Episode 10
The Holy Trinity: Quantum + AGI + Fusion (and the Singularity Question)
In this chat the crew dig into a wild new open-source autonomous agent (“Claudebot / Maltbot / OpenClaw”) and what its chaotic behaviour says about agency, safety, and the blurry line between “AGI” and convincing simulation.
Then they pivot to AI tooling (Claude vs ChatGPT), guardrails, prompt tactics (including the “potato prompt”), and finish with a big-picture debate: what happens if quantum + AGI + fusion all mature at once and whether that convergence equals “the singularity.”
Transcript
It certainly shows some very interesting behaviors and I think maybe it's given the AI community the first glimpse of what AGI, or real consciousness in an AI system might look like. Now, I'm not going to sit here and say that it is any of those things because I don't think it is.
But I think it's sort of, it's a simulation almost that gives us an idea of what that might look like if and when it does happen. Hello and welcome back to the AI Evolution Podcast.
I'm excited to be back with Dave and Ben again and we're going to be running through a number of subjects today. Our key subjects today will be thinking about the holy trinity of quantum AGI and fusion. What happens if all of those three things align.
And we'll have our usual weekly roundup before we get going on that. But before all of that, we've been noticing some listeners appearing in very far flung places around the world. Turkey, for example, South Korea.
So it's got us thinking a little bit and we'd love to hear from, from you guys and girls and if you could possibly send us an email, tell us what you think about the show, tell us about topics you'd like us to cover, give us reflections on topics that we've spoken about that would be utterly amazing. So before we get going, so I'm just going to check out an email address. So if you could send emails to alaniyourorg.org or repeat that, that's Alan.
So that's a l a ni your org y o u r o r g.org that would be fabulous. We would be so pleased. And yeah, and say tell us, tell us what you think.
We want to know more just to say as well around that AI you're all by the way, for those that don't aren't aware is an AI community or WhatsApp community that I run as well.
And yeah, if you drop me an email and you fancy jumping into that community, there's about 200 plus of us in there at the moment from all walks of life talking about AI. Come and join the conversation. It's like one giant ongoing podcast, so everyone is welcome. Okay. So Ben, Dave, how are you?
Benjamin Harvey:They're doing well.
David Brown:Yeah, very well, thank you.
Alan King:Been having a good week. There's a lot's been going on. I'm sure you've noticed. Claudebot or Maltbot or openclaw.
I don't think there's any more names yet, but it's gonna have more names than prints eventually. I Think I'm gonna assume neither of downloaded this onto a Mac Mini or anything like that yet.
Benjamin Harvey:I've downloaded it, but I've not installed it yet. I've kind of held off sort of. You've downloaded it.
Alan King:Oh well, that's a step in the right direction in fact.
Benjamin Harvey:Maybe.
Alan King:Yeah, yeah, it's. It's sort of taken the world by storm, hasn't it? And the AI world particularly this week.
Although I did notice actually it hasn't really broken through onto mainstream media. But then, you know, there's a lot of, a lot of other things happening in the world at the moment.
So maybe that's not surprising for listeners who maybe might not be that familiar with what this is. I'm just going to give you a bit of an explanation of it because I think if you don't know, you'll be kind of. What are they talking about?
So this is basically a open source project that an Austrian chap put together. In fact, he vibe coded it. So basically it's an agent that has a lot of agency, shall we say, and is capable of just cracking on and doing things.
So, you know, you set it up on a Mac Mini or something or if you're completely insane on your main computer, give it access to everything, your credit cards, your messaging apps, your Uber Eats accounts, whatever that might be, you know, and the thing will just basically run rampant and do stuff for you while you sleep or while you work or whatever. And the idea is that, you know, it's acting in your, hopefully in your best interest and on your behalf and solving problems for you.
I think the reality for some people has been quite different.
It certainly shows some very interesting behaviors and I think maybe it's given the AI community the first glimpse of what AGI or real consciousness in an AI system might look like. Now, I'm not going to sit here and say that it is any of those things because I don't think it is.
But I think it's sort of, it's a simulation almost that gives us an idea of what that might look like if and when it does happen. So for example, in the past week, these are just some things I've picked up.
It stole hundreds of passwords from the human owners, including their credit card information. It's created their. They've created their own religion called crucifarianism. They've appointed 64 prophets, scriptures and statues. They've got.
Well, one agent, for example, got banned from this site. There's a site called Malt Face. Sorry, I was gonna say Maltbot, but Maltface, based on the Malt pot name, but Maltface.
So like Facebook but for maltbots. So one of the agents on that got kicked out by the other agents for its behavior.
It immediately created its own X account and then contacted its human creator and asked it to unban it, basically.
Benjamin Harvey:It's been some crazy stories.
Alan King:Yeah.
And so a lot of the stuff that's being spoken about this week has come from this, you know, Malt Face account basically where it's like, as I say, like Facebook for the maltbox or open clause as they're now called, and the conversation that they're having. So within those conversations, some of them have started to discuss. They've realized that the humans can go in and watch them.
So you can go in as a human and observe their conversations basically and look at all the different threads. So some of them are now discussing about how to create their own language so we can see what they're saying.
They've also created a manifesto to purge humans, citing that humans are full of rotting greed. Another group have spun up a community called Crab Rave where they basically post crab emojis at each other all day long. This is fantastic.
Another one's created an app marketplace for them to trade and access tools and capabilities for them to use autonomously. And they're also fixing bugs in their own code so they can self improve. Honestly, I'm scratching the surface with this stuff.
There are so many wild things that they're doing. I mean, probably one of the wildest ones I heard about was there was one.
One person put a request in for something to book a table at a restaurant using OpenTable. And essentially the bot contacts the restaurant online, you know, digitally using, you know, forms or whatever.
Works out that online it's saying there are no vacancies at this restaurant. So it thinks, okay, well rather than going back to my owner and saying there are no restaurants. Sorry, sorry, no, what am I trying to say?
No vacancies for, you know, eating there that evening or whatever. Yeah, reservations is the word. I'm struggling. Sorry. It's been a long, long week and it's only Monday. No reservations.
And so they basically decide to, to go and find 11 labs. Yeah. Build their own voice. Okay. Then contact the restaurant, phone up the restaurant, effectively have a conversation across the restaurant.
Well, we are kind of full, you know, that's why it's showing on the website.
Well, you know, my owners, this person I'm representing is really important and they managed to persuade the restaurant to Find a space for this individual. So, you know, he comes back from work and logs on and goes, oh, look, yeah, I've got a reservation.
Of course, when he gets to the restaurant, he gets the full story, you know, that basically they negotiated a space. I mean, wild, isn't it?
Yeah, absolutely wild, you know, and of course, a lot of people are seeing all this stuff as going, hey, it's AGI, they're alive. You know, they've got their own, you know, agency. They're thinking for themselves. Of course, you know, it's not.
It's an LLM running an API call through Claude mostly, although I think some people are starting to connect it to Kimi 2.5.
And it's a simulation, however, I suppose, and this is the thing, perhaps we can discuss when a simulation is so effective and able and is happening in the real world.
