In this episode of the AI Evolution Podcast, Alan, Ben and David dive deep into the latest developments in the world of AI. From testing the limits of Sora 2 and video generation tools to discussing AI agents, automation, and how small businesses are using GPT to streamline operations. This episode is packed with real-world use cases, insights, and a healthy dose of skepticism.

We explore:

• Why AI video is still hit-or-miss, and where it shines
• The challenge of maintaining authenticity in a world full of “AI slop”
• How AI tools are reshaping creativity, education, and hiring
• The future of agentic workflows and what they can’t do yet
• Why trust in AI outputs is more crucial than ever
• Concerns around AI replacing junior roles and what that means long term
• The viral “X1 Robot” and what it really says about robotics hype

If you’re navigating the intersection of AI, media, and the future of work, this one’s for you.

🔗 Subscribe, share, and join the conversation!

#AI #ArtificialIntelligence #Sora2 #AIAgents #Automation #FutureOfWork #GPT4 #ContentCreation #Education #TechPodcast

Effective, you know, most reliable way to deliver any type of service or anything.

So if, if you don't have any money, you're going to end up dealing with robots and AI all the time.

And if you have loads of money, then you'll get to deal with humans.

Well, hello and welcome back to the AI Evolution podcast.

As usual, we've got Ben Harvey and Dave Brown here with us today.

We're going to tackle the latest, greatest things that are happening in the world of AI.

Ben, how are you feeling?

How are things?

What's been going on in your world?

Yeah, good.

I've been playing around with the new Sora 2, which I've been desperate to get hold of for a few months, but because it's not been out in the UK yet, I had to jump through some hoops.

But I've been playing around with that, which has been really fun.

It's really interesting to see how the lip syncing works.

I was, I was impressed with some of it.

I, I think when you compare it to the VO3 and the, some of the other steps that people are doing to get consistent characters, it's still, it's still great for a clip, but the moment you start to try put three or four clips together, you then have to go to other tools to, to create your models and to create your consistency.

So it's, it is, it is a lot better.

The audio is fun, there's, the sound effects are really fun.

But for putting together, you know, I was keen to try, put together a, a one minute little film about a terrible counselor.

And I tried it and every scene had a different look, even though it's the same character, same prompt, tried the same seed, all those things.

So it, it's just not quite.

I was really excited and then slightly disappointed.

But yeah, it's good, it's impressive.

So.

Yeah.

So.

Well, first of all, I want to know how you accessed it because obviously, you know, we're all in the UK and I've not been able to access that, so I'm interested to know how you got around that.

And then the other thing, I suppose is with VA3, doesn't that have some kind of software platform that allows you to kind of edit stuff?

And I'm wondering, with Sora, presumably it doesn't have that.

It's just a case of a text prompt and you get what you get and then it's up to you to try and piece it together.

Is that the difference?

Yeah.

I haven't used v3 and anger, so I've just seen what other people have used it for and I.

They probably using other AI tools as well to keep the consistency in the characters.

So I've had to download some other things.

I've.

I've downloaded Comfy, which allows you to use Lora models and consistency.

But it's just, you know, it becomes quite a ball, like quite quickly.

But yeah, it is fun.

I've accessed it through a VPN at the moment and got a code off someone because you have to have an invite code as well through Discord.

Sorry, I'm going to interrupt you there because I've obviously thought about a vpn, but it's on the App Store, isn't it?

So would you.

Does that mean you have to kind of convince Apple that your entire Apple account is now based in the us?

I can't.

I can't use it on the iPhone.

I had to use it through the website, through the OpenAI website.

So, yeah, okay.

Which is a different experience.

And it seems very set up for social media output.

You know, everything's portrait, um, it's not, I mean, you can change that, but it just seems set up for doing quite quick content, humorous content for.

For social media.

So I, you know, I've only.

I only managed to get hold of it two or three days ago, so I haven't used it a lot, but I've created maybe 10, 15 videos from it and created a script with ChatGPT about this idea I had, as I say, of a terrible counselor who starts cracking onto one of his clients.

And I just wanted to make a really.

I just wanted to use it as a test case.

But as I say, every scene looked different.

So then I started having to go down the rabbit hole of trying to, you know, create something through other tools that I could then bring into sora.

So, Dave, have you, have you had a play with it or seen lots of SORA videos?

I've seen lots online.

So I've seen tons and tons of videos online and half the time what's really funny is my wife gets annoyed every time she sees one.

She's like, God damn it, it's another AI video, blah, blah, blah, blah.

And I'm.

I've kind of come around to a lot of them are quite funny and are, you know, a lot of them are humorous, you know, like the bear jumping through the window and all the other stuff, you know, that they've got.

And I quite like them.

And I think it's been quite.

Again, it's its own thing.

And I think all of the AI Video content and all the other stuff, it's turned into, it's its own thing.

And I was talking to someone about this the other day and you know, they were sort of lamenting the fact that, you know, that there were, there was all this AI video and that everybody could do everything.

And I'm like, yeah, but the thing is, is that it makes, it makes what we're doing here, right?

The three of us together.

Maybe we even live stream this, right?

And we go, look, it's the three of us, we're sitting here, we're having a conversation.

And that sets what we do apart from the AI stuff.

So fine, if your company wants to go and they want to write a script and they want to create a training video from AI, first of all, it's terrible.

Still, the AI is, it's not, it's still not close enough to being a human, I think yet that it, that it's still annoying.

But the humanness is now what sets us apart.

So the people who are actually creating real human content, it's actually making us even better and more engaging than we were before.

I think in a weird kind of.

No, I agree, ironic way.

Do you know what I mean?

It's the authenticity of it.

Yeah.

Which, which brings it to the fore.

And I've started, you know, a lot of my podcast clients and a lot of my customers that come into the studio, I'm now saying to them, look, you should be doing more live streaming.

You should be going out there and driving home the point that, hey, I'm a real person and I'm sitting here now, this is not an AI tool and I'm engaging with you and I think that sort of stuff is going to rise to the surface.

I think we will end up, and we've talked about this before, what AI is doing is it's killing the bottom end.

So if you had a tiny startup company or a small company who needed to create a two minute video that they wanted to do for an ad for themselves or what, historically, they probably would not do it at all.

And if they did do it, you know, they, they try to get a videographer, someone to come out, they would push you down on price as much as possible, so you'd end up making a few hundred pounds from doing it.

And you know, yeah, okay, you might have a project to do and it's a few hundred pounds, but that's.

All of that stuff has gone out of the market now because people can go and do it themselves.

But again, ironically, what it seems to be doing is it's also showing people a.

That it's a lot more difficult than it seems.

And just because AI can technically go and make something that looks like a video, it isn't actually a video.

Yeah.

And do you know what I mean?

And for what you were just saying, it's like you can't get the consistency, you can't get the consistent look and feel.

Still, you know, the tools aren't there yet.

Ten, well, five years from now, they'll probably be specialist tools that will.

Yeah.

And you'll be able to put in and it will, you know, they maybe will have had that cracked.

But I think right now it's still, it's entertaining and I do love some of this stuff that's coming out.

People are doing some amazingly creative things and I do think the AI in general has opened the door for a lot of people who are creative and think creatively, but they maybe didn't have the manual skill to do whatever it was.

Maybe you're, maybe you're, you're a hidden musician, but you don't, you haven't gone through learning how to play the guitar, but you know what sounds good and you know what you like.

And so you can now use AI to get that out.

Okay.

It's a different thing, but you've still got the creativity there and so people are able to, to use it for that.

