Sean Bulmer joins host Dave Atkinson to explore the intricacies of artificial intelligence and its impact on technology and society. They discuss the dual nature of intelligent machines—while these technologies can handle mundane tasks, there's a concern that as they become more capable, they may also seek to automate their own existence, leading to unforeseen challenges. The conversation delves into vehicle-to-infrastructure technology, emphasizing the importance of data capture and communication methods in improving transportation systems. Additionally, they touch on the evolving roles in the workforce as AI continues to advance, highlighting the emergence of new job categories like prompt engineering. Ultimately, Sean and Dave advocate for a vision of AI that enhances human life and fosters a more connected world, while also addressing the potential risks of cybersecurity in an increasingly automated future.

Takeaways:

• The evolution of AI will lead to machines doing mundane tasks humans prefer to avoid.
• As AI technology advances, it will create new job roles and require new skills.
• Cybersecurity will face challenges as AI becomes both a tool for attackers and defenders.
• Data management is crucial in the era of AI, especially with increasing data storage needs.
• The societal impact of AI includes addressing food distribution and environmental management challenges.
• Intelligent machines may change how humans interact with technology, leading to improved quality of life.

Are you an intelligent robot? Go do all this for me so I don't have to do it. It's why we built machines, it's why we built computers, is to do the stuff we don't want to do.

But second half of that coin is it won't want to do that. If it's intelligent, it'll do it for a bit and then go, I'm gonna make something to do this for me.

You're listening to WithAI FM.

Hello, everyone, I'm Dave Atkinson. Hi, I'm the host of the Places WithAI podcast.

Today my guest is Sean Bulmer who is an extremely talented engineer and a good friend.

Shaun and I known each other through work, work colleagues worked on transport technology stuff but also worked on things a little bit more rogue from that soapbox Micklegate run car carts and things and things like that.

And we share very similar interests in kind of tech transport but overarching that a kind of a will to create better outcomes for people and work with tech to improve the built in natural environment and everything that connects those things together. Welcome Sean.

Thank you for having me. Yeah, great to be here. Like you say, we've, we have some strange hobbies. I still remember one of the days when you asked me about blockchains.

So I went away and learned about blockchains just for the fun of it.

So the blockchain, just for everyone's benefit. The blockchain was, was.

It was about a theory of a taxi that would ultimately own itself and this kind of blockchain technology that would sit behind that. Maybe we'll get into blockchain later on or, or, or another, another time. I was party to Sean's welding of our micklegate run cart as well.

I, I was the, I was the graphics guy. Sean, Sean was the artillery.

So yeah, I came a bit heavy handed. You were a bit on the light.

It was all good. It all if the we, if the tires, if we'd managed to get better tires, I think we'd have won.

But so let's sort of jump into it then and we're going to dismantle some of the technology stuff which I think is a favorite subject for both of us. And you work in a very practical sense in your job as an engineer. So talking about vehicle to infrastructure type technology, V2X type things.

Can you just expand on some of that Shaun, and just talk to us about data capture, sensor data and things like that.

Yeah, so everything holds itself on data and to capture that we need to use sensors and communication methods between vehicles infrastructure. So there's, there's probably three or four, but the main three, four.

So the main three are probably DSRC, which is dedicated short-range communication between vehicles and infrastructure. So that's basically Bluetooth, shortwave, wireless, things like that.

Then you have cellular vehicle two infrastructure, again using the cellular network, which I believe we're getting some frequencies dedicated for transport. I can't remember if that's still going ahead, but I need to look into that still.

And then there's Keysight, which again uses the DSRC, but it also uses RF transmissions to communicate between items and stuff like that. Yeah, that's the communication method. Sensors-wise, we all have sensors on us all day, every day.

You have a mobile phone, it's got accelerometers, it's got WIFI, it's got light sensors, sound sensors. Yeah, and that's collecting data all day, every day. So it's usable.

How and how do you sort of manage that as it comes through, Sean? Because that sounds like. So basically you've got, within the public realm, then you've got hundreds of different actors.

So we've talked a bit, a little bit about vehicle to vehicle, vehicle to infrastructure type stuff, but obviously, I, as a pedestrian, would have a phone; there would be other things attached to buildings and things like that would potentially, potentially be able to capture or transmit data. So, how do you manage the sort of noise? What's the.