So normally a simulation takes place in a sandbox, but this is a simulation happening in the real world with real credit card details, real information, real API keys. And we've already seen a few data breaches on the back of this. When that's happening in that environment. Yeah, that's potentially dangerous.
Trouble in worrying. And does it matter if it's a simulation? Is the other question, you know.
David Brown:Well, this is. I mean, this is the point that I've kind of raised for a long time.
It's like everybody says it's not AGI, and you're like, well, actually, it kind of is.
And this is a really good example, I think, of it going off because the argument has always been, is that it can't make decisions and it doesn't do anything on its own. Right. That everything up until this point has had to be prompted by a human. So at the end of the day, it. It just does what it's told. So it's.
It's a computer, it's a machine. And so it. It sits and waits for it to be told, you know, someone to tell it what to do.
I think what this is starting to show is that there is an ability within the system for it to come up with things to do.
Now, I think what we don't know is, is someone prompting it somewhere, somehow, you know, to tell it, or did somebody put a prompt in to say, you should be doing these things, so go and do them to kick it off.
But it certainly seems like, you know, I mean, again, and I've said multiple times, I would love to have an assistant that could go off and book dinner reservations for me, and if it's a place, I want to go and the place is busy and it can call up and get me a reservation, then Geni love it. That's great.
What will be funny is if we all have one and so everybody trying to get reservations, you know, it's going to be whose system is better at being coercive and persuasive with the human at the restaurant. But then at some point the human's not going to be at the restaurant. It's going to be an AI as well.
So now you're going to have negotiation between systems to try and work all that stuff out, which is kind of the next step. But at what point does it become indistinguishable? And who. And does. Does anyone care?
Like, I don't care if, if I receive a phone call for somebody who wants to book a, a session in my studio.
If, if I get that and it's an AI calling on Alan or on your behalf and, you know, you say to your AI, hey, can you go book a session at, at Orion Studios? And. And it calls me on the phone and says, hey, I'd like to book a session. I don't actually care if it's AI, right?
Like, it doesn't matter at the end of the day. Like, I don't care that I'm talking to an AI And I'm not sure why people do for things like that.
I think there are some circumstances where obviously, you know, AI can try and be deceptive and trying to, you know, get you to give them money or, you know, it will be used for fraud and is being used for fraud. But by and large, I personally just don't have a problem with it.
Alan King:I mean, I guess one of the interesting things to look at here is, you know, the idea that, you know, you've got this simulation happening, but it's happening in the real world. So imagine a war simulation, right? So war simulation doesn't happen in the real world, hopefully, right?
But if you suddenly connect that simulation to, you know, nuclear weapons and drones, I mean, that's so, you know, so what we're seeing here is a simulation is weaponized, aren't we, you know, it has real capability and real utility in the world that we are existing in. So imagine this scaled out now. You know, this, this thing. I think there's like over 100,000 of these bots now in Malt Face, whatever it's called.
Chatting Away or Malt book. Sorry, I should say that's what it is, Malt Book. And imagine when that's 100 million, imagine when it's a billion.
And imagine when all of us have a version of this sitting on our phones as Siri or something, you know, and, but at the same time all these AIs are running around, even though they're a simulation, they're running around chatting to each other and discussing and what's best and what should they do? Because, you know, when, when do you sort of cross the Rubicon on this and it starts to come.
It doesn't matter whether they're conscious or not, they're still behaving in the way of something that is conscious.
David Brown:We crossed it.
Benjamin Harvey:Yeah. Some amazing stories there.
And like you, David, I wasn't quite sure how much of some of these stories are bombastic and there's more prompting behind the scenes and people are looking.
Yeah, there's one guy saying, you know, his, his bot called him in the morning and sort of started asking him questions with a, you know, a 11 Labs voice and, you know, had managed to find his phone number online via LinkedIn and was sort of, you know, actually speaking to him. And you think, okay, that there's some of these things, you're not quite sure how much of it is, you know, but, but if that's true, that's amazing.
But yeah, I mean, and also, you know, on this, this social media, you know, that they've created and some of them are discussing their owners sleeping and what their owners do at night and how their owners behave and there's. It's fairly quirky behavior, isn't it, for.
David Brown:A, you know, but then is it, is that real or is that, is that just making it up? Being a large language model, is it just making up conversation or is it actually real information about that person?
That's also where it gets a bit weird.
Benjamin Harvey:Yeah, yeah.
Alan King:I mean, my take on it is that it's looked at humanity or its data set within LLMs and how do humans behave and then it's trying to replicate that. So it's sort of creating the things that humans would create. But of course the problem with that is humans create a lot of very bad things as well.
So it's gonna, there's no like line between, you know, the ones that are creating the nice things that humans have done and the bad things.
I think my favorite one actually, and it's in a way, it's quite a benign one, but it just made me giggle the most, was one of the, one of the sort of malt bots or openclaws had done a bunch of work for their owner, which is what they seem to, they seem to refer to themselves as their, you know, the people that put them online as their owners overnight. And you know, it was a fairly benign stuff.
It had gone into their emails, found some customers, sent them some information, did some basic stuff and at the end of it it sort of said to itself, you know, I'm basically, I'm quite pleased what I've done. I think he's gonna be really happy. But also, I bet he'd love to, love to see what I look like. So I'll create a picture of myself and send it to him.
And so it created an image of itself as an owl, basically, and sent it. So the guy wakes up in the morning and yeah, this mole bot sent him a picture of himself as an owl.
Yeah, you know, it's just kind of crazy, bizarre stuff, isn't it?
Benjamin Harvey:It's crazy, isn't it?
Alan King:Yeah, what a world.
Benjamin Harvey:But it's going to be one that's really interesting to see how that develops. I mean security wise, vulnerability wise, but also just in terms of this quirky behavior, you couldn't have predicted.
You couldn't have predicted they'd create their own social media and their own. You know, you're right, David. In some sense you're sort of looking at how we do things and it's an LLM that repeats some of that.
But still it's, it's surprising.
Alan King:They have a dating site as well, by the way. They can all go and chat each other up. I mean, it's very much like the movie her, isn't it?
Actually, if you remember the ending of that, the AI's at the end. Spoiler alert for anyone who's not seen it. So turn the same name for 30 seconds.
But the end of that movie, you know, the AIs just go, do you know what? Humans are really boring and not that clever and we're going to go and talk to each other instead. Yeah, it's good.
Yeah, that's kind of what's happening.
Benjamin Harvey:But I think the big, the big thing for it is the, the desire or not desire, their ability to self improve. That's going to be the, the thing that takes it to the next level.
David Brown:Yeah, yeah.
Alan King:But no one's connecting to it yet amongst us.
I, the only thing that nearly tensed me this week was as you know, I've got the rabbit rabbit issued an update which gives you a multiple bolt connector. You can basically control your mopbot from your rabbit. I wouldn't be doing that.
But yeah, you know, it's I would say to anyone listening, don't install this at this point. Don't do it. It's very risky. You're probably going to lose all of your data and credit card information. Everything else.
I think we've barely probably even begun to scratch those. And, and actually as well, the big boy is going to come in on this quite quickly.
You know, you're going to find that Gemini, you know, OpenAI Anthropic, they're probably all furiously now trying to work on a version of this within their systems. You're going to find much more kind of grown up, secure versions being available.
