Artificially Speaking – Exploring AI’s Impact – [FULL INTERVIEW]

David Marr discusses the evolving role of artificial intelligence (AI), particularly large language models, in society and industry.

This interview highlights AI’s transition from a niche interest to a pervasive topic, exploring its impact on jobs and societal changes, the waning novelty, and the shift towards finding substantial, impactful applications.

Find out more about Artificially Speaking -> Here.

Key Points

  1. AI’s popularity surged last year due to major releases from companies like OpenAI.
  2. Interest in AI, particularly large language models, has shifted from novelty to seeking practical applications.
  3. AI is increasingly integrated into customer support, replacing human roles, as exemplified by Klarna’s use of AI to perform tasks of 700 employees.
  4. Discussions around AI include its broader societal impacts, particularly in job displacement and economic shifts.
  5. AI’s adoption is likened to historical shifts in technology, such as the Industrial Revolution.
  6. The rapid development of AI technologies challenges human adaptability and job stability.
  7. Ethical and practical implications of AI are debated, including the balance of automation benefits against potential job losses.
  8. The implementation of AI in various sectors promises efficiency gains but also raises concerns about the pace of change and societal readiness.
  9. AI’s role in content creation and customer interaction is seen as a double-edged sword, offering benefits while posing risks of unemployment.
  10. There is an ongoing discussion about the need for a legal and ethical framework to manage AI’s integration.
  11. The potential for AI to exacerbate inequalities or disrupt employment patterns is a significant concern.
  12. AI’s impact on traditional industries and roles is compared to significant historical economic shifts.

Key Statistics

  • AI performing the job of 700 people at Klarna.
  • Human tasks being completed by AI in 2 minutes, compared to the human average of 15 minutes.

Key Take Aways

  • AI’s rise reflects significant technological advancements influencing multiple aspects of life and work.
  • The shift from novelty to practical applications in AI indicates a maturing field with increasing integration into everyday business and social functions.
  • The rapid evolution of AI technology requires a reassessment of job roles, skills, and education systems.
  • Societal adaptation to AI’s capabilities and impacts remains a significant challenge.
  • There is a critical need for comprehensive regulatory and ethical frameworks to govern AI deployment.
  • AI’s potential to displace jobs, especially in customer support roles, is becoming more evident.
  • The benefits of AI in enhancing efficiency and decision-making in businesses are counterbalanced by the risks of increased unemployment.
  • Historical parallels with other technological shifts suggest potential long-term benefits alongside short-term disruptions.
  • AI could lead to significant changes in employment patterns similar to those seen in past technological revolutions.
  • Public and private sectors must collaborate to ensure equitable benefits from AI advancements.
  • Continuous dialogue about the implications of AI is essential for fostering informed policies and practices.
  • Monitoring and managing the social impact of AI is crucial as its applications expand.
Interview Transcript

0:02
So hi, everyone. I’m here with David Marr today. So David, I know you’re busy doing a podcast called artificially speaking. And you got a new one now, which is really interesting podcast, I think his thing is called right. So we’ve gone back a little while, at least anyway, we’re having a great conversation around AI. And so that’s a bit of sort of your passion and interest and just really a bit of a catch up ready. Yeah,

0:22
exactly. I think AI started as a passion, but it’s now I think it’s almost on everyone’s lips. And I don’t it’s easy to say it’s on it was it but every day, there’s just new stories that involve AI. And it’s almost like it’s creeping up now on everyone, which I know sounds quite dramatic, but it’s starting to feel that way.

0:43
Yeah, where do you think because like, last year, it felt like, it was almost like all you’re talking about, and particularly large language models, and even then it’s not new, is it right, because it’s been around for a long time. But it obviously it just burst on the scene with open AI last year. And everyone was that’s what everyone was talking about. Last year, we were all talking about it. Why do you think it was so popular? But then it also, it seems feels like it’s almost reduced in terms of the intensity of conversation this year? I know if you see the same?