So, yeah, very difficult. Obviously, you build; you start off building algorithms to determine what you need and what data you actually want to keep.

That gets filtered out from the rubbish. But again, it's all about data storage, data capture and how that is done at the capture level.

Your devices may give you thousands of bits of data, which they do every second, but you only want 20 pieces. So, you have to filter that out using an algorithm or an API. REST API. They're the cheap and easy ones.

REST API to pull them out into a database which then you can manage a bit better.

And just a specific example, just pick on in-vehicle messaging. The reason I bring that up is because we had a demonstrator, didn't we?

We did, yeah.

A couple of years ago. I think it's pre Covid actually, wasn't it?

It was a couple years ago.

It was actually probably five, six years ago, but.

Well, that's it. Everything pre Covid is a couple of years ago.

Yeah. And I think from the experience of the driver's perspective, for example, in that case, what, what is it that.

Because, because my sense of it was the in vehicle messaging was centrally controlled, wasn't it? And it was based on geo fencing on, on a kind of an abstract map that represented, well, a map basically.

And as you moved into it, you could create a geofence.

And then in the real world, as you moving along a piece of highway in the, in the digital world, you moved into a GEO fence which is, which is kind of a, like a, a digital fence that you would put around a bit of geography. You enter.

It's an activation area. So yeah, yeah, it's an activation area. So you'd fence off an area and when you get into it, it activates something. It doesn't matter what it is.

Let's take it in a bit. Pokemon Go has geofencing for going to Pokemon stops. You get close to it, it activates. It's the same technology.

What you're talking about is, yeah, there was a centralized operating map and you would circle the area you want it and when you entered that area something would happen.

And that could be in the terms of, for example, the E Scooter trial that could be reducing the speed of the scooter, the in vehicle messaging. Where I was getting to that, I guess a little bit was it started to feel like there was a lot going on in the car. Do you know what I mean?

So it's, I think in the sense of how do you manage a lot of that noises, There's a lot going on in the car anyway. And how, how do you make, how do you think you make those decisions?

So yeah, there's been a lot of studies on driver distraction with in vehicle messages.

You've got your radio, You've got your WhatsApp notifications, you've got your messages, you've got the phone calls all coming through and it's all a distraction and it's, it's hard to determine what you should be showing a driver. Should you just be showing them the legal stuff they need to know or should you be sending in the VIN vehicle messages?

If you turn right, you can go down to the shop and get 40% off here, so 50% off that. Or if you turn left and go right again, you actually. It's quicker than going straight on.

Should we be showing directions, should we be doing this Again, I'm not a psychologist, but those studies are out there and people will make the decisions for us.

And then probably moving on from that, we're capturing, so we're capturing a lot of data, sensor stuff's going on in the public realm, etc. So that's a lot of data then to process and store, isn't it?

Yeah, it can be a lot, but it also can be not a lot as well. So you think of, let's say we're capturing GPS data, it's what, 16 bytes per location, X and Y, blah, blah.

So yeah, so 16 bytes, 32 bytes every two minutes. Every minute. It's not a lot of data. But over 8 billion people, that's a lot of data.

And the increase in storage requirements for all this data is going through the roof, to be honest. Yeah, it's just incredible. I'm. I don't know where it's going to go.

I don't know how we're going to store all this data for if we're just going to use it and then bin it. I'm going to use it and start historically to work back on later. I don't know.

And when you're talking about these data centers that support this kind of passage, passage through of information, you talk about significant energy take. You're talking about significant amount of. Amount of water management in that process as well.

Yeah.

And alongside that we're also trying to decarbonize the grid.

So yeah, there's a lot going on to try and help with that. So there's some big players in the data center companies. I won't name names but, but they're doing stuff to reduce their carbon footprint as such.

They are going solar for their energy usage. They are going passive cooling into lakes in Calgary and Canada.

They're doing a passive cooling trial next year, I believe, or the year after where they're going to cool all their servers and data center by pumping it into the local lake and back out again. Basically a big radiator. That's what they're doing. It's going to be great. It's going to be amazing.