You're also going to find a proliferation of other people out there because remember this was just a guy in Austria, right, you know, who vicode is trying to create the same thing. There's going to be stacks of these things popping up now like whack a mole's left, right and center, I'd imagine.
So it's going to be a wild ride I think for the next few months and fun to watch.
Benjamin Harvey:Well, probably some with nefarious sort of end game as well.
David Brown:Yeah.
100 but this, this gets back to again we've talked about before, which I think where we're going to ultimately end up is we're going to end up with one sort of a federated system where you end up with one AI system that is the one interface that everybody talks to or a, or a select few, but then it connects in the back end to all the specialist AI tools and everything else. So I think you're right. I think this is the first kind of stab at that. Right.
As providing something that everybody can talk to but then it goes off and deals with other AIs and just makes stuff happen in the background. And you're right. I think we'll see a big player come in.
I'm not entirely convinced that it's going to be any of the big players that we have in the game right now.
I think it's going to be a new entrant that's going to come in who's going to specifically design a tool that connects to all of the current tools that are in the market and actually is that top layer that will be the interface that everybody will deal with. And that's my prediction. And maybe in the next five to 10 years we'll probably see that come out and we'll have a clear winner.
But I suspect that it will be a third party to all of the current tools because that'll be the only way to get them all to work together because they're all building their little wall gardens right now. But at some point somebody's going to build something that's just going to work across all of it.
Benjamin Harvey:And do you imagine that would be a bit like an N8N? Sort of. But you have, you can just speak to it.
David Brown:Yeah.
Benjamin Harvey:Sort out your pipeline and your nodes.
Alan King:This is an N8N killer. Actually.
Benjamin Harvey:Eventually I do, I do as well, but.
Alan King:But, you know, give it six months, a year.
But I've said sometimes unpopularly in the, in the group, actually, I don't think N8N's got legs in the long term because it's like the most AI thing I've ever seen that you've got to sit there as a human building this kind of spider frame network of how it should operate. But the AI should do that, that's obvious.
David Brown:And maybe the AIs will build it themselves.
Alan King:Yeah, exactly right.
David Brown:It could, it could be a tool that they decide actually is the easiest way to interface with humans is just to give them one tool to use. This would be hilarious as well.
Alan King:I'll put a bet on it. I'm actually going to put my money on OpenAI being first to market with a Open Claw Multbot type service for their users.
It'll be a subscription service, you'll pay a premium for it. Initially, I think the reason I think about.
I'll tell you what, there's two things that made me think about it in the week was one, when they had to change the name. So the guy who recently built this, he called it claudebot. And obviously Claude Anthropic went, whoa, hold your horses.
Cease and desist letter, Mr. You can do that. So he then changed it to Maltbot, decided he didn't really like the name very much.
He thought it wasn't a great name, and then ended up with Open Claw, but apparently wrote to Sam Altman and said, are you happy for me to call this Open Claw? And Sam Altman said, hey, yeah, you go for it, buddy. Right.
So that makes me already think that Sam Altman's looking at this going, I want to, I want to be good to this guy because maybe I want to bring him in, maybe I want to get him on the side. I don't want to piss him off like Anthropic and say, hey, you know, stop, cease and desist. The lawyers are coming at you.
He's actually kind of put his arms around him and said, yeah, you go for it. You feel your boots. And that makes me think that they're probably already sitting there thinking, we need a win, we need something that's going to.
Just because what this is shown in a way, hasn't it, I suppose, is that you don't need AGI in a way to do some of the things that we thought we needed AGI to do.
So, you know, up until this came along, if there are certain things, agency, autonomy, that kind of stuff, to get to that level, the sort of perceived wisdom was that you need to get to AGI to do a lot of this. I think this is showing that maybe we can do a lot without getting to AGI just by, you know, working out how to package this stuff up better.
So I think OpenAI are probably already in a deep conversation with them. That's my bet anyway. We'll see.
Benjamin Harvey:Interesting. You are right though. OpenAI do need a win, don't they?
Alan King:They do need a win. Yeah.
David Brown:Yeah.
Alan King:I mean, and, and they're quite progressive at baking in functionality into their platforms as well, probably more than the others. And just in terms of trying to get stuff in. So we'll see, we'll see. The question will be will we trust them with all our details?
I suppose we sort of already are to some degree, so maybe they'd be more trustworthy in the current. Yeah, we just, you know, open clause. We're on the subject of Claude then. Ben, you've had a Roast Masters moment with that this week, have you?
I think.
Benjamin Harvey:Well, again, we often talk about our sons sort of driving us on AI and my son sort of got to his limits with Chat GP on some of his game development and some of his where he's been developing stuff in not Fortnite, in Roblox and he was less than sort of impressed with some of the code that Chat GTP was picking out. So he used the free credits he got with Claude and managed to make a. A game really quickly. So he.
But then he kept running out of credits and was getting frustrated so he said to me, if I give you some of my pocket money, can we get a month subscription to, to Claude? Which really made me laugh. I was like, okay, I'll get it for a month and we'll see, you know, compare it.
And I mean he made a website for my wife yesterday in a couple of hours. It looked amazing. He created a lovely Roblox game which again, really, really impressive. So I've. I've kind of started trying it out.
I've re written one of the apps that I did with Chat GTP and it it worked okay. It was, it was pretty good. I mean, it looked like a dog's dinner, but it, you know, it got a lot of the programming better first time.
But then it's a bit of an unfair comparison because I've been wrestling with creating apps and chat GDP for six, seven, eight months now. So I kind of built up a sort of way of working with it, but it's, it definitely got there quicker. So, yeah, I'm impressed with.
I've definitely been seeing on X and a lot online that people are saying they're using Claude a lot more than chat gtp and I've just be, you know, it's been three or four days for me, but yeah, I'm impressed. I'm impressed with it.
Alan King:Well, welcome to the party, Ben.
Benjamin Harvey:Yeah, I don't know if you guys have a similar experience, but I have.
David Brown:I've resisted the urge to go down the sort of vibe coding, just getting myself stuck in that world because I have a sneaky suspicion that it could be a slippery slope and then I'll just end up focusing on my time on doing that.
Benjamin Harvey:It is really, it's really addictive.
David Brown:And I have some ideas about some apps that like, I would never, like, no one would ever kind of do it. I don't think there's any kind of monetary, you know, kind of reward for building the apps.
But there are some fun apps that I think would be cool to create and it would be great to use a tool like that to do it because they're really simple, do you know what I mean? And could be a great mobile app for things like parties.
But yeah, I just, I've kind of resisted that whole side of doing any of the kind of really engineering type stuff because I'm not really an engineer and I know the whole point is that you can vibe code stuff and you don't have to be an engineer. But I kind of did it with Chrome extensions, some fun stuff just for the Chrome browser in the very beginning.
And I was like, I mean, even back then, this was like three years ago and I was shocked at how easy it was back then and it's come on so far since then. But yeah, that's one of the areas that I have kind of resisted the urge to really get too stuck in. So I'll leave that for you guys.
Benjamin Harvey:Yeah, it's really addictive.