1:10
Yeah, I think it’s like with any kind of, as you say, it’s a lot of it isn’t new. It’s just the the medium that people are harnessing it from, and I suppose with chat GBT and then whatever Google Gemini was called before. It’s given people the kind of the tool to play around with and then do something with and I suppose the way I’ve seen it is that people have been trying to utilise that or find how they could harness it. But I think it did people have, I’m not saying it’s gone away. But people maybe have moved away from doing that sort of thing. And we’ve been looking for the first or major kind of implementation of it, that actually resonates with most people. And I think we’re starting to see that now. Really, especially in the last few last week or so,

1:58
I suppose some of it was almost like a consumer LED product. And we could all do it. So all of a sudden, it’s like complex things that can only be done using coding before like, all of a sudden, we could all do it. We got keyboard warriors coming up with usually like translating everything in the style of Shakespeare, or I’d like my policy written in the style of rap, or something. Because we all had fun doing it. But I suppose then almost like the novelty wears off, doesn’t it? And you get to the how do you get the real use cases sitting in the background, which is a little bit more complex laid out?

2:26
Yeah. And that I think you’ve summarised what I was struggling to there, people were playing with it. And there was an even like, I was looking at some of the startups that have raised funds. In the last 12 months, some of them are dabbling with different ideas. And they’ve managed to raise funds to try and take it through to a product. But they’re still I don’t know that they’re trying to find market fit. But the one of the big story for me that’s made this all seem real, if we call it that was Klarna, announcing last week, that they’re now using AI to do the job of 700 people. Now, they were quite open about what they were doing or not, it was it’s mainly for customer support, and dealing with issues problems with that kind of where payments have gone awry or anything like that for their customers. I don’t know how many people Klarna would really be have or would have in place dealing with that sort of thing. But they must know that it’s carrying out the function of 700 people that one of the stats they were pushing around was that a human doing the role would fix something in 15 minutes on average, whereas AI can do it in two minutes. So that kind of that’s that making it seem real and actually have some sort of business usage. Yeah. Do

3:42
you think that’s any different from just automation? Anyway, you go back to having let’s go back to 50 years ago, let’s say like and everything was based on paper, you go down to your bank branch, and everything will be based on who the shuffling papers around, you have your own your own fire, wouldn’t you actually sit on the shelf. And what’s happened is that’s gradually being computerised. And now if even a call centres have gone from being like 1000s to being hundreds, in the space of what 20 years, do you think it’s any different than just automation is anywhere and we’re presenting it something new, but maybe it’s not?

4:09
I agree, it’s, it’s a bit like as well, like I was chatting to someone else about this. It’s a bit like offs, like outsourcing or offshoring of stuff. So say if you’re moving, say in accountancy function to India, I’m not saying it makes it more efficient just makes it cheaper on that regard, but it’s all what people might regard as progress. But I suppose where AI is changing slightly is the speed at which is bringing in this change. And I think, I suppose that’s the biggest issue, I think, for society in that you’re bringing in all this change, and you’ve always had problems with change anyway. But I think it’s just the size of the change that’s coming and how quickly it potentially could be implemented. So I don’t think it’s any real different from automation, but it’s just it’s happening at such a faster rate. But what’s going to happen that foster it

5:01
was quite an interesting discussion, I suppose in terms of like we have human beings have a certain we have a certain rhythm, don’t we? Yeah, we wake up every day we go to sleep at night, we’re on this like daily, monthly yearly kind of cycle, which is the rhythm of the earth to a certain extent Exactly. And almost like the digital is almost a digital process is almost like operating on different timescales all together, work 24 hours, they go into, like, microseconds, and they get even, like, even shorter timeframes, right. And it’s so much faster. And as humans, we can’t really keep up I don’t think, no,

5:31
I don’t think we can. And but also, going back to what I was saying this the impact of change, I think, as you say, we’re creatures of habit. And I think change is just the big thing that humans just don’t handle very well, I know, we always activities to try and improve that. But I just think if you increase the speed at which that changes happen, I think that’s the biggest impact we’re gonna see in the short term with AI, because it’s just I just don’t think we as a species, if we call it that and able to deal with it that well,

6:03
what do you what do you think we go with it. So there’s been a lot of discussion, I mean, it promoters and detractors have particularly large language models, I think, because of that change the change readiness to certain extent, but one hand is going to, if it gets connected into the physical infrastructure is going to cause all these problems in society. And then other people like just absolutely fantastic, I’m going to use it to basically do my washing at the weekend, or whatever it is. So the basic means I don’t have to do any work, write my essays, mark all my essays. And I don’t know, I can just sit back and watch Netflix. I don’t know. But it’s when you sit on that, what do you sit on it from? The discussions? I suppose you had, I just gonna be the impact.