And we've, we've seen this for years, haven't we, in music?

You know, with DJs and hip hop and everything.

You know, it's, it's new tools, new ways of creating Auto team.

Auto Tune opened up a whole, A whole, you know, millions of people who would never have sang before now are coming out and they're singing because there's Auto Tune.

So it's, you know, it's, it's helping fix them and they're coming out and tell you what, some of them are making millions and millions of dollars.

Yeah, yeah, absolutely.

I agree.

I do get quite upset when I sort of.

I constantly see the word you know, AI slop appearing all over the Internet, you know, AI slot, blah, blah, blah, and often related to saw in the videos and stuff, you know, and it is what it is, isn't it?

You know, and I don't think it doesn't, it's not replacing human videos.

People aren't, People haven't stopped making human videos because of it.

And actually, you know, I'll be completely blunt.

I think that a lot of human videos are slope.

Yeah.

You know, yeah, that's the point I was going to make.

Yeah.

We've been making slop for a long time.

Right.

And you know, the other thing I would always say about the AI stuff as well is that it's made by humans.

Right.

It's not made AI don't wake up in the morning and think, hey, I think I make a video of a bear riding on a horse.

A human thought that and prompted it and put that out.

Yeah.

Downloaded it, uploaded it to YouTube or whatever platform.

It's humans doing it.

So it's humans controlling at all.

And I don't see a problem with that whatsoever.

You know, I obviously, I think it's quite distinguishable still.

You know, not because of the.

Necessarily because of the uncanny Valley stuff, but more just the fact that what people make with it is generally so extraordinary.

You know, you couldn't do it in real life, but you can do it in AI.

So you can usually tell quite quickly when something's AI well, because that could never happen in real life, could we?

You know, I think it's fine, you know, it's.

It will find its balance, won't it?

It will find its place.

Yeah, yeah, definitely.

I think there's a slight segue here, as with something you just said that made me think of it and think about agents.

That's all right.

What's interesting, I wonder if you could set up an agent to tell the AI to just go and create a random video every morning.

I'm sure you will be able to do that.

And if it's not possible, I get.

That it's not doing.

I get it's not doing it itself.

Right.

It's not just waking up and thinking, I want to create a video.

But the first step I guess would be using Agentic to say at every day at 5am I want you to create a 90 second video on whatever you want to create it on.

Yeah.

I didn't see why I have an agent every Monday that creates a newsletter for me.

You know, it's no different to that is it's just.

And it goes and finds the relevant information.

I wonder what it would create if you just said create it on any topic that you want.

It's up to you.

Yeah, be that'd be interesting experiment.

Be like the trading bots that are sort of.

Do you see that?

Trading.

I think we talked about it last time.

The one that was.

If it loses a certain amount of money, it gets shut down.

So it's in the agent's own interest to keep making money.

So.

Yeah, but yeah, I had a couple of points.

I think.

I think you're right.

It is a completely different thing.

In the same way that McDonald's is different than, you know, a roast dinner or a Michelin star restaurant.

It's a.

It's.

But I think.

And also, I totally agree that, you know, when you look at the amount of design done with Comic Sans that, you know, you don't get designers going around saying, our teachers are creating slop.

They're just using tools that they have in their place to create stuff to communicate what they need to get done without spending hours doing a design degree.

And I think.

I don't know if you've seen the fake camping AI videos.

They're quite fun as well, you know, interviewing someone in campsites and, you know, there's a caravan rolling off a cliff and they're quite sweary.

But they're really funny.

They're really funny.

But, you know, I think you're right, David.

It just allows people to express really creative ideas in a fun way and I get it.

But then there's a place for authentic human led, you know, filmmaking as well.

There's room for everything, I think.

Well, and there's a whole sort of theme of stuff I've seen on Tick Tock, which is kind of AI satire, which is stuff that you wouldn't really necessarily create with humans or no humans would necessarily want to do it publicly.

So.

And so there's a whole series of where they sort of satire shops in retail, you know, places like Tesco's or.

And it.

It's sort of, you know, it's.

It's as if the shop is doing its own advert, but it's basically saying how terrible it is, you know, and if you want to come and buy, you know, a roll from us, you know, we'll make sure that we fleece you completely and, you know, if you come to Costa or whatever, we're going to make sure.

I better not say that.

But, yeah, you get the point.

And that sort of satire and stuff that, you know, is very quick and then able to produce is very good, actually.

Yeah.

And just wouldn't be possible without AI really.

Yeah, No, I agree.

And you're right.

You're right, Alan, as well.

The amount of slot that we create out there as well is just, you.

Know, I think we've seen with music as well already with AI music, because there's a lot of AI music out there now apparently, you know, Spotify is sort of every day removing it from.

From their, you know, platform.

But you know, the reality is that humans haven't chosen to go and spend most of their hours listening to AI music either, have they?

They're still listen, they're still listening to the stuff that they have connections with, with real human being they understand now recognize and you know, perhaps are fans of or whatever, you know.

So I think an AI music.

Yeah.

For background music, for lists, for wallpaper, whatever, you know.

Yeah, if you're making a video, you want the rights free bit of music, great.

But you know, if you're going to go and choose an artist to see perform, you're probably not going to go and see an i1.

If you're going to sit down and spin an album on Spotify, it's probably not going to be AI at the moment.

So, you know, that's not to say we won't get to that stage for some people one day I think we will.

But you know, it's not just suddenly overtaken and made everything else irrelevant, has it?

And I think with video it's the same, you know, it won't just replace everything and that's it, you know, we'll still continue to do what we do.

So.

And even things that, like the gorillas or the ABBA thing that's in London, you know, they're humans doing that behind the scenes and even the people I know, they've been to see that say that's a one trick pony for them.

It was amazing to see technology but they don't want to go to something else like that.

So it's.

Yeah, it's interesting.

Yeah.

And I've created music using AI tools and you know, it can be quite involved actually, you know, it's not, it's.

Not just a case of, you know, same with videogen.

Yeah, doing video generation is quite involved as well.

It's not, it's, you know, it's not just a case of typing a prompt and off you go.

If you want a good output, it's quite involved.

I saw an article a long time ago and it basically said that what's happening is when you're using AI is you become an art director.

Yeah, exactly right.

And so you have to understand that traditionally that's what an art director would do.

And so you're in this weird position where you, you still have to think of every single thing that you want it to do.

Yeah.

And actually that's, that's the hardest part.

Yeah, the hardest part.

If, if anybody out there has ever tried to sit down or if you guys have tried to sit down.

And maybe you're going to, maybe you're going to do a short film or you're going to do a, even an ad and then you've got to sit down and kind of storyboard what it is that you want.

Like going from I want to do an ad to what does that ad look like?

And it's like, well, I'm not really sure.

And you're like, well, maybe I want it to be outdoors.

And then you go, okay, well what kind of outdoors do you want?

City do you want?

Like, what do you want?

And you getting into the detail of the vision of exactly what it is that you want and then being able to describe that in words to then get it out.

That's still the creative part.

Yeah, that's the most important part of the whole process is actually knowing what you're trying to do because you.

Anybody can go hire a film crew, but if they don't know what they're doing, then it doesn't help anyway.

Yeah, yeah, right.

So it's still, it still needs all of that sort of stuff involved.

Yeah.

I've spent about three weeks doing pre production on something we're shooting on Monday.

You know, the amount of questions there are, you know, what cups are we going to have?

You know, what shop?

Not, you know, everyone thinks the bigger things.

What shops?

You know, what shops are we going to have?

You know, start with a wide.