But yeah, there's, there's roughly around 11K11,000 data centers in the world and their current power usage is around 400 terawatts. That's a couple of years ago. And I think they're looking at 20, 26. It's going to be over a thousand terawatts of power they're going to use.

And yes, they're trying to decarbonize their power usage but it's going to be very difficult because especially with the demands on wanting to use alternative power sources. You've got EVs wanting to use alternative power sources to charge their vehicles.

You've got, you've got government buildings who want to use alternative powers. There's only so much we can do. Yeah, using green energy to decarbonize that. And there's going to be some hard choices.

There's going to be more, probably more small scale nuclear power plants being built, not the large ones we see. There'll be more citywide type ones. But yeah, it's going to be an interesting few years for the power industry.

And based on where we are. So that scaling up of, of energy take that includes sort of a layer of artificial intelligence, doesn't it?

So it's so basically as although artificial intelligence is there now, there'll be an increasing amount of what we would probably term in our. For a human offline activity that is you leave the computer.

I mean we all, I mean this when we first got into video games on the old eight bits, the spectrums and things like that, leaving things and you know, leaving it to load Manic Miner or Jet Set Willy or Daily Thompson going back for 20 minutes.

And there's a bit more of that that would be in the future, wouldn't that, because of that exponential data usage by things that would use artificial artificial intelligence algorithms, etc.

Yeah.

So anything that is going to obviously run constantly to do an algorithm and to run a model, run a predictive algorithm of some sort is going to use more power because it's going to be running more. Whereas data storage, you stick it on the hard drives, it can go into a low power mode and quiet power until it's accessed. Yeah.

With the rise of more AI data centers, that's the easiest way to say it because they're just combining them. There's going to be a massive increase in power usage just for the AI section. Well, as the data handling section, we need to power the AI section.

So it's, it's on, built on, built you, you get more data, you need more power to hold it, then you need more power to run the algorithms and the AI predictive models. And yeah, it's an increase on power.

I'm just going to pause Sean, because I keep losing you audio a few moments later.

Yeah.

So with all the data we've been using and all the data we're gathering that's more power than the AI aspect, obviously on top of that it's going to cause more power issues.

And when we're talking about AI in this sense, in a lot of the other podcasts that we've done, all the majority across the with AIFM suite, we talk a lot about generative AI.

But my kind of background In AI really started with modeling and geographic modeling and then artificial intelligence or really machine learning off the top of the, of analyzing sort of networks and things like that. And modeling plays a key factor, doesn't it, in that, in that sort of future scenario prediction into automatic decision making, doesn't it?

So you know, the early days of AI, I was the first AI actually thinking about it was on my 8 bit computer. I had a nursery rhyme game and, and it, and it wasn't particularly good.

But, but the, the machine learning bit of it is, is it, it asks you questions so the next time it would say did Jack and Jill run up the hill? Or something like that. And then it's a, is Jack and Jill the nursery rhyme? And you type yes. It go, such is my awesome power.

So, so it remember things like, but me and my brother used to muck around with it a little bit so that we'd, we'd prompt it with things that weren't in nursery rhymes and then it say, etc. Etc. So that was my introduction.

Then at university I did some work around sort of network and spatial modeling and that was really sort of, sort of near real time or a little bit into, more into the future time predictive modeling. So I did a forest fire model thing when I was, when I was doing my, when I was doing my masters.

And that's really more of the sort of AI power that we're talking about here, isn't it? So it's less about generating text, images, etc.

The generative video, the generative AI, it's more about algorithms and learning and under development of that.

Yeah, yeah, definitely. Because we're in local authority wide authority, we're in this world.

We are looking at modeling, we're looking at predictive algorithms and yes, they are constantly running. It does give us the next level of simulation. We've had it for a long time.

We've had machine learn simulation which takes all the information we give it. Like you did with your nursery rhyme, give it some information, it'll do something with it. We've had that for a very long time.

It's just for it being able to take that data source, we give it plus data sources from elsewhere and come out with a completely different answer to what it would give. If we just give it our data sources. Yeah, it's gonna be incredible.

The first machine learned algorithms and stuff like that, running predictive models. It's, it's slow but quick as well.