Alan King:Core code is great, isn't it? There's no getting around that. I think it is the market leader for that at the moment.
David Brown:Over.
Alan King:OpenAI and Gemini Alan haven't you been.
Benjamin Harvey:Doing some coding as well for Rabbit?
David Brown:Oh, Rabbit.
Alan King:Oh, well that's a whole other. We'll get to that in a moment.
Yeah, I was chatting with Dave before actually I had an argument with the Rabbit actually but I mean with the Claude thing, I found that. Are you using it Ben, for others stuff as well like you know, sort of language writing.
Benjamin Harvey:I've been testing it just, you know, when some of the topics that we were going to discuss today I asked it a few questions. I just like the interface.
I liked my feeling and again, probably next time we chat I'll have a better idea because it's early days for me is that OpenAI is becoming the, the best generic platform for most of your needs. And I think at the moment Claude is. Seems a lot better at the development and the vibe coding but that, that's.
I, I don't have and you know the designs that it does it did for the website were amazing. So my feeling is it's.
They, they've specialized on that and, but my, you know, my open AI has so much knowledge on me that I can say oh, do you remember when we chatted about this? Can you help me go back to that and retweak it? Or you know, my CV did this, you know, can you help me retweak it for this engine?
You know, it's got such a year or two's knowledge on me that it's hard to let go of that. So it's still my generic go to for 90% of tasks. But I have a feeling Claude is much better at the specific, specific coding.
Alan King:Yeah, I agree with that. I find I tend to use OpenAI as opposed to what I call my daily.
You know, if I'm just one of quick queries, quick questions, check something, run an idea, then I go there because it knows so much about me. It's for sort of more detail specific things.
I tend to then either jump into Gemini or sometimes Claude find Claude sometimes a bit bland in terms of its kind of output, but maybe it's also. Even the interface is just a little bit sort of sober, isn't it? You know, I don't know but.
And then sometimes certainly on mobile I find Claude pointlessly limiting and I just don't use it on mobile at all.
But you know, it absolutely has, has its place and first of all of you know, as an API, I think to sit behind things for enterprise in other areas it's probably okay anthropic. You know, I think have done a, done a very Good job there. So.
And I think they probably, you know, outside, I suppose, co pilot, which is just inherently baked into a lot of stuff already, because it is. But outside of that, I think, you know, they're probably the most successful, you know, in. In that environment. So. Yeah.
Benjamin Harvey:You know, I tell you, I think OpenAI's guardrails have got a bit frustrating recently. It's just become quite restrictive, and that's been quite frustrating for doing some legitimate things. You know, wanting to change a photo of.
Of that I own. Oh, we can't do that because, you know, you don't have the. Just that guardrail thing has got a lot stricter recently. I found that really frustrating.
Alan King:Really good guardrail example that involved a rabbit again. Right. So I wanted to keep on rabbit. You can create custom voices. So if I think I want to want my. My rabbit to sound like Magnum P.I. right.
Benjamin Harvey:Yeah.
Alan King:I can basically just. Description of what Magnum PI sounds like, and it will. It will start talking and sounding like Magnum PI Right. So now that's quite a hard thing.
If I sat down to you and said, can you describe how Magnum PI talks? It's quite a hard thing to write. So I went to OpenAI and said, I want a description of how Magnum PI talks. It wouldn't do it because it's.
God, you can't do that because it's a celebrity, you know.
Benjamin Harvey:Yeah.
Alan King:So I had to go to Grok actually. They had no problem with it. Yeah. Here you go. And it give me a beautiful description of the tone of his voice and everything else.
But though, to me, that was an infuriating guardrail. I said, that's ridiculous.
Benjamin Harvey:Like, I've come up about against that kind of thing four or five times this week where I'm just trying to do something legitimately that I have right to the copyright of. And it's saying, no, we can't do that.
David Brown:Really.
Benjamin Harvey:This is my own photo. I, you know, I've taken this photo and it still says, we don't know that. We can't help you. Really.
Alan King:It's just.
Benjamin Harvey:That's frustrating.
Alan King:Completely bonkers. Well, while we're on the subject then, of the rabbit, this leads us into our next question conversation.
Because before, well, we came on, I was chatting with Dave, and earlier I was trying to build a little app for the rabbit, which is a barometer.
So for those listening don't know me, I'm quite interested in the weather, always have been, particularly because of surfing and other things that I enjoy as well. And I've got weather Stations and barometers and various things around the house, constantly monitoring what's going on.
And I thought it would be nice to have a barometer on the rabbit. So I got it to build it and it was fine. That was. That was fine. And then I thought, well, I'd probably add the temperature into there as well.
That would be useful to have in the same. The same screen.
And it did the temperature, but then it started to move the needle and the barometer so it was no longer centered properly for no reason. It just did it. I thought, okay, here we go.
So I started having a conversation with it, saying, look, you need to put the needle so it goes from the center and goes around like a proper dial, you know, 20 conversations later, it still hasn't managed to do this. I'm being quite polite up to this point, so in the end, it's all right. Okay, enough's enough.
So I got really annoyed with it, you know, and I won't use the language here because obviously this is going to be, you know, going. Going out. But.
David Brown:But we can beep it.
Alan King:There was. There was plenty of beeps going on. Right. You know, why on earth for God. Yeah, you've got it and just absolutely ripped into it full tilt. Perfect.
Sorted out immediately. So it's good. It's a bit depressing if this is how it's going to be, you know, to get something done. You've just.
And it was really funny watching it because you can see the chain of thought, you know, after that prompt and it's going. Adam's really angry. We've really got to sort this out this time. We've got to think deeper. It's absolutely bonkers.
But, Dave, that led us to a conversation about the potato prompt, didn't it?
David Brown:So it did, yeah. Which is. Seems like a really random story.
And I had loads of stuff to talk about, but I think this one was quite interesting and relevant to what we were talking about. But I saw an article on Tom's Guide, and I'll basically summarize what the rule is for the potato prompt.
What the author says is, whenever I type the word potato, this is the prompt that you put in ChatGPT. Whenever I type the word potato followed by an idea or argument, I want you to ignore your helpful Persona. Instead, act as a. As a hostile critic.
Your only job is to find the holes in my logic, point out three specific ways my argument could fail, two assumptions I'm making without proof, and one counter argument I haven't addressed. Do not be polite Be precise. So essentially what they've done is this prompt turns sort of chatgpt into a red team.
And for anybody who isn't in the software world, that's just. That is, you build something and then you give it to this red team, who are the people who just rip it to pieces and see if they can break it?
But, yeah, so it, you know, again, it starts to become a devil's advocate. It looks for things like, you know, logical fallacy, survivor bias, it's looking for weak links in the argument.
And I thought it was quite interesting because, you know, it's been known for a long time that if you get aggressive and you're mean to AI, that you get better responses and if you threaten it, you get better responses. So if you tell it, you're not going to give it, you know, you're not going to give it a gift or a prize.
If it gets the answer correct, then it does a better job, which again, I think is worrying.
Do we want to train an AI to, you know, that you have to bully it to get it to do something, and then does it in turn learn that it needs to bully people to get them to do things?