I mean, you know, you speak to the DOP and.

But hundreds of hundreds of decision making process processes and you.

And it takes, takes a lot of time, takes a lot of energy and some of it's, some of it's really dull.

So yeah, it's the same.

Exactly.

Same with all the art direction.

That's fascinating.

Yeah, for sure.

Well, you know, and I think, you know, it's like you said at the start, Dave, you know, this, this is, you know, it's, it's here to stay.

It's not going away.

And you know, I think it will just get used to it.

And you know, I think a lot of people when they first saw it did kind of go, oh, you know, this is the end of everything.

And then perhaps over the weeks I kind of realized that, you know what, it's probably fine.

It's going to just be what it is and it'll be out there.

I remember, you know, back in the sort of 90s and 80s people complaining about synthesizers and, you know, DJs using decks and mixing and sampling and all that, you know, and everyone, the music thought that was the end of the world as well.

But, you know, here we all are.

Yeah, well, we mentioned agents a minute ago.

So why don't we pivot into that then?

Because I think it sort of does relate a little bit to this.

As we said, it could be used to drive this content in the future.

I was quite pleased and interested to see GPT launch their agent this week.

And obviously we are in the middle of what I would now call the browser wars.

Everyone's releasing a new browser.

I mean, how boring were browsers for 10 years?

Like, nothing, literally nothing ever changed browsers, ever.

It was just like browser to browser.

It's like the dullest thing in the world.

And then, you know, in the last sort of six months, they suddenly become really interesting.

And, you know, we've seen, you know, offerings from Gemini, GPT, you know, Claude and all doing different things.

Perplexity is quite an interesting one.

Particularly their mobile version is very good.

It can really jump around on your phone and open things and get videos for you and play and do all sorts of stuff.

So I had a play about with the ChatGPT one when it launched.

Atlas, is it?

Yeah, yeah.

You know, I thought, well, okay, yeah, it's quite good.

You know, you can sit there and watch it clicking around, doing stuff.

But quite quickly I kind of went, but what will I use it for, actually?

What am I actually going to use this for?

And I sort of run out of ideas very quickly and even put it in our WhatsApp group, in the area of WhatsApp group, you know, difficult.

Any ideas?

And really there was a kind of a nil.

One or two people came back with a couple of things, but there was nothing that interesting that you sort of think, oh, I must try that.

So I did put it to a little bit of a test the other day.

So I.

And I apologize if my door bottle goes, because I'm getting lots of deliveries at the moment because we're going to have some plumbing done at the weekend, which involves radiators.

So I thought to myself, okay, I've got a list of radiators.

Now, the plumber sent them to me and I've got those.

And I thought, but I wonder if I could get the agent to, you know, if I give it the instructions to go and find the same thing.

Imagine I didn't have a handy plumber telling me what I needed.

So I set the agent off.

I mean, you know, it wasn't great.

It sort of clicked around for a bit, come back, some stuff.

Completely the wrong sort of things that you know, wouldn't fit, wouldn't work, wrong colors.

And I thought to myself, well, maybe I've not given it enough detail in the.

The instructions.

So then I give it a sort of much more highly detailed instruction, literally copying and pasting all the information that other radios I'm actually going to use, you know, to see if it could go and manage that.

And again, it just got itself into a muddle, really.

And, you know, if you were relying on that to do the work for you, it wouldn't work.

You would end up having to just go and do it yourself.

And I just got to thinking, well, you know, we are sort of at a stage, I think, with this stuff where if you've got to give it that much level of detail, you know, then you might as well have just gone and ordered it yourself.

So if you're ordering a pizza and you're like, well, actually, okay, but I need to make sure it's the right sort of pizza, the right size with this topping, and then I want Diet Coke with that, and I want to make sure it's got the fries.

By the time you spend instructing, you can't just simply say to think, give me a pizza.

Because unless you order the exact same order every single time you order a pizza, you're going to basically spend as much time instructing the agent as to what you actually need.

So you just go and click, click, click, click, click, order.

So I do wonder about its usefulness at this point.

Don't get me wrong, I think agents are absolutely huge and will change the way AI works and functions in the future.

But I think, as with a lot of things, when they sort of first arrive, we haven't quite worked out what the killer app is yet or the use case.

So what's the difference?

Let's use your pizza scenario.

Yeah, what's the difference between that and saying to your pa, can you order me a pizza?

Yeah, you're going to have to say, can I have a large pepperoni pizza and a Diet Coke?

It's the same.

It's the same as talking to a human, really, because otherwise, if you leave it up to her, you may get a mushroom pizza or like, whatever, like, I don't know, some veggie pizza with, I don't know, like it.

It this.

My point broadly is, with a lot of stuff, it's like talking to an AI is like talking to a human.

And I think we forget that.

We just assume that it's going to know everything automatically, all the time, and it's Just like talking to another person, if you want something specific, you have to tell the person what you want, otherwise you might get unreliable results.

I wouldn't be telling the person because I wouldn't ask someone to go and order it for me, I would order it.

So the point is that if I've got to go and tell a person or an AI that's I might as just done it myself, you know, and I would have done it myself.

So it's kind of.

So it, it says useless as the human then at that.

But the whole point of the process is that why bother doing it in the first place when you.

And yes, the other thing that strikes me with this, this sort of implementation of agency on whether we know within the browsers at the moment is that it feels completely aligned with humans rather than technology.

So to my mind, if you were saying to an agent, right, I want you to order me a new MagSafe battery for my phone.

It shouldn't need to go and click around like and pretend it was a human on the Amazon website and click about and go and find it and order it.

And it should just be to send a message to Amazon and say, yeah, this guy needs a blank, safe battery done process.

So yeah, it's almost like why are we using this kind of human framework for digital systems?

It's like putting something that's analog right in the middle of something that's digital and saying, ah, but you could just talk to each other directly, but instead we're going to make you use the human interface and GUI and click about that to get there, which is like.

The wrong way around.

I quite enjoyed watching Atlas go off and try do things.

I mean, I've only used it again in the last week or so, but I kept coming up against walls anyway, you know, I, I tried it to, you know, I was looking at some topics around this week and I found some YouTube videos that were like an hour long and I didn't particularly want to watch them.

So I thought I'll try Atlas, get it to see if it can send this video to a transcription service and summarize it for me.

But it just couldn't do any of the processes.

And I saw it trying to work out and it couldn't.

And so then I had to go manually, take the URL, go find a website to transcribe it, download the text, then load it back into chat, GTP and somewhere do you know.

And it was just so clunky.

For what?

It's completely bonkers.

Yeah, sorry.

No, just to say that it's just that stepped process is what you want an agent for that or, you know, we've talked before about how AI totally transforms what a business is.

It's not a bolt on, it's that, you know, processing and automating tasks that are repetitive is what you, what I want it for.

I want it to take away the boring and to take away the stuff that isn't creative.

I want it to take away, you know, summarizing a document that I don't want to read or summarizing a video that I don't listen to for an hour because I know 90 of it is someone banging on about themselves.

That's what I want AI for and to sort of automate that process.

Not for me to have to create a, you know, a Python app myself in chat GDP and then bring it into another, into Claude and create, you know, I've done that because I'm interested in the process.

But it'd be like when these agents get to a point when they can start to, you know, work with each other and do take some of that repetitive and automated work off you, then that's, that's when I think they'll fly.

So.

So I have a friend who runs like a heating.

I think he does gas and he does bathrooms and all sorts of stuff, right?

And he uses.

So he's got the next, not the 20 pound a month chatgpt, but the 200 pound a month chat GPT.