6:39
I think the impact is gonna be is huge. And I know that sounds dramatic. But I think, obviously, that the impact you I think you have to look at it in two, two ways, really, the impact of the technology and the improvements it will bring. So there is positive, so it may makes me sound like I’m doomed. Kovic, but we’re getting closer to the Terminator kind of what we’re going to be driven off this planet. But the I suppose the there is, I think there is a lot of benefits to us with the AI technology coming in. I like one being I think medical and healthcare type. Benefits, I think we’ll start to see in terms of analysing health data and stuff, I think that’s actually going to be big. But I suppose the biggest challenge is the other element of this, it goes back to people having to deal with this challenge, I think like, I think, and this is really just my perspective, from speaking to different people through the podcasts, but also other discussions I’ve had with others and seen AI being used day to day. It’s the bit that worries me is just the human aspect of this in that I just don’t think a majority of the population are going to be able to handle dramatic change that we may see over the next five to 10 years. And I think it will be that quick, suddenly, even quicker, I think it’s going to unbalance the norms that we take for granted. It goes back to what you were saying about people having structure following the natural rhythms of life, I think it’s going to affect people that I think most people regard that there’s not many jobs for life anymore, but I think they, they assume they’re going to have some sort of job stability. But I think in some professions and some jobs, AI will have quite a dramatic effect. And it’ll actually it could lead to more redundancies in a short amount of time. And I think we’re estimating and that’s the bit that’s going to cause the big impact. And

8:44
I suppose if you look at such as the Industrial Revolution, if you look at the digital revolution, if you look at like even the internet, the web revolution, that kind of happened, each one of these has resulted in there being a big change in terms of who was employed, what kind of jobs there were, you just wonder if it’s the same thing gonna happen. But I think what was interesting with those was, there was some people who did just incredibly well. So I remember when I was young, I wish I’d become a game designer. I used to love programming, but yes, it was that X 81 or Zedeck spectrum, I can handle that. But then it’s just accelerated the early days of the web with HTML, and just, it’s just accelerated to be much more complex. And if you don’t keep up, or if you’re not in it, then you fall behind a certain extent. And you just wonder if the people who do keep up like Bill Gates, like Steve Jobs, they obviously they did very well for themselves, you could say, but there was a whole slew of people who didn’t keep up and they fell by the wayside and I just wonder if the same thing is going to happen again with large language model where we’re at the ground floor. If you keep up then there could be fantastic opportunities in if you’re in for that change. If you’re not then that’s why there’s a danger you can fall behind ready to to speed society. Yeah,

See also  Data Dominance: The Real Power in AI Development

9:52
massively, but I think where this is going to have, and again, this is my personal view, it’s not necessary. Really what others are thinking, the way I compare it to is, if you think like in the 70s and 80s, when we had the decline in mining, and obviously it was accelerated, with the mines being closed and stuff. And what you then had was areas of the UK that where, you know, people just didn’t have any other job options to them, because mining was the major part of industry in those areas. Now, that obviously, has been a major issue that’s still going on now, where sub multiple governments have been trying to instigate economic change in those areas, that’s in areas where mining is was dominant. But I think with AI, you’re going to open up areas of the country that have traditionally been quite prosperous, in the last 20 years, where there will be lots of people unemployed, and I think that’s where it worries me, it’s a bigger kind of version of that. And that’s, as you say, that’s always happened as technology has moved forward. It like you say, since the industrial revolution, but I think it’s gonna be on more people’s doorsteps, if that makes sense. In that, if you think like, the example that I’ve seen numerous times is Newcastle in the northeast, there’s so much call centre, there’s so much customer service type functions that are in those areas in that area. If you jump ahead with this Klarna type model, where you’re replacing 700 people, where does it leave places like that? If that’s my worry, it’s gonna be on more people’s doorsteps.

11:33
And I suppose just in terms of timing, and I suppose historical acceleration, I think it’s called right. If you look at that, it’s like when it was the industrial revolution, it was almost like on a generational level, right? So your job might not have gone, but your children will probably train for the new technology. And that’d be fine. That’s quite liberal, isn’t it? Because that’s the kind of scale, if you go back to the some more recent times, it might be that you do like a 10 year scale or a five year scale, it feels like we’re almost like a two or three year scale. Now there’s the speed at which things are changing. But it’s even faster than that. It seems like it’s developing fast, and you can keep up as a human being in terms of those kinds of things.