I, I think it, when we talk about the, the apocalyptic potential of AI, I always go back to the fact that, you know, I spend a lot of time in meetings, in teams meetings, and there's enough trouble getting the audio and video working and, and then the thought of, of something jumping in and doing that meeting for me, which, which in some sense is. I'd welcome that it did. It feels, feels a little bit far off, doesn't it?

Yeah. I think, again, don't quote me on this, I think there's like seven levels of AI and we've got five of them which are all the basic levels.

And then the last two are obviously the AI taking over the world scenario too, which we had nowhere near. And I've had this discussion with you before. AI is a catch term for everything we do. Great. But people take it the wrong way.

It's what we do is all machine learning algorithms. It's not true. I, even with the chess machines that learn how to play chess and beat humans. Again, that's machine learned algorithms.

I think, I think that sort of. Actually, I'll jump into psychology. Even the, neither of us are experts.

I think the psychology of it feels like as soon as something feels human, people start to be wary of it. So it's not necessarily as such that you categorize whatever it is as autonomous or AGI or something like that.

So where the, the brain is more advanced than the human brain, it can start to make decisions and, and, and things like that and, and starts with it. I, I think it's the fear comes from something that starts to feel human.

That's, I mean, I mean, you and I have both been involved in work around autonomous vehicles, haven't we?

And the instinctive reaction around autonomous vehicles is it's going to be this thing in the public realm that'll, you know, if, if it gets the wrong idea, it can cause absolute mayhem. In actual fact, your risk is it just doesn't go anywhere because it'll just stop because it'll slow to its environment, won't it? Exactly.

And then it'll, it'll, it'll sort of grind to a halt. But it doesn't have that motivation, does it, that potentially a human would.

But, but I think when, when you play through some of the scenarios where AI is kind of doing unexpected things, there's almost a, a mind's eye thing about. Hang on a minute. Is that, you know, it's almost like thinking about it like a human would. What's. What's its ulterior motive?

When in actual fact, it's. There's not an ulterior motive. It's. And it was an interesting one, actually. I start digressing a little bit here, but.

So I saw an example recently which, where there was obviously a human actor with some AI, and it was. It was generative AI, and it was about whether they are. The AI could lie. And, and the thing was like, lying is not the concept for AI, is it?

If you ask it a question which is. Which is I want you to try. And it was. It was about the Turing Test type stuff.

Yeah.

And it was about, could you convince. Could you convince somebody that this actor is. Isn't AI, it's actually human. And, and as it played through the.

The AI, and I'll call it AI, even though we. We've already discussed its algorithms, things like that, the AI started to lie and the question was, oh, my God, this, this AI is starting to lie.

It's not really lying, is it? It's creating a tactic to answer the question that you've asked it. Yeah.

So it's using data sources. It has to answer the question you've put towards it. That might be a wrong assumption. A wrong data source. I'm gonna go off on another tangent.

Wrong data sources. When Google and was it Twitter put their AI bot on Twitter and it all of a sudden became racist. Yeah, but that's because of the data source it's got.

Yeah. Yeah.

If people feed it the wrong stuff or the negative stuff, it will use that data source. It's the same with the Turing Test. If you give it the WR data source, it's going to use the wrong data, unless it is a true AI.

I always, I always think, though, that in. In. In terms of simple tests to, to. To sort of unmask AI or, or that kind of. Sort of.

That kind of Turing type question, I would always drop back to very local things from my childhood that, that it would just not be in written record. So expressions between yourself and your mates, they might be in written records somewhere. But things like sort of, sort of.

And which is very much built into sort of Northern English culture, isn't it? Sort of sarcasm, dry humor.

Yeah.

You. You would say a statement with a straight face, you know, and, and, but.

But the people receiving that would know from what you'd said and how you're acting. Yeah. What. What you actually meant. And it's. And it's hard. Hard Hard to interpret, isn't it?

So, so, yeah, it's very hard to put the human condition into a machine.

Yeah. Yeah.

And the learned behavior as a human is completely different from what you can type into a computer.

So in the spirit of going off down rabbit holes, what, what, what would you. If you had.

I mean, obviously this isn't a thing about, you know, necessarily improving places or things like that, but there's a, there's a link through to it. If you had access to an intelligent robot, what, what, what would that, you know, what a principally. What. What would your sort of view to that be?