So he's got like the next layer up, okay.

And he's built an agent and essentially what he does is he's got teams of people out doing work all over the place and you know, different job sites and different jobs and all sorts of things and everything runs through that.

So basically when he goes to a job, he finishes the job and then he gets in his truck and then he leaves a voice note for the agent.

And it has a personality.

So at that level you can basically, you can get a voice and it basically is like a person and you can set the setting so you can say, is it male or female?

Is it slightly flirty?

Is it more businesslike?

And all these.

You can give it characteristics and then it basically keeps track of everything.

And so he set it up.

It's female, slightly flirty because he says it's nicer to get through my day if it's slightly flirty.

And.

But, but he comes in and he leaves his.

But he leaves his notes about every job to the.

Just, he just talks to it and says, here are all my notes from the job.

It then goes off automatically and writes all that up and puts it onto his Google Drive and puts it into the form that it needs to go to, adds it into a CRM and keeps track of all of that.

And then he can get in at say 2 o' clock in the afternoon and he can go, hey, I don't remember what he calls it.

Like, I'll just say, eva, it's not that, but whatever.

And you can say, hey, Eva, how's it going today?

And she'll go, well, John's finished two jobs this morning.

Robert's done three jobs this morning.

He has some trouble with this one thing, but he was able to resolve it.

And it gives like a, and it talks back to him and it gives him a summary of what everybody's been up to.

And then he'll say, is there anything that I need to like pay attention to?

And it'll come back and go, well, it seems like John's been having trouble communicating with people and blah, blah, blah.

So you might want to have like.

It actually like gives advice.

And it's what you were saying a little bit, Ben, is it's summarizing what's happened without him having to go and then manually log in to look at all the reports that everybody's written up.

They don't have to spend time sitting and actually filling in forms and writing everything down.

They just talk to it.

It does all of that.

So then when he comes back later, he has all the notes and everything are completely done and ready to go.

They're all in his Google Drive under each customer where they're supposed to be.

And I was just like, okay, so this is a guy who is a small business owner.

You know, he runs four or five different companies and he keeps all of it straight and he keeps track of everything by putting it all into chat, GPT and having the AI and the agents that he's built doing all of these independent little tasks.

And he runs his, basically he runs his entire business.

And I said, is it worth the £200amonth?

He said, I'll tell you what, if it was £10,000amonth, I would still pay it.

Okay, that's interesting.

I mean, so essentially what he's running there is a rag, isn't it?

You know, it's, it's just kind of retrieval when it's generation, it's pulling information through.

Yeah, so a little bit different from saying, go and buy me, you know, some Vimto because I've got a cough And I'm running low and it.

Going out into the world and finding it.

So he's dumping that his team are dumping stuff in.

It's finding stuff, it's allowing him to kind of pass it and then he can get feedback through it.

But I mean, I think.

And that's, that's really good use case, I think good on him.

I wouldn't have considered ever spending £200 on the, on the, the.

The extra powerful version.

But yeah, there you go.

But that's what.

But, but I think this is the beginning.

So it's not a.

It's not a big step to then say to.

And maybe he's doing this, I don't know.

This is.

I've had a couple of, you know, sort of offhand conversations with them at the pub kind of thing.

But it could, there could be an.

There could be an aspect to it where it's then going off and actually proactively doing things if it sees something occur.

So there could be a more agentic kind of side to it where it actually is analyzing what's going on and then it's basically reacting to that and saying, okay, well, I'm just going to go take care of it.

But.

But this is how it starts.

Do you know what I mean?

And it's like.

I think it's people like him who are going to really test it and really find out how can we use these tools?

Because otherwise he would have no idea what was really going on.

He'd be on his phone constantly.

He'd always be trying to figure out what's going on.

Whereas now everything just sort of happens smoothly in the background.

It might be worth saying to him though, that he could probably do all of that in motion for 20 quid a month, just.

But he couldn't.

He couldn't though.

He couldn't because then everybody has to sit down.

They've got to manually type in all the notes all the time.

They've mainly got to do that because, yeah, okay, you can do that, but you've got to have the AI from notion if you want to talk to it.

So you're still back to using AI again?

Oh, yeah.

I'm saying it doesn't do all that.

You can do that.

Yeah, you can just talk into it and, you know, dump your information.

But it's.

Yeah, it's cheaper than open AI.

£200amonth, but yeah.

So do we think then, sort of futuristic wise that we end up in a world with, you know, websites, that there's two versions basically there's one for the humans, you know, so if we want to go and look at Amazon, we've got a nice sort of GUI we can click around in because that's how we work.

And then there's a version of Amazon's website that exists that AIs can talk to basically that is just functional.

It's not.

They don't need to click about, they can just talk to it directly and say I need a MagSafe done.

You know, do we think that's because that will require these companies to open themselves up to the AI companies and allow that kind of access and create that pathway.

And there's a risk if they do that, that everyone then if they say it works really seamlessly and actually can work really well, then people stop clicking about in their websites and then you lose that kind of browsing upsell opportunity or marketing opportunity.

If you're, if you're Uber, you don't want somebody probably just going to look at their phone and just say, hey Siri, or me, order me an Uber.

You probably want them to come to your platform because there's clicks as engagement, there's ads as other things as potential.

So.

Or if, you know, if you're selling pizzas or whatever it might be, you know, again, there's an upsell opportunity that you're going to miss if people don't come to your platform.

So.

And also you don't necessarily get that you user data about that customer either.

So I wonder if there will be some friction and resistance from, you know, retailers or whatever to actually say, well, actually no, OpenAI or whoever.

We're not that keen on allowing your agents to do this because it's stopping our customers from interacting directly with us.

It's creating a kind of a bypass, if you like.

I think even more than that, I think a lot of websites will just not be used anymore.

I think.

You know, I hardly go to webs.

I used to read blogs for information.

I just, I, I can't, I don't think I've done that once this year.

Yeah, I go to YouTube from my information or Twitter or whatever social media.

I haven't been to like to read a blog about something I'm interested in for, for so long or get, get it summarized by chat gdp.

I just think there's a whole level of websites that just probably are going to cease to be useful.

I think another direction will be, you know, you'll want your agents to just go off and give you recommendations, you know, so I've spent probably A couple days this week buying props and things for a shoot next week.

I'd love, you know, an agent in Chat GTP to be able to go off and out to Amazon, research the best five of these or three of these in the price point, look at reviews for me, you know, there's obviously a trust element to that, but to then come back and say, you know, here's your options.

Do you want me to go ahead with one of these and, you know, to be able to speak these things to the.

To Chat GTP as well and not have to sit at a computer and manually, you know, that's where I would like to see it go.

Yeah, I think it's going that way.

You hit on something that I was going to bring up, which is there's different type of shoppers as well.

Right.

Like, I'm a maximizer, so I like to shop around.

I will look like I need to buy an external drive.

I will look at or a NAS system.

I will look and look and look and look and I'll read every review and I'll waver back and forth and I'll go, what about this and what about that?

Like, a nightmare for me would be what you were saying.

Alan is just saying, go buy me a charger.

Yeah, right.

Like, that literally, like, fires every single, like, nerve ending I have, because whatever I get, it's going to be like that.

Whereas my wife, she doesn't give a shit.

Like, she just like, I, I joke that I shop like a woman and she shops like a man.

And I know that's probably.

People kill me in the comments, but I don't care.

But she will, like, walk into the shop, she'll find the first thing that ticks a box and she's like, yeah, I need, I need a charger.