12:06
Yeah, exactly. That’s what I’m saying. That’s what I’m seeing is that, and also what I’m predicting is that the ability to change. And obviously, the pace of that change is what’s going to envelop what’s going to come over us. And that’s that and as you said, like there will be winners, and they’ll be losers in that. But my concern is that there will be we’re not just talking winners as in like they’re running multibillion dollar corporations, they’ve made lots of money, you’ll then bring that down to the Joe Public, your eye kind of thing is the survival of the fittest in the job world is going to be hitting every sort of market there is at some point anyway, that that’s the big concern I have already. Yeah.

12:47
What are some of the interesting use cases you’ve you’ve seen, I’d

12:51
say it’s a bit like what we started off with, there was a lot of kind of, look, we’re using AI can’t gimmicky stuff. And then it settled down a bit. But the the call instances that have got practical usage, were actually some of the cool ones. And actually, once you could use our round, like when we were producing podcasts and stuff like that, the creation of like things like clips for social media, short term video clips, that sort of usage of AI, I know, it’s not got mass appeal to a lot of people, but it actually works quite well. And it’s actually really useful for anyone that’s producing a podcast, I suppose that was the one that kind of you kept going back to because it does work really well. Riverside is good example, they’ve got component on there. Also some of the tools around content creation. Again, it’s very hit and miss. There’s some cool tools out there that could churn out emails and other types of content or blog content and stuff like that. But I would say a 10th of the tools we tested or test have got any sort of longevity or potential that gets you past the, that’s quite nice or quite cool to use. But they don’t necessarily get you back every time that that was I suppose the issue is you just get gimmick after gimmick with any new technology. And it was very prevalent in the last 12 months, I’d say. I

14:11
mean, I kind of find it that the use cases I find them the most useful least from consumer point of view, or almost like the ones where you think I could get someone to do that for me, right? So it’s Can you summarise this email? Or can you pull out the salient points from this email and put it in a table, you’re gonna go through this documentation and look through something? Those are basically I could get someone to do that, right. I could get someone to do that. But he will basically give me a helping hand to give me a leg up. Almost like I would ask another person and they’ll probably do it quicker as well. I’m not gonna repeat it. I’ll

14:38
be honest, those sorts of tasks at the moment, I’m seeing the most efficient way to do that with AI is through chat GBT or have not used Google’s since it’s changed to Gemini, but I was using it and as it went, it was bad to do that sort of thing, and that both of those are by far the most practical usage of Vega. IT technologies that I’m I’m using on a regular basis. Really?

15:03
Yeah, it just feels like that’s the opportunity, I suppose. And it certainly isn’t. I’ve seen it. And I suppose some of it is interesting because it’s fuzzy logic, isn’t it? I mean, it’s so we have to remember, you remember, we used to talk about fuzzy logic. And we use that fuzzy logic a lot for name matches, and those kind of thing is like the latest technology. Now, it’s like, you just put it into chat GTP, or one of these large languages, and he just does it automatically. But I suppose it one of the things I find strange is it comes out with a different answer when you ask it a second time, or you can try and get consistency is really hard. And it’s a hard to come to terms with that with a computer, right? So if you’re talking if you’re talking with a human, so if you’re almost like you relate to it as a human, yeah, you say, David, I want you to like, give me this article five times, each of the times, you might come back slightly differently, because the fifth one might be the end of the day, and you might be tired, you’d run out. And it’d be like, maybe it won’t be the same quality as the second one, at least for me. But it’s almost because it’s a computer. And I suppose, because it’s technology expects to be the same every time and almost like our impression of computers. But actually, it’s much more like a human experience, which is like changes all the time. And actually that consistency, I find a bit of a problem, which is I want to summarise something and it summarises it differently in different times. Yeah,

16:10
I agree. And there’s times where I’ve thought like, is that just whoever’s developed the model? Have they deliberately put that in there to give it a more of a human kind of aspect? It’s weird, because it I know exactly what you mean. I was just getting, I was trying to get description of something. And I, it was late in the day, so shooting chatting with Dieter Reuter. And I was just getting multiple versions of it. And you’re right. In some respects, I’ve got less patience with it, because I know it’s a machine that’s doing it. But I was getting a different version of it. And I was of the opinion what I was getting back was getting steadily worse as what was being given, it was less than the original one became more and more valid because of the future bit. And I say I don’t know if that’s down to how the models been built. It has to behave like more like a human. I don’t know, you

