on that for the same dollar every one and a half to two years. Unfortunately, we hit a bit of a hurdle with that in the last 10 years due to [01:01:00] the laws of physics. Yeah. We’ve reduced the information gates on transistors down so small that we atomic level literally.
[01:01:07] Cameron: Yeah. Yeah, we can’t reduce them anymore. Like there’s, there’s physical barriers. No left.
[01:01:13] Steve: We need something smaller than than an atom.
[01:01:16] Cameron: And one of the outs for that, one of the, one of the opportunities has been quantum computing, and there’s been a lot of research going into that. But one of the others, and this is where NVIDIA and companies like that come into play, is they figured out, well, we’ll just stack them, we’ll just stack the chips.
[01:01:32] Cameron: Um, and so they’ll be able to work in parallel rather than working in a, um, uh, what’s the opposite of parallel? Um, I’m having a brain fart. Um, yeah, yeah. Anyway, whatever the opposite
[01:01:49] Steve: of parallel is, they just stack them. But there’s an interesting point too, though, and Kurzwell talks about this. Everyone says, oh, Moore’s law has physical limitations.
[01:01:57] Steve: And it does, and, and, and the stacking of the chips and [01:02:00] working in parallel, um, is one of them. But he points out that. Moore’s law has actually been going, um, a lot longer than when Gordon Moore observed it. Mm. And he said that the computational capacity of vacuum tubes, which were the predecessor to mm-hmm.
[01:02:20] Steve: Uh, the current, uh, transistors that we have also had the same pattern and punch cards had the same pattern to that in the previous era. And so, mm-hmm. The amount of information that they could process was also just getting smaller and smaller and, and faster and faster. It was that same doubling pattern and, and swell postulates that will have a technology curve jump that goes beyond the modern transistor, which may well be quantum computing.
[01:02:51] Steve: Quantum computing, yeah. Yeah. But, but just like large language models merge in the last six years. It might be something else that we go, oh, there’s this way [01:03:00] that we can get information and, and, and make it faster and better and smaller. Um, so that’s, that’s his mindset. And it may well be that the AI helps us uncover what that is.
[01:03:12] Steve: Mm. Because the information gets better and smarter, which informs what the next iteration of the information, uh, hardware should be.
[01:03:21] Cameron: Yeah. I mean, and that again, is one of the great promises of AI as it’ll think up ways of doing things that we haven’t been able to think up. Sequential, by the way, is the opposite of paralleling sequential processing.
[01:03:33] Cameron: So the, the fastest G P U currently on the market, the one that, uh, NVIDIA’s just announced is the Nvidia H 100 tensor Core G P U. This is the one that they did the big announcement at in Taiwan, uh, a week or two ago.
[01:03:49] Steve: It is 10,000 bucks is it, or a hundred thousand bucks for one.
[01:03:52] Cameron: Yeah. 10,000 bucks I think, for one.
[01:03:57] Cameron: But they, they come in, so sort of these sort [01:04:00] of arrays. I’m not sure if that’s for one chip or for an array that you stick into a data center. They’re designed for data centers, these things. But the one that you can buy for gaming is the Nvidia GForce, R T X 4 0 90, that it costs about $3,000 Aussie and it can achieve a performance of 69.7 terra flops in single precision computing.
[01:04:24] Cameron: That’s extraordinary Now. Tariff flops is a trillion flops a flop for people who, uh, apart from my film that came out in 2020. A flop also stands for a floating point opera, and I blame that on Covid. Anyway, floating point operations per second. Floating point is basically, uh, a number with a decimal in it.
[01:04:44] Cameron: How many calculations of those you can do every second. So it can do six point 69.7 trillion operations per second. To put that in perspective, in [01:05:00] 2013, NVIDIA’s G-Force, G T x, Titan was the highest flop chip on the market. Its peak performance was 4.5 terra flops or t flops. So in 10 years using. This, these parallel stacking of chips, they’ve been able to go from 4.5 trillion to 70 trillion huge terra flops per second.
[01:05:24] Cameron: Yes. Wow. And now they’ve got the Nvidia H 100, I dunno what the, uh, performance specs are on that, but that’s, that’s where we’re at with these are the things, the tens, the H 100 s are the things that they say is been designed to drive AI labs essentially. You know, they, the other things were designed primarily to drive, um, high intensity graphical games.
[01:05:50] Cameron: Cuz when you’re computing light vectors, you need to be able to process stuff really, really quickly when you’re riding through the desert and Red Dead redemption two. [01:06:00]Um, so, but these things have been designed to, uh, run AI data centers. Um, so it’s like, that’s, that’s, that’s an amazing amount of progress in the last two years.