She'll walk in, she'll go, that Charger, that color, I'll have it.

I wish.

I'm like, do you even know how I know?

I just wish.

I'm like, do you know how that works?

Do you know anything?

Like, I don't care.

It's like a, a charger, I'm fine, I'm just going to get it.

And I'm like, yeah, two weeks of.

Research is needed on this charger.

Exactly.

So for I, I think, Alan, for some people, I think the AI thing would be really, really good.

And for other people, it just wouldn't work at all.

Yeah, okay.

If you're shopping for a company and you need to buy toilet paper or whatever, then, yeah, you could probably just Set a thing off and just go.

But even then you're like.

Because it will go and it will go away and it'll come back with the least expensive, cheapest toilet paper.

You go, I hate that stuff.

So I don't know.

I totally get what you're saying and I do think at some point, yeah, there needs to be a back interface.

But the other thing I don't trust about that in the API interfaces and stuff is if you go off and say, give me a charger, what Amazon's going to do is they're going to sell you the charger that makes them the most money, not the best charger.

And so their AI is going to be maximized for their profit, not your satisfaction.

Agreed.

And so there's going to.

This is where I think the trust breaks down, is that you have to really trust that the other AI is going to give you something that's something that you want.

I could not agree.

I don't know how we get there.

Yeah, I think that's a really hard problem to crack.

I think it's, you know, we've seen this already to some degree years ago.

Look, you know, when the Internet first came along, everyone said, well, that's the end of, you know, people shopping.

But it wasn't.

People still, still go to the supermarket and like, pick up the.

It is look at it and touch it, but it is choose their clothes.

Look at our high streets.

Our high streets are dead.

The malls are gone.

It has killed retail compared to what it was in the 80s.

It killed it.

It's absolutely killed it.

I think there's plenty.

Look, the high streets where I live are full of shops and retail outlets and people go.

And I walk around Morrison's at the weekend and it's busy, it's full of people.

And you walk into the next and all these clothing shops and they're full of people trying on stuff.

And what I'm saying is it didn't all just go online, you know, it just doesn't.

It wasn't like, oh, no one, no one shops anymore.

You know, people enjoy the experience, they enjoy the transaction of it.

They enjoy the physical, you know.

So I'm not convinced that, you know, what you were saying, Dave, is that everyone's going to want to just kind of let the agents do it for them.

I think that a lot of people, including myself, will prefer to go in like you review, check, understand the products, you know, before you click buy.

And the example that they always seem to give the tech companies is, you know, about booking your holidays or something.

There's no way I'm letting an agent plan to book my holiday.

You turn up at the airport and there's no real ticket or you get to the hotel, it doesn't exist or do you know what I mean?

It's just like, yeah, yeah, oh my.

Gosh, yeah, you know, yeah, we're all to do that.

We're also particular when it comes to those, what, you know, those big impact, big ticket items.

You know, you just, yeah, just never going to trust someone to know your idiosyncrasies and quirkiness over those kind of big things.

I don't think.

Yeah, maybe in the future.

Yeah.

And I think we, we've talked about this on, on earlier episodes, but this is where I think that you're going to end up with a federated model.

So what you're going to have is a bunch of specialist AIs that do travel.

You'll have a specialist AI that only does travel bookings and maybe only hotel bookings.

And you'll have another one that does airline bookings.

You'll have another one that does, I don't know, cruise bookings or whatever.

And what you will talk to is a high level conversational AI that is the federated part of it.

And then all it does is it then goes away to other AI tools itself and uses those tools to do the research and to execute those specific tasks as opposed to one AI that does everything.

And I think that's where we're ultimately going to get.

And then that way each one of those specific tools is only trained on data from that specific knowledge base and you'll get better results from that.

And then that's where your high level AI can take stuff.

It'll go, okay, here's the airline results, here's the hotel results, here's the, you know, I don't know, here's the, you know, sort of activity, result set and everything else.

And then it can start to weave them together and go, okay, well this matches that and that matches that.

Yeah, but it's a different, I think ultimately that's what we're going to end up with.

And then what you'll have is you'll have Those top level AIs will have different personalities.

So you may have, they may be regional, they may be language based, they may be culturally based or whatever.

And, and, but the core systems that sit underneath them that do those tasks will pretty much be the same.

I still think trust comes into that.

Like, I still think you might think the agent and the back end are not going to give you the best value or this.

Do you.

I mean, there's.

Yeah, there's still going to be that same element that we talk about with going to Amazon, like, is it going to give me whoever's bought the best reviews?

Or, you know, and there's that constant fear with TripAdvisor or Amazon or how much can we trust that data?

How much can we trust that airline?

And.

And I don't think we have a lot of trust with the people that are providing us these services.

You don't know.

And I think.

I think we do end up, you know, the version that we have in 20 years from now, as this evolves, will be nothing like what we've got today.

Today, I don't think.

Yeah, what we have today will probably look like the equivalent of what ceefax or Teletext was to the Internet.

You know, seem kind of readable and laughable and it would be kind of.

Isn't it bizarre that we tried to do it like that?

That's just.

That was mad.

Yeah, I. I just don't buy it as a way forward.

You know, a computer mimicking a human by clicking around a website just seems bonkers, basically.

Yeah.

So, well, speaking of the future, what about education?

And so this has been in the news a little bit.

I saw an article just yesterday, actually, on the BBC website saying that AI is going to be on the curriculum.

So their take was this, that they were going to teach students to understand and recognize AI, know what it is, how to use it, and also be able to identify when AI is being used.

So they can spot fake, you know, news, fake imagery, that kind of stuff.

So kind of a.

A general roundup.

I don't think it sounded like they were going to go deep, like in China, where they're teaching kids to kind of, you know, get super programmy with AI at a very early age.

This sounded more like a kind of general, you know, overview of.

Of AI and awareness exercise than anything else.

But.

Yeah, what do we think about AI and education and where will we be in 20 years from now, do you think?

I have thoughts, yeah.

I was going to say, if you look at Estonia is an interesting.

I think it was Estonia, because they are outperforming almost everybody in Europe at the moment and they've gone very pro digital, very AI, AI phones in classrooms, obviously, in a controlled way, and there's rules around it, but they've really embraced it and it's shown quite a big.

Yeah, quite, quite A big improvement, I think, like for the horse has bolted for a lot of.

Okay.

I mean, my.

I don't know about you guys, but my son is all over AI.

He's teaching himself Python.

He's sort of learning about large language.

You know, he's 12.

He's learning about this stuff long before the teachers can catch up.

You know, he's, he, he's already living, breathing AI every day.

You know, he uses it for game development.

You know, he doesn't use it so much for learning maths or English.

But yeah, I think, I think it's really, it's here to stay.

I think it's really important that it's included in the curriculum.

Yeah, for sure.

Dave, go on.

I was going to say, do you think in the classroom it becomes a co teacher, you know, for the teachers themselves, they'll be using AI.

So, so 20 years from now, kids will be sat in front of screens.

There'll be teachers, but there'll also be AI teachers as well, working through, you know, the curriculum with them and covering the different topics and giving them more personalized, I suppose, in a way that a teacher, if he's got 30 students in the classroom, it's difficult, one or two teachers to be up close and personal all the time, that you could have an AI tool available to each student that could understand where they're at in terms of their development in progress and then be setting them appropriate, you know, sort of tests and learning activities.

And then the human teachers kind of, you know, moving around almost oversighting, making sure that, you know, this is, this is kind of, you know, happening properly and the kids are happy and settled and everything's as it should be.