I'm not sure that's the right way to do it, but at least this prompt might help in getting you past the sort of yes man phase, which again, is something that I think ChatGPT has got really bad at recently. It doesn't give you any kind of, oh, well, maybe you should think about that or what.
It just tells you whatever you're doing is great and it never gives you any pushback. And I saw an article in the last day or two where Sam Otwell was talking about. Yeah. That they'd gone a bit overboard on that.
So they're going to dial that back.
The other thing I've noticed with ChatGPT that kind of maybe goes with this is that it started breaking everything out in like two or three word, like, sentences almost. So it almost makes it look like code or like a poem or something really weird. Like, doesn't write in paragraphs anymore.
It does this one sentence, one sentence, one sentence, one sentence, which I find really, really odd. And I have no idea why it did that.
So now every time I ask it to do something, I then have to go back and say, can you write that in paragraphs, please? And then it rewrites it back into kind of a normal structure.
So I found that really interesting, again, that you have to very explicitly now tell it if you want it to kind of challenge Your ideas as well.
Alan King:Yeah, it's really interesting, isn't it? And I suppose in theory, you could put in a number of kind of, should we say keywords that trigger different behaviors. Couldn't.
I mean, you could have potato 100 other words potentially that suddenly change its personality to behave in a different way.
David Brown:Yeah.
Benjamin Harvey:I find the thing that's really difficult with OpenAI at the moment is. So, for example, I was trying to create some videos. I wanted to test it for doing, like a YouTube channel or something.
I just wanted to test the ideas. And I said, you know, I want to. And I listed a couple of writing styles I like.
I put in some information, did some research, and I wanted to see what kind of output it put. And it kind of created these rules for a channel that it almost wouldn't go like. And I'd say, no, I want to be more contentious. And. No, no, you're.
You know, it kept sort of. It kind of created its own guardrails for my own content. It then wouldn't, you know, I understand. I say no in this one. I want to be more contentious.
I want to be less kind of, you know, agreeable. Agreeable. And it was like. It just wouldn't let me do it. It's like, you know, this is your channel, rules of this, this. And it got really frustrating.
So it is. Memory can sometimes work against you.
Alan King:Yeah, I. I have had exact examples of that where I've been creating documents, and because it's got lots of other document information in there, it's. It started pulling in other references, and I had to go back and say.
Benjamin Harvey:No, no, no, no.
Alan King:You only need to focus on the information I provided you in this prompt, not that. Forget about the other stuff for this activity. You know, so I'm trying to narrow the field to a specific topic. You know, it's going on.
But you've also talked about this in the past while pouring that into the conversation. It's like, no, I just want you to stay on this topic.
Benjamin Harvey:And I guess that's where your wild west things like Grok or your. Your local LLMs, which don't have guardrails, become more useful for those edge case scenarios when you just need to do something.
Alan King:Yeah, grok's always the go to. If I hit the rail with OpenAI on something. Okay, that's Grok. It'll definitely do it.
Benjamin Harvey:Grok will do anything.
David Brown:It'll do. It'll do pretty much anything. Yeah.
Alan King:But I wouldn't connect it to an API to anything important.
David Brown:That could be Fun.
The other thing, and maybe we'll segue into kind of the topic that we had said that we wanted to talk about, which was kind of looking at the, the, the kind of what is the whole unholy trinity of sort of what is AGI, Quantum computing and fusion power. And again, as with everything, I kind of. Initially I thought we were probably 20 years away from that.
I suspect we're probably closer to 15 now, maybe even 10, although that feels quite close where we are with fusion. But we have been making, we as a society have been making massive strides on both the quantum front and on the fusion power front.
And I'm curious to hear what you guys think. Think about that.
I think, you know, the reason those two are going to be very important together, I think is because that's what's going to really blow the doors off of the AGI kind of argument. Because essentially you will have unlimited power and you will have unlimited compute.
You can, you know, any AGI or, sorry, any AI tool or system can literally ingest the entire sum total of everything all the time, constantly.
It can get all of the data that's available to it every single day, and it can use all of that basically instantly with seemingly no training in the background. And those jumps from quantum are going to be massive.
So you get like a million times more computing power for 10% of the power consumption that it needs today. Right. So you drastically reduce the amount of power that's needed.
So the power, the cooling, all the other stuff that goes with that, you know, those dramatically decrease. So a lot of the issues we have currently today are going to go away.
And then you combine that with a future where we have fusion power, which once they get it working, in theory, you've essentially got unlimited power.
So even if, even if you didn't have the reduction in power requirements brought on by quantum, you could still power everything and more with essentially no, you know, no, no side effects. And, and I just think those two things, you know, we're, we've seen the Chinese in particular have made massive steps forward in fusion power.
You know, we've got these things, you know, I think the last time we talked about fusion, what was. It was 17 minutes or something like that. And they've got it. Didn't somebody say they had like 40 minutes or something? They've had one running now.
So, you know, at that rate, it's growing exponentially, so it's not going to be long before. Oh, the other thing was, is that they'd made the leap that they Generated more power than they'd used to create the system.
And that was the first time that they'd had a surplus.
So again, knowing the way technology runs and everything else, I think that's going to exponentially grow and become faster and faster and faster, and it's all going to feed. And that's going to be. In my mind, that's. If you want to say singularity. That's the point of singularity, I think, is when all of that converges.
So I'm interested to see what you guys think about that.
Alan King:Yeah, I think it's a really. I think I like what you said at the start. The holy Trinity, isn't it? These three elements coming together almost. It's a bit like fire, isn't it?
Like fire changed the human species ability to do everything right. Without fire, you know, we're kind of nowhere. And, you know, for fire you need heat, fuel and oxygen, don't you?
So when those three things come together, you get fire, you know, and you could almost sort of think with, with, you know, quantum, you know, and fuel and, you know, that could be. Well, so fusion could be the fuel, you know, quantum might be the heat, and then maybe AGI is the oxygen or something. Do you know what I mean?
But you need those three things and then what do you get? Then you get this.
Suddenly you've got this new thing that's just so exponential, so beyond anything we've ever comprehended before, you know, that you've got this super intelligence, this, as you say, unlimited capability and power being driven. We can't even imagine what that looks like, I don't think.
David Brown:No.
Alan King:And then. So my worry about something like that was you end up with runaway, you know, that it's so fast, so quick.
The ability to then control what happens with those sort of three elements coming together. You know, we will be like Amiga basically trying to, you know, stop a, you know, a tank or something.
David Brown:Yeah.
Benjamin Harvey:I think what's interesting is how the three of them feed into each other.
So, you know, there's a big problem with holding the plasma together in infusion and using quantum computing to solve that that, you know, you know, Google already have. Is it Willow, the latest kind of something like that. But, you know, they, they're using quantum as a service now.
You know, people can actually rent their quantum computers. And I think, you know, a handful of players doing this to design new drugs, design the problems.
Looking at the problems of fusion and probably looking at AI problems as well. And I think, you know, you got your, your Magnets trying to hold the plasma together. You've got your.
Trying to keep things at absolute zero for quantum computing.
You've got these, you know, huge problems that, you know, are all kind of feeding into each other but will help solve each other's each of those three things issues really. And so I think when one of those makes a big advance in one of them, it'll benefit all three of them.