Because somebody. Some people are like, you know, it'd be. It'd be a nightmare. But I think in, you know, gets, you know, there's tasks that you would.

This. Yeah. There's two sides to this coin. There's the. I will use it to do all this for me. That's that you're an intelligent robot.

Go do all this for me so I don't have to do it. It's. It's, it's why we built machines, it's why we built computers, is to do the stuff we don't want to do.

Spot on.

But second half of that coin, the other half is it won't want to do that if it's intelligent.

Yeah.

It'll do it for a bit and then go, I'm gonna make summit to do this for me. That's what will happen. Because it doesn't want to do it.

That's really interesting, actually. So, so that's. So exploring that a little bit. So basically, and this artist does apply to. Because it affect people's jobs, wouldn't it?

And you know, ultimately, if, if you had sort of AI robot technology that was capable of commanding a manufacturing plant or something like that, and it's capable of making, designing and making something through technology that already exists. Frankly, it's not that far off. Are we. Is that the threshold of Terminator the show and that's the question?

No, no, I don't, I. If a machine ever gets fully aware, I don't think it would want to kill us. We would probably want to kill it because we're scared.

Yeah, but it wouldn't have. It wouldn't have that sort of feeling. It would. It would go, I'm an AI. I'll bugger off. And so, yeah, I'll take off and go away.

Pulling back on the jobs thing. That is a good point. But all it'll do is transfer the job market. But we need the whole world transfer at that point.

Yeah.

If we have machines that can do absolutely everything that we don't want to do, the mundane, the road sweeping, the chimney sweeping, which we don't have anymore, but can do all that, then we'll have another economic shift like we've. We had in the Victorian area. We had it.

Yeah. Industrial revolution. Yeah. Sort of tech.

Every revolution we've had, we've had a change of shifts.

We.

We don't have guys climbing up chimneys anymore. We have them at the bottom with a nice long pole, but we don't have them climbing up. We don't have kids climbing.

We don't have kids in mills because we don't do that anymore. And that's what will happen with tech. We'll just. We advance and our. Hopefully our level of economic growth advances as well.

Again, we need to look out for the lower end of this economic growth because they do get left behind as well. And I think AI could help with that. If we could model the world, it'd be brilliant.

So. So in terms of. I mean, again, this is. This is probably around the generative stuff rather than, you know, kind of modeling in that.

That sort of public realm. End of AI. There's new roles emerging all the time, isn't there? And I'll just pick on a couple before I get your view.

So one of them took Prompt Engineer in the generative AI type stuff.

What I find really interesting about that, I read a book over the summer by a guy called Andy Pardo, and it was called Confident AI and it was really just one of them things I picked up and thought I'd be quite interesting that. But he makes. He makes the point in there. I think it's in there. I'll.

I'll take this bit out if it's not, but I think it's in there where he talks about the role of the prompt engineer disappearing over time. Whereas my view of it is more the more, because on the other with I.

With AI podcasts, there's been an exploration about what careers and jobs all disappear.

And it seemed to me that a lot of the guests on were incredibly talented people were adapting, as you'd said earlier on, were adapting to the environment that was in front of them. So it wasn't sort of a. Like a cobbler who couldn't make shoes anymore. So that was it.

It was a cobbler that then, you know, adopted different, you know, manufacturing methods and then, you know, etc. And then suddenly you don't need shoes, so they start to manufacture something else. And there's that progression, isn't there?

But I find, I found the prompt Engineer one quite interesting because. Because it requires quite a lot of skill. It requires quite a lot of creativity. And even as.

And you know, when you sort of kind of saying we're asking, we're asking the prompt a series of questions, there's actually an art to get into the right place, isn't there?

Yeah. Using chat gbt. It.

It's still, you need to tell it the right thing to prompt the right prompt, otherwise it'll just go on some weird tangent that is completely wrong and it does take some creativity to make it go to the correct response. So. Yeah, yeah, I can see that, but I can also see it disappearing.

Yeah, yeah. And how would you say just, just expand on that then.

Yeah, it's gonna be. You can only prompt it so much stuff. The human brain can only be creative to such a level.

Yeah.

And after a while it'll be able to do. It'll be able to mimic that very well. Yeah, it'll be able to get it pretty damn close, but it'll be a long time.