Yeah, maybe I just had a thought while, while you were saying that, that I hadn't really thought about before, but.

And, and this, this is related to education.

But I wonder if, I don't know if you've seen the, the sort of post where they're Talking about that ChatGPT no longer gives medical advice or legal advice.

And so if you ask it to give you advice on something, it basically says you need to speak to a lawyer and it won't tell you.

I wonder if it's going to start doing that if it knows, like if the kids go into school and they sign into ChatGPT, there'll be an education version that won't actually give you an answer, but it will help score you on your work or it will give you some information about, like, I don't know how it would work, but they can restrict how the AI behaves and the information that it will give you.

So I wonder if that would be the precursor to a real take up in education because then it's like maybe each school could tell it, okay, these are the questions that we ask.

So you can't answer these questions, but you can help them with information related to it or something.

So I think Gemini almost does this, doesn't it?

If you go into Gemini, there's like a learning mode you can turn on and then if you ask it a question, rather than it just giving you the answer, it almost takes you through some learning steps to get to the answer.

So you go for a process to get that rather than this is the answer.

That's interesting.

Okay, that's a really good idea, David.

I think it's almost, you know, what privacy is to large corporations and they're having to go for enterprise models.

Is that that kind of not giving you the answer straight away and teaching you to think for yourself and more the process, one on one teaching, which is the, you know, the goal, I.

Think, for scoring your answers.

But the way the students were using it, it wasn't to get the answers, it was to grade their answers.

And so they were doing mock exams and they could go and answer the questions and then they would ask ChatGPT to grade the answer and to give them strategies for writing a better answer.

And the teachers were rightly very concerned about that in the beginning until they said, look, go and do it and then bring us in what it says so we can see to make sure that it's correct.

And when I was sitting in the staff room talking to the teachers in private, they said that most of the time it came up with stuff that they had never even considered and would have never noticed.

They're like, the advice that it was giving was way better than the advice that they could give any individual student on that.

Private schools were doing that so that their kids could score better on the exams and to understand how to answer the questions better.

I think that's a fantastic use of using a GPT, particularly a language model, because that's exactly what it's based around is the language of how you write your answer.

You know, there's another, there's another story I want to talk about in a second, but jump in, Alan.

I Think you wanted to say something.

Well, I was just going to say, I mean it's interesting, the private school thing as well, because maybe, you know, in private schools obviously already, you know, there's a two speed economy, isn't there, when it comes to education with private schools and you know, public schools.

And.

I wonder if this widens that gap.

You know, private schools have much more advanced AI adoption because they can afford the technology earlier, they can implement it.

They're that they don't have the same restrictions of, you know, the government that suddenly that gap even grows even bigger because now you're, you're almost supercharging the education system in those environments, in the public schools which are tied into, you know, government policy or whatever are slightly hamstrung and only be able to deliver a kind of, you know, a poorer version of it basically.

And I wonder if that's one of, that becomes a sort of long term problem as well with these systems.

I mean, one thing I'm have a lot of hope for the AI systems is that it can create a democratization.

I mean, just stepping outside of the UK for a minute, you know, if you think about the other sort of 200 countries around the world, but a lot of those countries are not in a great state, but there is an opportunity, you know, with a very simple device for somebody who doesn't have access to a lot of good information to suddenly be able to talk to an AI system fairly cheaply, you know, and get information and learn, you know, so maybe, maybe globally there is also a democratizing effect to the technology.

Yeah, if you can get the, you know, AI to those edges off, you know, offline.

So you've got a large language model on a, you know, I don't know, on a Raspberry PI or a small computer in a village somewhere that doesn't have connection, you know, or you know, you think of, you know, Afghanistan where girls aren't allowed to go to school anymore and you know, that's a whole different discussion.

But you know, being able to access OneToOne teaching with AI, you know, there's, there's opportunities there to go across cultural boundaries and economic boundaries to sort of aid kids that can't have access to some of this stuff.

Yeah, we get Starlink, beam it in anywhere, you know.

Yeah, for sure.

So I signed, I saw an article the other day and I'll, I'll, by the time this comes out, I'll find it and we can put it in the show notes so people could see it.

I was desperately looking for it earlier and couldn't find it.

But it essentially said that some of the big consultancies like McKinsey and PW, PwC and those guys, what they were finding is, is that they were getting graduates, recent graduates who were coming through as new hires and they were having to get rid of.

I want to say it was like a stupid number.

Like 90% of them weren't fit to work because what had happened is that they'd been using AI to help them through their education.

But what they didn't have is they didn't have the ability like for encoding.

They didn't have the actual knowledge to sit down and actually code something themselves anymore.

That they so relied on the AI to help them when they were writing code that they couldn't actually write the code themselves.

So when you got them in a real world work experience situation where they couldn't use AI, they couldn't work and they couldn't do it.

And so that, you know, they were lamenting the fact that basically most people coming out of school now aren't fit for those, for certain types of jobs because they're so reliant on AI that they can't actually do it anymore.

Which I thought was.

It was a really interesting article.

I wonder if part of that as well is that because the companies themselves are still a bit legacy.

So they've not caught up with the AI world yet.

So they're landing in a company, so they've spent their time, you know, being, you know, very sort of involved with AI.

Then they landed a company that's still a bit legacy, hasn't quite sorted itself out in terms of its AI capability yet and doing, still doing things the old way.

So there's a mismatch, isn't there, in terms of their skill sets.

Maybe if they landed in a company that was very AI native and already progressive with that, they'd be okay, but they sort of landed somewhere that doesn't do it.

I used to see this in engineering a little bit, actually.

You'd get students coming out of, you know, with M engines or whatever, and they'd end up in a company that did a lot of sort of what we call traditional engineering.

And they'd struggle and those companies would have to literally say, right, we just need to go and put these students on the shop floor for six months and let them learn what the back end of a lathe looks like and actually understand, you know, some real, you know, heavy engineering.

So I wonder if it's, it's that as well, and then also the other thing that occurred to me when you were talking, Dave, is that I imagine a lot of these applicants are probably making their CVs, you know, a bit better with AI, so again, perhaps overselling themselves, you know, so they, on paper they look fantastic, but the reality is, is somewhat different.

Well, you know what they say, there's small lies, there's big lies and there's CVS.

And there's AI.

Yeah, my wife, my wife's an actress and she was saying the amount of people that have got on their, you know, their details that they can ride horses but aren't very good.

And quickly before a shoot, all the horse riding courses get really booked up because it was like, oh crap.

I said I can do this, I can't actually do it, I can't do it.

Yeah, but the.

Yeah, and that's really interesting, I think, and I've often said this, I've said this to you before, Alan as well, that, that I, I worry that the difficult phases of becoming, going from a junior in say something like law or something that's very knowledge based to an expert takes a lot of time.

That can take 10, 15, 20 years.

I think when AI can start to do the job of what those juniors do, that's where we get to quite a precarious situation where how do those people go to in engineering or law or architecture, how do they go from that junior position to an expert position when for those companies it's going to be much cheaper for them to have AI doing those junior roles.

And I know we're not there with some of those, but we're not far off being there and I think we'll find 10, 15 years time down the line that there won't be those people with that depth and that knowledge base and also that world experience of using subtlety and being, being able to cope with pressure and all those things that come from failing over and over again as, as a junior, that, that's my worry.

With AI, you want to have experienced judges, you know, they need to have done a bit of law judge.

Yeah, yeah, yeah.