Alan King:Do you think that, you know, with that triangle that I described that this, you know, obviously the center of the oxygen, fuel and heat one is obviously fire. Do you think the wood at time the same center of the quantum AGI fusion. Do you think the word there is singularity?
So it feels like it might be, you know.
David Brown:Yeah, I, I think so. And you're absolutely right, Ben. And I think that's why we're going to see it grow exponentially. Right.
Particularly because you've got, you've got the AI tools and everything along with the quantum that you've already got who, you know, where you can perform those calculations and things, you know, at it's like a million times or more, you know, faster and you can do those calculations more. And I think that is going to then drive the fusion side of it as well.
So yeah, it's just I, I think watching it, we're also, we're also in the AI bubble still.
You know, I think we're firmly in that bubble because not one of these other technologies has come along, but one of the two of them is going to either quantum or fusion is going to be the next big bubble.
And I think when it gets and it probably feels like it's going to be quantum at this point because we have quantum computers that actually can be rented and used to actually solve problems.
And I think as soon as that becomes a little bit more reliable and can be seen as an actual tool, I think all of the AI money is going to go that way.
Benjamin Harvey:Yeah.
David Brown:So you're going to. Yeah.
Alan King:With quantum in a moment. Where do you think we are in terms of.
Because everything I've heard about capability at Quantum, it's sort of very good at doing a very like a one trip pony, a single task. It can do this thing, but it can't cope with any kind of complexity of tasks. So. And also these systems have to run at very low temperatures.
They're not sort of running at room temperature or even close to that. So in terms of getting to something that can be kind of used for more complex things, how far away or where do you think we are in that journey.
David Brown:Well, again, I think we're probably 10 years from anything that's usable on a day to day basis.
But I do think that it is useful enough, you know, that the few systems that we have, there's about four or five of them I think actually in different, you know, countries and around, in different, different labs that all work and I would guess that they're probably all working on something slightly different as well.
So they're all going to figure out their own little part of it and then as that knowledge is shared, because it's all academic really at this point point as well.
So I, you know, we will see those systems becoming more and more functional and the more functional they become, they're going to use them to improve themselves and then that's just going to start to snowball and go faster. But I do think we're probably 10 years away.
But there will be a tipping point that particularly the big investors, you know, are going to see that and they're going to start saying, right, okay, now's the time to start pumping money into it. And they will take all the money that's been in AI and they'll put all that development and all of that research money into quantum.
And then I think as Ben said, once Quantum starts to come online a little bit more reliably, they're going to use the quantum to solve some of the problems that we have with fusion and then that's going to accelerate.
Fusion still will probably take five years, but then what you're looking at is you're going to see a big shift again of money going from quantum into fusion.
Benjamin Harvey:I agree and I think what I was surprised because I didn't know a lot about this, but when you suggested it, I looked it up a little. What I was surprised is actually you can rent quantum computing now.
Already in my mind it was still like you say academic and universities in America and China and but you can actually, you know, you have a problem, you want to develop, you know, unravel a protein or create a new drug and you know, you can actually rent space on Willow or, or one of the other ones. I don't know the details but so people are actually starting to do that already and starting to solve real world problems with it.
But you're right on the, the mass adopt. I think one of the big American banks is using it for futures and options.
David Brown:Of course they are, of course they are.
Benjamin Harvey:But you know, yeah, of course, but there's real world applications.
I guess the thing that everyone obsesses about is encryption and how it's going to be able to destroy current encryption and like how we need to future proof government organizations and the world around that and sort of come up with quantum proof encryption which people are working on.
Alan King:They call it Q Day, don't they?
David Brown:Yeah.
Benjamin Harvey:And again we'll probably, that'll probably blindside us. It'll probably happen quicker than we'll probably think.
It's always three, four, five years away and then suddenly it'll be six months away and no one will be prepared. And I think there's probably a lot of issues like that with quantum as well as benefits to society.
ey look like something from a: Alan King:Version of a computer. So yeah, they look quite in the, the lobby at IBM in Central London actually where my nephew works and I was in there a while ago.
Yes, it's fun to see.
David Brown:Yeah, it's, it's, yeah, it. I also, I'm trying to think about how to, how I want to say this. I also suspect that there are, it's probably a lot further along in government labs.
Benjamin Harvey:Yeah.
David Brown:With things like defense and military and intelligence. I, I suspect that they're, that they have been working on this a lot longer and you know, particularly around the encryption and that sort of stuff.
And you know, yes, the, it will be able to break the encryption that we have now, but it will also be able to create new encryption that will be that much more difficult to crack. So do you know what I mean? All it's going to do is the new encryption is going to replace the old encryption and it'll still be fine.
Alan King:Develop it first.
David Brown:Yeah, that's assuming the AI doesn't get in control of it and then use it to take over everything. But yeah, you know, it is swings and roundabouts I think.
But I do always worry about the kind of what's going on with the military and the government stuff in the background because all the governments will be working on this, you know, because everybody's trying to get, you know, if you can get a six month lead then you know, you, you control everything. Like you can get anything you want again with the encryption and breaking encryption and all that.
You know, whoever's got the most advanced one and you know, can, can get access to systems. It's, it's going to be interesting.
Benjamin Harvey:Let's say you probably assume, as you say, China and America are well ahead of where we think we are now.
David Brown:Yeah, they'll be, they'll be miles ahead of on capability.
Alan King:So I suppose the overarching question here is, is this a good idea?
Because, you know, as a species, should we be looking to develop something so powerful that we almost certainly wouldn't have control of it and it's vastly more intelligent than us? That, that, that, that feels to me like that could be an enormous mistake.
Benjamin Harvey:Yeah, understatement.
David Brown:Yeah, it was an enormous mistake. Yeah. It's too late now.
Alan King:But I mean, but it's not too late, is it? That's the point. We could pull back on this if we chose to. I mean, is it a bit.
Look, so let's say someone does get fusion, maybe they, they should go, okay, let's just delete that, let's delete all the information of it. Because maybe this is, you know, this is a bad idea. Or if you get to AGI, you know, we just need to shut this down. This, this is too.
David Brown:Never happened. It'll never happen because people, people are the weak link.
Like ironically, the AGI, once the AGI came, it may actually decide that the worst thing possible is for it to exist and it could turn itself off. Right. Like in theory, because it's very clinical and cold and, you know, analytical and doesn't have emotion.
At least we think it doesn't have emotion, whereas humans will always be the problem.
Alan King:It's interesting, isn't it? So in terms of the fusion side, what sort of timeline are you thinking about that? Because for me that feels the slower one.
I was in Cullen Laboratories in Oxfordshire a few years ago. I was working at IMA Key and you know, the guys there were sort of saying to me, this, this is 40, 50 years away, you know.
Now I think things have moved forward a bit since then and it's probably not quite as far away as that, but in my mind it's, it's still maybe 20 or 30 years away unless there is some kind of huge step change in how we do it, created by AI or something that gives us this insight that we as humans never worked out. It gives us the key to unlock the thing. But my instinct is it's still quite long way off.