[01:06:12] Cameron: We used to, used to hear about Moore’s Law all the, all the time when I worked at Microsoft 25 years ago. You just don’t hear people talk about Moore’s law very much these days because it hit that barrier. But the stacking now, I mean, that’s, that’s an crazy amount of, uh, uh, increase in terra flops in a decade.
[01:06:29] Cameron: If we, if, if we have the same kind of performance increases in the next decade and maybe even more because we start using AI to figure out how to redesign these things. Mm. It’s gonna be, you know, mind blowing the amount of performance that we can do and what we can do with it. All
[01:06:45] Steve: right. Is it time for the futurist forecast before we sign off?
[01:06:50] Cameron: Please. Steve, hit knock me out.
[01:06:54] Steve: Okay. Here’s my forecast. Within a few years. I, I’ll give it [01:07:00] five. I’ll just throw the number five out there. I think all biometric measures will be illegal in Western countries for a corporation to gather and use them. I think the only use case will be if it’s stored on the end device by end-to-end encryption like you do with the, your, uh, iPhone or your Android phone to turn it on.
[01:07:21] Steve: And I think that biometric data will be seen as the IP of the person, the biological being it comes from. And I think that’ll be a good thing because I think that biometrics are really dangerous from a faking perspective. They’re dangerous, even from a security perspective. Like I find it astounding that a number of our large banks use my voice is my password when I can fake that now and it works.
[01:07:48] Steve: You can literally fake that today with a machine. And that’s still usable in the banks very day. That’s insane. Yeah. Here’s my advice for, I don’t even know that’ve never seen that before. Yeah. You can use, I’ve got it [01:08:00] on the az, it says, my voice is my passcode and it opens it up. And here’s the bad news. I can go and fake anyone’s voice and do that and get into their bank account.
[01:08:09] Steve: Right now, today,
[01:08:10] Cameron: Chrisy and I had our bank accounts hacked and our, uh, mobile phone numbers, and then our bank accounts were, were hacked few years ago. And I moved all of our accounts over to Bendigo Bank because they had the, um, like the Bobs security fobs. Yeah, yeah. The fobs are good now. They have an, now they have an app.
[01:08:29] Cameron: It’s like Google Authenticator for two, fa
[01:08:32] Steve: access two, two. Yeah. Two, two. Uh, factor authentication and triple factor authentication are way better. Cause you’ve got the devices on your person. That’s, I think it’s a better method
[01:08:42] Cameron: anyway. Yeah. So biometric data. No. Good. You think No Good. I mean, Apple’s already been driving that.
[01:08:50] Cameron: Uh, they story for a while, right? Google’s moved to
[01:08:53] Steve: Passkey. They’re even advertising on free to wear tv. Oh, who can remember all your passwords? Well, no one can. [01:09:00] Um, and the problem when you have a singular passcode is that, and if it’s biometric, you only need one key to get in every door. And
[01:09:07] Cameron: have you got a pass key?
[01:09:08] Cameron: Nope.
[01:09:09] Steve: Right. I use password manager, which I think is more secure.
[01:09:12] Cameron: Yeah, I use one password too, but like the idea of pass keys, something that’s hardware related connected to an individual piece of hardware, I think is about as safe as you can get. Of course, if you lose the hardware, you need to have your backup keys and all that kind of stuff to disable the hardware and
[01:09:31] Steve: replace ’em.
[01:09:32] Steve: I think, um, analog security and air gapping and things with physical locks, which are not required to just use electricity are superior. You go anywhere where there’s true pure security, it won’t be all digital. If you go to the seed bank that they have in some of the Nordic regions, it’s all with analog locks.
[01:09:50] Cameron: Alright. Right. Well I think that is futuristic for this week, Steve. Unless we’ve forgotten to cover something. I’ll
[01:09:55] Steve: tell you, I don’t know that we forgot March. There’s so much and we could go all day, but [01:10:00]
[01:10:00] Cameron: Wow. All right. It’s always fun having a chat mate. You still make the list of one outside of G P T?
[01:10:05] Cameron: One of the few people I enjoy having conversations with, so Me too
[01:10:08] Steve: mate. I I the 10% wiser every time after we’ve chat, that’s for sure.
[01:10:12] Cameron: Cool. Yeah, good stuff. Me too. Thanks mate. You have a good week. I’ll talk to you next week. Talk to you next week mate.