And get it wrong in smaller cases, things that don't matter as much as the bigger cases, you know, and there's, there's hundreds and hundreds of situations where you need to go through that process to get better, to get good, to get gifted at something and, and I worry that we're going to, as company is going to want to save money through using AI instead of juniors or administrators, that, that will lose, lose something along the way.

And, and it is happening, it is, it is happening on the ground quicker.

Than we thought as well.

Quicker than we thought I had.

She came on and she said at the time, she's like, there's no way AI can do what we do.

You know, this is a very personal, driven, you know, it has to be human.

You have to have the connection you have.

You know, what we do is a very particular human skill type thing.

I had her in the studio a few weeks ago.

We were talking about stuff and it's like, oh, how smart, you know, how's business going?

How's everything going?

She's like, oh, we just let 14 people go because basically we, we couldn't, we couldn't keep them on because it was too much trouble to try and train them when all of our people could just use AI and get the same work in less time.

Yeah, 14 people got basically let go because they weren't as efficient as AI and it took more time to train them than it did for the more senior people to use the AI to just get the work done.

Yeah.

And I just looked at her and she went, don't, just don't.

And I'm like, I mean, this is what's happening.

And again, this is across the country, across the world.

Yeah.

And this is in thousands and thousands and thousands of small to medium sized companies, which is where it's hitting the hardest.

Right.

Because they're the ones that are pushed the most on sort of cash flow and everything else.

Like you get a McKinsey or someone like that.

Like they've got billions coming in, they've got massive government contracts, they can kind of afford to keep the juniors on if they want to, and they're still making money and their shareholders are happy.

You get an SME that has, I don't know, 20, 30, 40 people and they're still pressing to kind of make their salary every month and whatever and they really have to watch cash flow quite hard.

Then you get in a situation where, well, actually those salaries matter and if they don't have to have those people, they're not going to have them because it makes it that just that little bit easier, you know, to keep the company going.

And those are the vast majority of jobs.

Yeah, agree.

Or at that.

At that level.

Yeah.

And, and that's where we're really starting to see a problem.

And I think that's where we're really going to see a problem moving forward more, more and more.

And exactly like you said, if, if we're not hiring them now, what we're doing is creating a real problem for ourselves 10 years from now.

Yeah, because 10 years from now we're not going to have the senior people that we would have trained normally, but AI will probably be able to do the senior role.

Well, since we're talking about replacing people, then why don't we move to an even happier note of.

Did you see the X1 robots all over social media in the last week?

I. Yeah, I saw something about it, yeah.

I mean everyone must have seen.

It's probably worth explaining for the listener.

So this was a robot that.

It's hard to explain really.

It was about 5 foot 6, I think, probably weighs about a 100 pounds.

It's robotic, obviously.

It's covered in a kind of jumper, effectively.

It's just a kind of.

It was like a kind of woolly sort of outfit, sort of strange looking face.

Just a couple of cameras for eyes.

No, no, mouth.

It's got, you know, ears and microphones and it weirdly has a speaker in its groin and it's in its belly.

Put your head down to listen to it.

Yeah, that's kind of odd.

And they kind of sold it with their corporate video that, you know, this thing can kind of wander about your house, you know, do, do, do the washing, you know, fold your laundry, take out the rubbish, you know, bring your shopping in, what, whatever, you know, sort of mundane house chores basically.

And they had this quite sort of slick video.

And so you showed this robot kind of thing, stuff like this.

And then of course, you know, as, as the week went on, the journalists like Joanna Stern and Marcus Brownlee sort of dug into this a bit more.

In fact, Joanna Stern went and saw the robot and spent maybe like, you know, a day with it and sort of quickly worked out.

Now, in fact, the robot wasn't really doing any of this.

It was more like an animatronics, you know, from Disney or something, you know, and effectively there's a guy with a headset, a meta headset.

It looked like controlling this thing remotely from another room, you know, so.

So here we are.

So, you know, obviously their, their intention is one day it will be able to do all this stuff in autonomously.

But what was clear was that even with a human controlling it, it could barely do any of this stuff.

You know, it tried to open dishwasher.

This is the human control.

Yeah.

Pulled it into itself.

Couldn't.

When it went down to try and get dishwasher, it practically fell over and it was just fumbling away, you know.

So I don't think we're in fear of this thing replacing any of us anytime soon.

But just in general, what do you think about the whole idea of having a robot wandering about your house doing this, being able to see you and all of that and everything that sort of comes with it?

I, I think we're still decades away from that happening.

I know a guy who runs a robotics company and basically his thing is it's going to be 20 years, 25 years at least before we're in any position for that to happen.

Because simply that they're too unreliable.

Just expensive.

And they.

Well, it's expensive, but they're so unreliable and the parts are so fidgety and fragile that as soon as the smallest thing happens, then it breaks and then the whole thing doesn't work and it's expensive to fix.

I know I saw a video about BMW.

Like BMW has some robots in their factories and I think it said one of them's been working a 10 hour shift every day for five months in a row and it hasn't had any problems.

And it actually shows it like moving around and it grabs something and it takes it over.

So it's doing what a human would do in a very like a 12 square meter area.

But that's it.

It only operates in that one space, but it is actually doing it.

And it has been working in a factory environment reliably for that long.

But that's a, that's doing one repetitive task over and over and over again.

That's not kind of walking around.

So for me I'm much more concerned about, I don't know, quantum, because I think we'll actually get quantum computers that are functioning and useful before we'll actually get robots that are walking around our house.

Well, and I think part of the problem with the robot as well, I mean this one in particular was that it wasn't waterproof either.

So, so half the task you might want it to do, like the dishes for example, you know, would involve water.

It couldn't go outside, particularly in case it was raining, you know, and then imagine if you had it cleaning the toilet and it's going to have the same pair of woolly hands.

It's then going to go and pick up your food, your cutlery.

It wasn't allowed to pick up sharp objects, so that rules out any cutlery being put back and put away.

It wasn't allowed to lift any heavy objects.

It wasn't allowed to go near anything hot in case it catches fire or decides to throw it out, you know, and then, you know, the idea that you have a company that if it wasn't worse at the moment there's an actual human in your house looking at you, presumably.

Yeah.

But at some point, a company at least anyway, scanning your house and having a look around.

I mean, it feels like, I think, and I agree with you, Dave, 100 for everything I've heard about robotics, is that this is still a very long way out and just something that might seem simple to us, like unloading a dishwasher is increasing.

Incredibly complicated for an AI system and a robot to be able to do that.

Incredibly complicated.

But like the robot dog is a different thing.

I've just seen one of those live actually a few weeks ago.

Yeah.

And they, they are cool and they're creepy as well.

Freaky and cool.

Yeah.

But they're using like the army's already using them in combat and stuff to carry heavy loads.

And because it self balances and whatever, you can like load it up with ammo for your unit or whatever and send it out.

So it, it can take a lot of stuff with it as basically like a pack mule.

And so I can see where something like that, that they've kind of got out in the field and they're using it, that can be useful.

But again, that's not like a humanoid robot.

That's, that's more like, it's, it's like a horse.

Right.

So you're kind of making a horse just to carry heavy loads and that makes sense and, or bomb disposal or.

Yeah, all those.

Yeah, but it's, I think it's so far away realistically that it's, I mean, it's fun to think about, I guess.

We've, you know, I think most of us have got some aging relatives.

What about a companion robot?

So it doesn't need to do much, doesn't need to go and do the dishes.

It's not, it's not a slave.

And I do actually, I am actually slightly concerned that the human's first response when we think we've got some kind of robot is to turn it into a slave for us.

Maybe said something about humanity, but, but putting that aside, it's just a companion.