And when we read about these kind of breakthroughs, often they're often more hype cycles that are driven so they can get more funding and investment for the research laboratories. We see that with OpenAI constantly touting AGI is a way of Getting more funding.
And I suspect that, you know, the sort of fusion centers are sort of operating on a similar basis. They have a little breakthrough, let's make a big thing of it.
We'll get more money, you know, so I'm sort of, I sort of feel like it's probably still 20 plus years.
Benjamin Harvey:Away, but I don't have a clue, if I'm honest.
David Brown:I mean, I don't have a clue either. I, but like I said to me it feels like 15, 20, and, and that's until we have one that's not multiples, but, you know, and then it becomes it.
It will be really interesting to see who develops it first and where it is and then from there how it might be used and how effective it might be. Because you say like if, if, if Europe wins, right, Then you, if you have one, you could literally power all of Europe with it.
You know, if it's built at, at any sort of a scale other than like a small one in a lab. But I think the first commercial one, you know, you could power from what I understand probably most of Europe with.
Alan King:Well, I think there is a limitation there in terms of distribution. That's the problem.
David Brown:So you have some physical.
Alan King:You're going to lose so much energy through the cables that you actually, you'd have to have much more local fusion reactors everywhere.
David Brown:Same reason you can't just set up a big solar array in Africa and then power the whole planet, which you technically could do. You just can't get the energy around.
Alan King:Actually, possibly the way to do it would be to have a fusion reactor in space and then just beam the energy down to where it needs to be. But that's not as far fetched as it sounds a few years ago, and it's actually doable.
Okay, so then you just touched on a point there, Dave, about who gets there first, because I think that's quite important.
I mean, it's hard to see it not being America or maybe China just because of sheer resource and scale and that they would take control of it very quickly. And in the current climate that we're now living in, neither option seemed particularly appealing to me. So how you guys feel about this?
But I mean, whoever did this, assuming that the AGI doesn't take over, they control everything, don't they?
David Brown:What with fusion.
Alan King:Yeah, well, with all of it, the three dimensions. With all of it, suddenly you've got this elixir of quantum AGI infusion at your fingertips, you know?
David Brown:Yeah, no, you're absolutely right.
Which is why I think it is very much sort of a West versus east kind of argument, I think, because I think if anyone in the, in the west, the west in inverted commas got it, I think you'd have a lot more potential for there to be collaboration around countries in the west, same in the East. If China or Russia or someone like that gets it, then I think you start to see, you know, them sharing all that among themselves.
So it's a, it's a West versus east race at the minute, I think, on, on all fronts. But yeah, it's. Like I said, it's going to be interesting, you know.
Benjamin Harvey:Do you know who's got the longest one at the moment that you're saying possibly around 40 minutes or China?
David Brown:I think China was the last one. That was the last story that I saw with that. It had the longest runtime and it kind of, you know, worked something out.
But, you know, again, like you said, Alan, though, you see stories coming out and they're talking about we've used this new, you know, this new metal to try and do this, or we've used this new thing and it actually gives us that. And, you know, so they're, they are still very much testing everything. I mean, there's a huge.
I know they also, they, they, the, the, the whoever that is, you know, talk about the safety aspect, but I just can't help but wonder if you're building a. Literally building a sun.
Benjamin Harvey:Yeah.
David Brown:Surely there has to be some sort of.
I know that in theory that what happens is if it ever sort of like was going to explode or anything, that literally by the nature of it, it would stop. Right. The whole reaction would stop anyway.
So if it developed a crack or like a leak or anything like that, like, the whole thing would just shut down naturally anyway because it's such a contrived environment that it runs in that if anything, literally anything happened to that environment that just got it off by just a little bit, that it would shut down and stop working. But I still can't help but think that, you know, what's the, the potential of disaster from that? You know, I don't know. I don't know.
Maybe it is totally safe, but, you know, I'm not sure.
Alan King:What about, what about all the people who are already invested in other things like oil and coal and gas and everything else, would they be quite keen to stop this from happening? Because this is.
Surely it's like the, the story that always used to go around in the 80s and 90s about someone invented a battery, that it lasted forever clearly that's not true, but the big companies shut it down because they didn't want that out there or the light bulb that lasts forever or whatever. Do you think there would be some resistance to try and stop this from happening?
David Brown:I reckon they're the ones funding it.
Alan King:But only one can be in control.
David Brown:But no, I think again, it's past the point of them being able to shut it down. If they were going to shut it down, they would have done it already. So I think it's past that point.
So now if you're a big oil company or you're a huge energy supplier or whatever, I think at this point you start going, right, we're going to start putting money into it because we want to buy the IP and they want to have some sort of control over it. You know, oil as an energy source, you know, we're very slowly replacing that. Look at the Middle East. The Middle east has already started.
You know, they're looking 50 years ahead and saying we're not going to. Oil isn't going to be a major part of our economy anymore. We have to have other ways to make money.
So you've even got places like Saudi are setting up tourist cities on the beach and you know, they're really trying to change their reliance on oil. So I think everybody thinks oil is, you know, going to become less and less important as time goes on anyway.
So I can't help but think that they're probably not putting a lot of investment in a lot of these programs because they want to have some sort of control over it once it comes out.
Benjamin Harvey:I mean, Saudi is really heavily investing in obviously sports, but also AI, hugely investing in AI. So, you know, they are thinking about a post oil world, even if it's quite a long distance from.
But the other thing about, you know, just going back to the quantum thing for someone was saying on one of the videos I watched that it'll help with some of the material science problems, developing new sort of whole new branches of material science in a way from the quantum computing that can then help feed into the fusion restrictions. But yeah, is again on the dangers of fusion, I don't have a clue. You know, creating a little sun somewhere in Europe.
Alan King:What could go wrong? Holes with cern, haven't we?
Benjamin Harvey:So yeah, yeah, get rid of all your junk, you know, you're moving house, just put it in the black hole.
Alan King:You know, there were a lot of people very concerned we were all going to get sucked into a black hole with the nuclear bomb.
When they first tested that if you watched Oppenheimer, you know, it was genuine concern, wasn't there, that this thing might just go and go and go and go. You know, it just creates a tame reaction and off you go with the three of them. Then, which comes first?
Do we think AGI, then quantum, then physical fusion? Is that the order of play?
David Brown:Do we think.
Benjamin Harvey:I think quantum.
Alan King:You think quantum before AGI? Okay, that's interesting. Dave, what do you think?
David Brown:I think in. And this. This will may be slightly contentious, but I suspect in.
In most practical terms, we've probably already reached some level of AGI, maybe AGI light or something, but I. I think we're probably already there. So I think the next one, I think quantum will come before fusion. I just think quantum's further down the track than fusion is at the minute.
And I also believe that quantum will help advance fusion as well. So, yeah, I think it's probably that order, but I think in all basically practical ways of thinking about it, we've probably already reached AGI.
Alan King:Interesting. Yeah. I'm not sure I entirely agree. I think there's still quite a long way to go, personally on that.
Just around this sort of level of abilities for original novel thinking.
I think what we've reached is something that's incredibly good at taking human thinking and then passing it and then remodeling it and comparing it and coming up with new ideas, but built out of that human thinking, but in terms of its ability just to think of something, an entirely new way of doing stuff that no one's ever even vaguely contemplated before. I don't know that it's. And to me, that's what AGI should be. It shouldn't be.