17:00
must have seen those it Dally and some of the new sort of like video, I think sores that was the was the open AI one, but there’s others. And there’s the Chinese one that came out the other day. And they’re so convincing. And I suppose part of the issue with the text videos, and example is you can’t edit it or Sony Pictures, you can’t edit the picture, once it’s been done, which you can with a graphic designer, you can even tweak it and those kinds of things, do you think you’ll ever get to the point where they, you will get the source file, so you get the graphic design file that comes out rather than just the JPEG or the GIF, or whatever it is, you actually get the you actually get the source file that comes out or you get the source file of the video, you know, 3d video that comes out. And it’s actually the wireframes that come out rather than just the video to a certain extent. Yeah,

17:43
I suppose that at the moment, that’s the secret sauce, isn’t it, then it would make so like I hadn’t an example at the weekend, I’d help my parents own a flat, which they rent out. And they were wanting to rent the flower. Again, there’s a platform where you can advertise to rent properties on. But the photos they take in this bill is not very good. But they’re mainly very dark. So I just went onto Fiverr to find someone, because I just thought that you get someone to do it very cheaply to dislike the pictures make them. So the person I got blatantly just push them through some sort of AI image improver type thing. That the only way, I assume they will be doing that anyway. But the only reason that made it obvious was there was actually a door like in the kitchen, there’s like a fitted dishwasher. And it’s got like a, what do you call it like the cover door, wherever. So I’d open it that was open in the picture. But whatever had manipulation, the Fidget made that door, probably about 10 times as long. So you had this massive door that was impossible to close on the dishwasher. And I was just like, so all these guys do. And I think he edited like 10 images for me. He obviously then went back and said, Look, this doll looks impossible. Can you just change it? And it changed it within seconds. But he just because I said what AI tool are you using to do this? And he just No, I’ve just built like a model myself. And the what he was saying was that he just had to adapt that he just had to change like you say the blueprint of what it was running off. And whereas as you said, if you had that kind of logic, if we call it that, that you plug into the language model, as you said, if you have that logic of how you do stuff, you’re you can do anything and then why not? Literally and then you’ve it once you get that information, it makes the AI tools so much more powerful, doesn’t it? Because you’re getting the expertise, rather than just relying on the model itself. Yeah,

19:46
but it’s also very like text, isn’t it? So you’re getting the raw information or so as the files that you can then tweak, right you then get that person then tweak. So it gives you that helping hand up, doesn’t it? You still got the human expertise, but it’s just a much less amount rather than actually doing the whole sort of design elements,

20:01
I suppose if you think about it, like you mentioned and about HTML and stuff, like if you think like, developers, if you wanted to build web pages, you’ve got like, interface to, you can use like an editor or you can edit the code. And they’ll always be there. And they’ve always been, I suppose more the accurate there’s always been people that have used those sorts of tools to build webpages, but then they also like to lift the bonnet up and amend it at the you know, at the code level. And the same must apply for the AI tools out there. Just there will be people that want to have a fiddle with them.

20:44
Oh, wait, so you have the pictures of your flat waist squat with the Bahamas, sitting in the background? Most of the things and whether that’s actually maybe overstating the nice as nice as Bournemouth is, right? But yeah,

20:56
I was gonna say the pictures were taken on a rainy day. So yeah, it was rain. And once it finished with it was like blue sky. Like that. That kind of gave it away was that you could see he couldn’t get the water off like the windows.

21:16
It’s like every, every every sip mini series TV show made in the UK for export, isn’t it? It’s like it’s like, it’s like a different country. It’s sunny. Yeah. Three days a year. All the filming happens. Yeah, good.

21:33
I was gonna add to that as well. And Joking aside, but he’s got quite serious is with what’s happening with Google in terms of like we were saying about who builds the model when you know, what biases they have and stuff like that. That’s probably the one threat but Google of their model seems to be causing absolute havoc. I don’t know if you’ve seen some of the things that know the bias.