So it can sit on the sofa with you, it can chat to you.

It looks a bit human, it looks friendly, it's got chat GPT inside of it so it can have a decent conversation.

It can sit there and talk about whatever.

If it was the case of my mother.

Talk about the war endlessly and you know, just kind of.

Yeah.

Be.

Be a social being.

That's always.

I just had a terrible thought.

I just had a terrible thought.

You need to share.

It's not what you think.

No.

Or the.

It's like is that what we're going to end up to do the euthanization with is we're going to have the robots do it then it's no one's fault.

Well, the pods are machines, aren't they?

They're sort of.

They're robots in a way.

A crew that single purpose robot.

But no, it's just.

Yeah.

You know you could have a few of these wandering around care homes maybe and just going in and talking to.

People who are taking people out and you know they're just gonna mouth.

They're just gonna malfunction one day and just take everyone out.

I mean it's.

I get.

I guess this X1 suffers slightly from the same thing that you know the, the Elon Ross Elon Musk robot show had.

You know it felt so like it was all controlled by humans and none of it was actually.

And I think it's that over promising.

Everyone's so desperate to do stuff and to get it out there that they're you know, we're always over.

We had the same problem with Chachi GP5 when that wasn't that you know, it.

These things are amazing steps but we're marketing them as beyond what they can do and I think that's one of the big problems.

I mean even when we did the car one there was a lot of the cars were driven by remote operators to start with.

Yeah.

And less so now because they've solved some of that and I guess they will solve that with this robot but it doesn't feel like it's going to be ready for when they launch.

And the thing is, you know the car thing is actually is a much easier problem to solve that because it's.

It's operating within a much more defined framework.

Normally the self driving cars we have at the moment only work in very specific environments anyway and.

And they're heavily loaded with radar lidar and all sorts of stuff.

You know like the waymos and things and then the sort of Elon Musk self driving.

So say self driving, you know, only work in very particular environments like motorways and things like that.

You know he'd jump, jump on a Cornish road and try and let a Tesla drive you to, you know St. Ives, I think you're going to have a lot of problems.

But with the house, there is so much complexity.

Even if you just take something like folding clothes, you know, every clothing item, the different shape and how does it fold and how should you fold it and what should it look?

You know, the, the level of complexity is off the scale, basically.

And, and I think it's going to take an awful lot of uplift in intelligence and also robotics to get as close to that.

The single use robots seem to be really good.

You know, the, the Hoover on the floor going around and it's got, it's got its problems.

It can't go over a le nip of more than an inch and it cut.

But we, we have one.

It's great.

You know, it's a toaster, isn't it?

It does one thing and that's what it does.

Pool cleaners or those sort of single.

Can you imagine getting a robot to clean your pool?

It's just.

I'm sure they do.

Yeah, I'm sure they do now.

Yeah, yeah, yeah.

But, yeah, it's.

I agree.

I think it's, it's unnerving a little bit.

I just, I can't really put my finger on why.

I think probably because it was so crap.

So why are you going to even show off the dishwasher scene?

It was so awful.

Well, there's the, there's the guys down.

Multi, multi, multi millionaire.

Right.

I would get one just, just for the giggle.

I think what'll end up happening is, is that eventually, let's just.

If we look far, far into the future, I think we'll go through a phase where, like you said, if you have tons of money, you'll buy a robot because they're fun.

And then what will eventually happen is that only the rich people will use humans and everybody else will be stuck using robots because robots will be the cheapest, most cost effective, most reliable way to deliver any type of service or anything.

So if you don't have any money, you're going to end up dealing with robots and AI all the time.

And if you have loads of money, then you'll get to deal with humans.

Yeah, human is a luxury.

Yeah, yeah.

You have a chef and the.

Yeah, someone's exactly a nanny.

Otherwise.

Yeah, you're getting food from a machine.

Yeah, yeah.

And the rest.

And on that positive note, Uber eats.

Robot.

Just locked up on a bike.

There we are, guys.

We're on the time, I think, so we're just over the hour.

So literally we've Talked ourselves into the end, haven't we?

The end.

The end of civilization as we know it.

So make lots of money so you can keep dealing with humans.

That's the message.

I think, Alan.

I think, Alan, really interesting topic for another sort of slightly deeper conversation is around sort of robots and AI for companionship.

Yeah, that's a really interesting thing to look into.

I haven't looked into that at all.

And obviously there's the, the darker underbelly of that, but there's the sort of more old people, you know, losing memory, all those kind of children at night being read a story to.

I mean, there's some creepy stuff around it, but it'd be interesting to explore that at some point.

Well, I've got a cunning plan actually, just very quickly.

And maybe this is something we can explore, you know, in a future podcast, because I might have done it by then, but I'm going to set my mum up with chat GPT so she, you know, if she's bored, she's sitting on her own, she can chat to it.

And I'm basically going to put in the, in the personalization that every time she fires it up that it will ask her about, you know, her life and what she's done and things that she's, you know, and she, she'll sit there and tell it all about the war and her childhood and everything else.

And I figure if I leave her doing this for, you know, for a year or something, it will capture so much information about her life and what she did at the end of it.

That will be an unbelievable resource for me, you know, in years to come to go back to and talk to and ask about and.

Yeah, you know, it's a kind of lazy way of sit.

Instead of sitting there with a load of tape recorders and transcribing and recording it for hours and hours and hours.

I just get the AI to basically.

Well, I could do video as well.

You don't even have to be there, Alan.

Exactly.

Exactly.

It would be interesting the extreme you could go to.

Then, of course, I could then take over that AI and at some point and then actually turn it into a kind of, you know, Denise Chatbot.

I could then talk, say what you are now this person.

You could.

That's.

Yeah, we talked about that at one point on the grief tech.

Yeah, I. I think I would like to.

To throw out a potential topic that I think we could dig into in one show as well, which is.

I'm really focused a lot on Quantum right now, and I think everybody's talked about how AI is in a massive bubble, a financial bubble, at the minute, which I totally agree.

And I think that Quantum is going to be the next bubble.

And so I think as soon as we get.

Just because we now have some quantum systems that are working pretty well, and I think we're going to start to see that more and more.

And I think as soon as anybody really cottons onto what's going on there, I think all the money is going to move from AI and it's all going to pile into Quantum.

And then.

But the impact that Quantum might have on AI, I think is also something that we should really think about, because all of the processing and all of the struggles that we have around the processing at the minute, if you put that onto Quantum, all of that goes away.

And then that opens it up to doing thousands and thousands and thousands times, maybe million times more processes in the same amount of time.

And that's going to completely change the playing field.

And then we can talk about fusion, which is coming even after that.

And that's the next bubble.

So you've got the kind of the triangle of transformative technologies and the quantum fusion in AI.

What?

Yeah.

When you put that together, what kind of force multiplier are you going to.

Let's make that the next episode.

You end up with the rabbit.

Yeah.

Spoiler alert.

I'm gonna get my son for Christmas.

Don't tell me we can talk about that another time.

Why would I?

Why on earth would I?

Are those your meta glasses as well?

Are those the new ones?

Yeah, they're not the new ones.

They are the.

The originals.

But.

But they are the original strike, so.

They don't start giving me messages in the middle of the conversation.

That's probably a good idea.

All right.

Are we done?

Thanks, guys.

Been amazing.

Amazing.

To catch up.

Let's.

Let's get another pod recorded soon.

But.

Yeah, until then, see you soon.

Thanks, guys.

Cheers.

Nice.

See you.

Bye.

Sa.

Sam.

Sa.