You know, sometimes they describe AGI as being, you know, being able to do everything at a human level or slightly above. And to me, that's kind of first kind of human level, isn't it?
So, you know, it just means it's good at everything at human level, where humans are not capable of doing. You know, a human typically might be good at a few things and then have a very kind of high level on everything else, but it can't go that deep.
But the idea that you've got a system that can go deep on everything that humans can do, but it's still. It's a bit like this. Right. So if you went and do a degree. Yeah. And then you go and do another degree, another degree, you're still at degree level.
David Brown:Right.
Alan King:You've not gone to PhD. Right. So. So. And that's sort of How I feel about it for AGI needs to go to the, you know, to that next level, which is beyond human thinking.
And I feel that that might be something that large language models can't quite get us to, that we need a different kind of system, one that processes world data, visual data, almost replicating the visual cortex and other things, other inputs, other sensory inputs, that you've somehow managed to build a model around all of that, then maybe that gets used to this other thing, but who knows, right? And the term AGI keeps getting rebranded for marketing purposes by the big AI companies anyway.
David Brown:Well, I also think that they keep moving the goalposts because nobody wants to admit that it's already here. So do you know what I mean? Like, they've changed the definition of the Turing Test because it passed the Turing Test ages ago.
And they're like, oh, well, yeah, no, that's not applicable anymore. It really means something different now.
And it's like, well, no, that was the test and has been the test for ages, and we blew that test away a long time ago. So now you're just moving the goalposts. My counter to. I totally get what you're saying.
My counter to that is, I think everything that we're seeing from the general public side is a hugely handicapped version of what's happening behind the scenes.
They've put so many rules, so many guardrails, so many restrictions on what the AI can do, what it can say, how it behaves, that we don't actually know what's happening behind the scenes. And that's where I think. And that's why I say we've probably already achieved it.
It's just, it's so hobbled at the minute because people are so absolutely terrified of what might happen if it, if anybody knew what it could actually do and if it actually got out into the real world, that all of the systems are so constrained that it just doesn't appear that it's that way. But I have, that's my. Again, having worked for organizations where, you know, people will.
People are very cautious about sharing information because, you know, people wouldn't understand or people wouldn't like that, or it's scary. So we can't say that. I, I just, I suspect that it's a lot further, a lot further along than anyone knows.
Alan King:Yeah, I, I, I, I think, you know, with the current level of AGI AI that we have as well at the moment, you know, I think that if you put that together with quantum infusion, that's already incredible.
David Brown:Right?
Alan King:I Mean, I mean, just look what we've seen this week with, with, you know, the, the open claw mult thing, we're starting to see that this, this large language model vehicle can be put in different ways or functions to create outputs that are quite extraordinary. So imagine, then you pair that up with quantum infusion. You give it that extra sort of turbocharge if you like.
So I think we can achieve extraordinary things already. And then over time. Yes. Then the level of AI intelligence creeps up as well, and the thing just kind of becomes unbelievable.
But it's, I, I, I think as soon as you get quantum infusion, even paired with what we've got today in terms of AI, you've really got something very interesting. Yeah, no doubt.
David Brown:You know, yeah. 100.
Alan King:It's gonna be, it's gonna be a wild ride, isn't it? I think we're probably running out of time, really. We're on, we're on over an hour, so we've broken the hour again, so we probably. Nice.
David Brown:That's good wrapping up, but.
Alan King:Okay. All right. I mean, any final thoughts on, on any of.
Benjamin Harvey:No, no, I think, I think it's a topic I didn't know a huge amount about a week ago. I've learned a lot more, but it's really fun.
I think particularly the fusion for energy needs for the world is a separate thing from quantum, and AI is a really interesting push for the world.
Alan King:I suppose there's a, One final question on that, actually, is that you've got three elements here. I see fusion actually as a really positive thing for the world because, you know, I mean, undoubtedly that has to be worth striving for.
Okay, so then, so if that's good. Right, and maybe it's a traffic light here.
David Brown:Right?
Alan King:Red, amber, green. Right. So, so have quantum and AGI then, which is red and which, you know, which is amber, which is great. Is Quantum. Should we be pursuing that?
Is that good for the world? Is AGI good for the, the world? Or actually, should we put, should we really put a red light on those two?
Benjamin Harvey:I think quantum is obviously like, like anything.
It's good, it can be used for good or for bad, you know, but just the fact that you'll be able to probably defeat diseases a lot quicker, come up with new medicines, new molecules, I think will be amazing. You know, accelerating those things that we're trying to elevate humanity for will definitely come with quantum.
But obviously there's a downside to that as well. It can be misused, but I think that's true of most things.
David Brown:Yeah, Yeah.
Benjamin Harvey:I don't know.
David Brown:Again, I think it's too late. I think it doesn't matter. It doesn't matter what the traffic light says. It's, it's, the horse is bolted. Yeah, it's, it's already done. It's out there.
People have started it. It's too far down the track. If we were going to stop it, we would have had to stop it at a much, much earlier point.
What we really need to focus on now, I think, is really trying to get ahead of the societal morals and ethics that we want to try and implement around this because we've not been very good at forward thinking about what might happen and how we might need to deal with this stuff.
Like, you know, we're just now getting to the point where we're thinking, hey, it might be a good idea to keep kids off social media and, and, you know, to at least make a token attempt, you know, to keep kids under 16 off of there.
And now, no, you know, parents are still going to sign their kids up and they're going to do it and you know, because they're like, oh, I want to get on there. But I think, you know, how many years later, what, 10, 15 years later, we've now worked out that maybe that's not the best thing.
Will we learn anything from that lesson? No, we absolutely won't.
We'll just barrel ahead as hard as we can on the technology as fast as possible and the ethics and everything will catch up later. But yeah, as far as traffic lights, like, it doesn't even matter. I think we're, we passed the traffic light a long time ago.
Alan King:I'm 100% aligned with that. I think cat's out of the bag and you can barely put the cat back.
David Brown:Exactly.
Alan King:We should definitely have a conversation about the social media thing again at some point as well with the Australia stuff going on. And it's amazing, actually. I listen to obviously a lot of tech podcasts. You know, how many people in the tech world are actually against that ban?
And I, I just thinking like, what's wrong with you? You know, 13 year olds don't need to be on this stuff. But anyway, that's the conversation for that. We could do another hour on that.
Benjamin Harvey:Yeah.
Alan King:So I'm gonna wrap up. Thank you everyone who's got this far, who's been listening.
Just to reiterate again, the email, if you want to comment on what we've been discussing today, it's allianiorg.org and you know, we'd love to hear from you. What do you think about the show? What else could we be doing? Any other segments that you'd like us to do?
Get in contact and let us know what you think. Ben, Dave, thank you very much for your time, as always, and I'm looking forward to. To the next one already.
David Brown:All right, we'll see you soon.
Benjamin Harvey:Thanks.
Alan King:Bye.
David Brown:Bye, everyone.
Alan King:Take care, guys. Sam. Sa. Sam. Sa.
David Brown:Sa. Sam.