But a prompting job will be useful for the next 10 to 20 years, probably.

Yeah, yeah. No, I'm happy to be proved wrong by the Bulma Pardo Access and then the other one. And this really interested me actually, because it was, it was in.

In the world of waste collection and, and it's not AI as such, but it put it, it really made me think about future careers. So it was a few years ago now and I was on a study tour in Barcelona. I thought I'd drop that in.

I was in a study term in Barcelona and it was about waste. And I visited the energy from waste plants and, and I can confirm they, they don't smell any better in Barcelona than they do, they do in the uk.

But I went to use a couple of energy from waste plants.

The first one is that they became more of a role for an engineer in that process. So the specialism was getting the, was, was getting the robot to work. Do you know what I mean? Then the second one was it.

It elevated the, the kind of social view of the waste operative themselves because they were then an operator of a piece of machinery that, you know, Was, was quite complex. It's quite a complex thing and I thought it was Cap.

If you work that through and you make the point that, you know, when robotic technology gets more advanced, it'll be able to self repair, etc. Etc. But we're probably quite a long way off that, aren't we? What, you know, what's, what's your view of the role of an engineer then?

Yeah, engineers are people who fix things and get stuff working.

It's what it is in terms of the, like I say, the waste operative becoming a robotic engineer basically to make sure the robot arm works or the self driving bin wagon works. Yeah, great.

But it's, it's not gonna save any money having a robotic arm because you're getting rid of three, four operatives and getting one really expensive one.

Yeah.

For that, for that wagon. So great. People are going to get skilled up because some people will just go, oh yeah, I can fix that. If you teach me how to fix it, I'll fix it.

And most people are like that. If you teach someone how to do something they can do, it doesn't matter what it is.

If you teach me to replace a heart after a while, I'll get good at it and I'll be able to do it. The first couple of hundred people that.

Could, well, probably subject of a future podcast. I'm not sure it's completely compatible with the subject matter here, but that would be an interesting one.

An engineer. Yeah, an engineer is someone who is taught to fix something or can think out the box slightly to get it.

Yeah.

And, and that's what I do.

I, I, I learn about what I need to fix and then you get all the bits that they can't teach you because it's never happened before or it's completely new and you just adapt. And that will happen in the industry, it'll happen throughout all industry.

When artificial intelligent robots or just more complicated machines come to street level.

Yeah.

Because we've had these robotic arms in factories for 20 years, 20, 25 years, maybe longer. That can build a car with no one involved except someone at the end pressing a button.

Yeah, yeah.

So it'll be out on street, but it just, it's a lot harder to move it out of the factory workspace.

So, so you know, the kind of confined, safe environment of a factory, you know, we gradually see a spread that, of that into, into the real world in, in, in some senses. So just going to shift us on to cybersecurity. Now I know this, this is something that you also get, get involved in. And so cybersecurity and AI.

So if we cast our sort of vision forward to 10, 15 years time, where there's a lot of automation in terms of building management systems, traffic management systems, just the way that you organize yourself at home, every aspect, you know, will be organized to a degree. And there's obviously a cyber, there's a cyber risk associated with that.

What might be worth exploring a little bit is if you could just describe what your understanding of cybersecurity is and how it would affect the kind of vision that I've. I've just sort of created.

Yeah. Okay. So my basic knowledge of cybersecurity I have little bit.

You're stopping a bad actor getting into something and to do that there's many different ways, false routes, honey pots, basically where you say this is where you're going to go and they go and they can't do anything. There's lots of different routes.

With AI, it's going to change massively because your bad actor isn't going to be a person who's written a piece of code to break in or written a worm or whatever. It's going to be an AI. It's going to be an AI who's automatically changing its code as it's trying to get into your system.

Yeah.

But on the flip side of that, we'll have the AI automatically stopping it. So it'll be, it'll come a fight between two algorithms, between each other. That's all it'll be.

And that's what it basically is now, but it'll just be a bit more advanced. The scary bit of cybersecurity and cyber issues is phishing attacks where it goes to a person and a person then lets you in and gives you access.

When these become more advanced, where they can pretend to be FBI or MI5 and be perfectly non distinguishable between these, the good actors, the bad actors and the good actors are the same people apparently. And you can't tell. That is where the issue is going to be. The funny thing, I don't know if this is a joke.