See also  Moving upstream?... Debt Advice in a changing economy - [FULL INTERVIEW]

21:56
And the bias stuff is almost like, yeah, there’s been bias one way they tried to correct for and then it was causing other issues. And it also felt like it was just feeding into, everyone seems upset about everything. And so, Wherever, wherever, wherever you do is people seem to be upset by so it’s they can’t win really, to a certain extent. But yeah,

22:14
I agree that they’re in a very difficult position. You’re upset. Some people with this, and then you’re upset other people with the changes you make it? It’s, it’s a crazy situation. And yeah, I I get there. I’m not saying they’re panicking. But they obviously the what tools are out there now is the net. They’re not there yet. But is the big threat to the cash cow of search revenue, isn’t it? Just

22:40
the speed of development that’s happening? I think it’s interesting, because it does feel like it gets accelerated. And the text video piece is just like, I mean, you feel sorry, for some of the folks out there, we’ve obviously spent a lot of time you’ve invested a lot of your expertise, a lot of money in developing models. Yeah. And it just feels like the speed of acceleration in some markets just feels like it’s just like, they’re just getting leapfrogged. You’ve not been doing it for six months, you’ve got a world leading product, and then you just get leapfrog by someone else. And it’s just, and you just I just wonder how much AI has been used to generate more AI? And are we getting almost like this acceleration effect, almost like within the AI community, as well as self reinforcing. And it’s getting faster and faster, and getting better and better. And as and we’re almost like saying, Oh, I could never do that. But it does, because it’s on a different timescale that we’re actually working on. Yeah, I

23:25
agree. It’s almost I suppose we’re like a human like, if you think like when you’re born, you acquire skills as you develop as a child then going into adulthood. It, it’s like you say every new instance, it’s almost like you’re taking as you would like the previous learnings from the previous instance, applying that and then if you like the developers, development of that new platform is the Day Zero is already better than or at a better level than what the previous model was. And then it just accept, I agree that it’s accelerating. The ability to do this as the, as you said, the video and imagery creation stuff. In six months, the technology has jumped so far ahead. And the difference between the outputs is so not noticeable. That’s the only way I can see they’ve got that massive leap forward. I just don’t see how else they’ve done, I suppose.

24:19
So there’s a bit of a concern there around whether the data that’s being used to train it or design it is almost like aI generated data or AI. So at some point, if you’ve got an error, and use like, self replicate for a long time, does it accelerate the errors and you’re gonna get select strange outcomes we don’t really understand because it’s almost like it’s developing faster and developing into almost like a, in a bias on one particular direction rather than another. And I just wonder if that’s going to come through, and it’s happening so fast.

24:48
Yeah. The way I see is it’s almost like and I don’t want to make it sound like it’s a military type thing, but it is an arms race in terms of it’s not necessary. The tool you’ve got, it’s the data you’ve got access to, and the data that you can run it, if you like, run over to get your model working. So like talking about Elon Musk, but like with Twitter, I don’t know how valuable the Twitter data is. But there’s a lot of it. And I’m assuming Google have a lot of data in which they can run their models across. But it’s just it. If you’re the person or the organisation that’s got access to the most data. Does that give you the best tool? Like AI kind of arms? I don’t know. It’s a weird one. But

25:36
you can guarantee if they’re using social media to train the models, then it’s going to give expert opinions on cats and dogs. Yeah, exactly. I asked her about all these crazy scenarios that were dangerous for the world. But really, what is really expert on is cats and dogs, particularly cats doing funny things.

25:52
Yes, cats are being scared by toasters.

25:58
You get a perfect picture that you get Yeah. I think go back and look at it wasn’t med journey, it was stable diffusion. So I’ve been I’ve been putting them locally. So you can now run them locally. So you can run a local copy of stable diffusion. You can run local copies of some of the other sort of open source large language models. But they’re not quite as good. I don’t think as as certainly as chatting up or the open AI ones, the online ones. But it’s interesting going back to looking at some of the older versions, and some of the pictures that come up and you get those sort of like strange kind of like people with five sets of teeth in the mouth and those kinds of things is

26:33
the other one. I see. It used to see a lot of people with no like, especially women that were like saying the dress, you’d sometimes get one foot or nothing. And they lit it was that doesn’t seem to be happening now with all the newest stuff. And you just got it like you said, are those other players that were the new thing six months ago? Are they just now completely blown out of the water? It just, it’s a dog eat dog world is getting thrown by the wayside?