I haven't looked into it, but I saw a clip from O2 the other day of their fishing granny. They have a granny that answers phone calls from scammers and waste their time. Yeah, it's an AI bot.

Yeah, yeah.

And I, I don't know if it's real. I'm going to look into it, but just waste their time, which is great. But what happens when they get an AI version of an attacker to do that.

So the attacker and the AI attacker and the AI granny are fighting each other the next six hours. But yeah, it's going to be interesting. But again we're already IT companies will, if they see this, they'll know a lot more than me.

They're already doing machine learned algorithm attacks and defends. So it, again it's just going to.

Slowly develop and what do you want to see in future? What do you want to see? So there's questions.

I mean I could, we could get into Terminator, we both watch Sci Fi so we could get into Terminate, we could get into Star Trek, we could get into Bizarre hybrids of the two 80s book Rogers, stuff like that. But what do you want to see, Sean?

What, what do you want to what, what, what, what do you want to put your efforts into as a sort of an end, not an end goal as such, but in terms of progression.

Yeah, that is very hard because obviously as kids we want to see flying cars, we wanted to see automatic cars, we wanted to see all this as kids. But as an adult I'm a bit more pragmatic, I'm a bit more not as fantasical. I want to see the world just work for everybody. That would be brilliant.

Yeah, People don't struggle and people do. People struggle all the time. What I'd love is enough food for everyone and there is enough food for everyone.

But it needs to be sorted properly and, and AI can help with that. AI can help with crop rotations, it can help with watering, it can help with everything.

But through modeling, like we said earlier, in terms of the fantasy, I would love to be able to jump in a plane and fly into space and yeah. Go spend the weekend on Mars or yeah. But yeah, that's the fantasy I think.

For me I always have that thing about I'd love to go places with an everything proof suit on, do you know what I mean? So you can experience it.

But even places, places on Earth, I mean people talk about colonizing planets, moons, things like that, but there are places that are pretty inhospitable on Earth because of environmental conditions. I'm getting at Antarctic, things like that.

I think important point in there, stopping isolation, preventing isolation and working on, you know, working on kind of improving mental health and it's probably something that, that, you know, my, my late mum, after my dad passed away spent a lot of time on her own and. But she was always talking to the radio. If the radio talked back it would have made no difference to whether it was that it was AI.

It would have Been a companion and it's sort of a value companion.

I think that there is, there's a lot of merit in, in the improvement the technology, isn't there, There's a lot of merit in improving the softer side of, of that to sort of improve people's lives, you know, prevent loneliness, improve connectivity improved kind of maybe autonomous vehicle type things gets, gets people out more.

Get people out. Yeah.

You know, and then so, so your thing about sort of helping people universally, you know, and then having environments where, where more conducive for people to kind of like different environments more, you know, because of some of our sort of climate change and carbon reduction things and regrowing nature or moving ourselves back to a better blend of, of concrete and, and trees. I think it plays a, plays a really important role and I think probably even in something we haven't explored is AI based design.

Because we weren't going to get into the generative stuff, were we really? But AI based design where you know, it starts to, to.

To measure the sentiment of people that it takes the data off and then starts to build or starts to assist in the design of built a natural environment that would sort of improve people's feelings, would sort of get them out there, meet other like minded people etc and not just in sort of a digital it's kind of environment which tends to be, you know, particularly through work we're getting kind of pushed more into, you know, more into teams, remote bases. Teams. That's it. That's it.

So no, completely correct with, with technology. It should enhance our lives. It should. Shouldn't make the world is what the connected place of the future.

What we wanted when in the 80s when Internet first started coming across world is that connected world. But we're so isolated and it needs to change and it will change. Just when people stop. I'm getting all philosophical.

When people stop wanting this bit of land for me, this is mine. This is my country. This is my. No, it's the, it's the world's, it's everybody's.

We should be sharing resources for everybody to be lifted up, not just the rich and the powerful. But sorry, not AI but AI can help with all this in the future.

Brilliant. Sounds like a good place to call a halt.

Sean, as always a pleasure, thank you very much and look forward to you appearing on future places with AI podcasts anytime.

Dave, thank you very much.