27:02
Yeah, I remember the time six months ago, we were like, Oh, this is amazing. That’s actually been done. And there’s no, we should be using this other ones. Yeah, it’s just changing so fast. So

27:13
the other bit that I suppose one of the things that through the podcasts, which we kept coming back to you always, all roads lead to, shall we say, like the big parts of the AI industry. So you ended up either BACK AT chat GBT, like with chat, like, either at Microsoft. And then obviously, open AI, Google, the, although you had these kinds of things coming out from different startups that ultimately you ended up all roads led to the big boys, if we call it that and Nvidia involvement at some point, because those big boys had access to the chips to do it. And it’s, in some ways, it’s almost like those bigger players and letting the smaller people make the mistakes or get their foot get crosstalk creating the groove. But then those guys are just going to swoop down and just, shall we say, envelop the market, because they, in some cases, hold all the chips to do it, but also have the resources and mainly people to corner the market. So

28:21
I just wanted an opinion piece just today. Actually, it was about Apple and Apple not not going into AI just yet and pre being the right thing to do. So it doesn’t make some of the early mistakes, being incredibly cash rich, but it was not hesitating but holding back so you can then come in later on. Right. So bit like it’s done with the the Google vision, not the Google vision, the Apple vision. Yeah. Yeah. So rather than Google Glass as far as I was going with that, I mean, that looks pretty incredible. In terms of like what it was three, three and a half $1,000 is quite a lot of money, isn’t it? Yeah,

28:50
it’s a lot of money. The reviews have not thrown three grand or whatever. Can you even buy them in the UK?

28:58
I think you can’t buy them. Yeah, buy them. The feedback from

29:03
people that have used it have said it’s like any new thing you pick it out and to be fair, a lot of them said the same. It’s like any of the VR headset type stuff that even like the metal one, you put it on you go wow that’s amazing. And then you put it back in the drawer and it never comes out again. But what they were saying with the Apple One was it’s really impressive. And actually you could see like, it’s like any Apple product is the first version has got bits that probably need to be improved. But they but one guy saw he said it doesn’t need that much doing to it like the headset looks bulky but when you’ve got it on doesn’t really seem that bulky. Watching a film they said is amazing on it because your lips is like sitting in a cinema. But you’re watching it in your wherever you watch it so you could be on the bus, you could be on a plane or whatever. But the other bit that they’ve all said every other interviews as they can see, and this is obviously what will get it over the bump of getting it out there. People can see where you’d actually use it in a business context. So like, the examples over time with things like if you’re sitting at a desk, and you’re working on different things, it’s a lot easier than working with big screens in front of you. Even like practical if you’re using your hands type jobs. So one of the examples was like, if you’re, they were showing, like people working in like, like those greenhouses that grow like crops all year round, and stuff that if you’re cultivating stuff that you need multiple bits of data, or you need to have information that it’s literally in front of you. And they said it is it is impressive. Have you seen it? What have people using?

30:45
I’ve not seen it live? I thought do you think the review one of the reviews I watched was the guy was playing he’s playing table tennis with it. I thought that was that’s, that’s pretty impressive. Yeah, I like the idea of having it if you get an eight hour battery life, and you can sit on the plane. Yeah. And you just have a beach or something like that. And you just sitting back having a drink or whatever it is yes. Like that I could see see that would be useful to you if you’re in a cramped environment, and gives you that sense of space or that those kinds of things. The other thing I think link to VR will be interesting is if you create these almost like VR environments, like 3d worlds, or a film that sort of rendered in 3d, and you could stand at any point in the scene, like you can in a video game, that’d be a different type of film that you’d never you’d never seen before. Like photorealistic you could be at any point of the scene itself and observe what was going on, and almost move around with from different points of view. That’d be interesting. Yeah.

31:36
And then randomly one of the people I heard reviewing it, for all the site had that same comparison point. And what they what they would point of what they thought was the most interesting was that they were saying was that as much as that sounds pie in the sky, we’re not actually that far away from this. And the example they gave was like Star Wars. So it said, Imagine if you were what I’m not, I don’t mind start with I’ve watched most of them. But like I said, if let’s just take the original one, episode four, well, three.

32:08
The first one is row one, row one. Yeah, the first one

32:13
came out in historical data. So you said if you think the beginning that like you’d see the ship that prints layers, on and down the Emperor come after, he said, Just imagine if you took the perspective of so you get the perspective of what it is Darth Vader come up. But he said, Just imagine, with a device like that you could be one of the people firing at the storm troopers, you could be one of the Storm Troopers, you could also be like on the bridge of the ship. What the hell is all that noise? Kind of Yeah. And so what he’d done was sitting

32:48
down there in the canteen, just having your pie and chips. What’s going what’s going on, are invading.

32:56
That’s the exact he did say on that ship for the selfie game was you can you can have the perspective of working on the Death Star as a very low rank working in the the HR department. What he was saying was the technology to do that is there. And he said, there’s this tool that they’ve been using for the more recent Star Wars films, where you can build the environment, and you can build out what the actors are then placed into. Now he said it’s not a massive leap to use the AI to generate the actors and the what they’re going through. The problem is at the moment is legally they can’t do that, because of what’s been agreed with the all the different acting unions and writing the writers unit. But that technology is there. So what he was saying was you probably not going to get it with Star Wars, but you’ll probably get an immersive experience where it’s driven off an AI engine. And like you say, what do you call that a goggle or like the Apple, whatever it’s called, is the platform where you would drive that from, because it’s just so immersive. It’s a lot easier doing than doing it on a screen in front of you. Just one

34:06
I don’t know. It’d be like a couple of days, but 24 hours 24 Remember that I was just waiting for but I know what happens in a building. But yeah, you can contain it. But then there’s like multiple storylines. You can follow and just experience the same thing like Groundhog Day. Yeah,

See also  A changing job market - [FULL INTERVIEW]

34:20
exactly. The other. The other one, which I’ve seen which again, it’s more it’s a it’s not the technology isn’t there it is. But it’s just like from a legal standpoint, it’s just very difficult to do is the way I refer to it. It’s almost like aI generated karaoke. So what you can do is you can get to say if you think like traditional karaoke, someone’s murdering a song, reading off the screen, but technically speaking, someone could be singing a song. The AI technology can then actually take their voice, improve it and actually bring it into line with what an artist would actually sound like. And then also incorporate them into say, a music video that or like the famous music via the issue is you can’t obviously get it go and get agreement from the artists or the record companies to do it. But the technology is there and it’s just it. It’s a bit like you were saying, you almost creating the story that you sit in, but you have different perspectives and the technology is there, but it’s just not a legal framework. So what’s holding it back? I do

35:32
wonder if an artist’s gonna it’s gonna be it’s gonna be art that does it first to a certain extent. Yeah. I can’t get beyond the Star Wars, the Deathstar HR departments, you’re talking there and they got to you got to bring Darth Vader in again, Darth Vader, you’ve got to stop doing that.

35:52
You strangle them? Is this right? You really got

35:55
to stop doing it as cold as

35:58
you get really that starts to do. Yeah. That’s what I mean, if you think this is the scary bit to some degree, if you think if you think the Star Wars universe, or whatever they call it, there is only plot lines and potential stuff that has not been explored that is just open to kind of interpretation that if you can generate your own stories, that’s going to be one of the first places where they’re going to want to do it. But then Walt Disney aren’t going to go okay. You can create your own. It’s a bit like those books used to get like wizards, if you want to go down this page there. Yes, that’s what you’re gonna get to, but on a lot grander scale, I suppose.

36:35
Yeah, yeah. So tell us a bit about the new podcast. I know, you widened it from being just now AI and those kinds of things. What’s the decision? The reason to do that, that decision around doing that? And then what kind of things you explore now.

36:47
So I suppose the decision was, and I think I’d hinted a little bit earlier in that we, me and Simon that do the podcast, we were, we kind of start up. There’s all roads lead to the big boys kind of thing. But also there was a lot of gimmicky stuff that was just coming out. There’s just in effect, clicky typing click bait, we’ve done this kind of thing. So we said, Actually, we don’t want to confine ourselves just to the AI piece, because we as I said, there was a lot of repeating of things, and you ended up back at the same point. So we wanted to broaden it out a bit to cover more topics. So one of the things that we were struggling with was a lot of the people we were speaking to in the AI world to things fit it into two categories, real innovators that were doing something and then shysters. He just tried to cash in and there was really no AI can and you can filter them out. But it the problem was basically the whole podcast just on that. It just, it just drains you. So what we wanted to do was expand out our focus. So we’ve expanded it up really to three main areas. So one, if we say over a week period, we would do like normally on a Monday, we do like interviews with interesting people. So we expanded out to just people with stories to tell in that regard. But we also picked up on exciting or interesting new stories that some AI some business or really whatever kind of took off fancy. And then on the if you’d say the Wednesday we talk about more business focus stuff. Again, this was mainly just down to feedback we had from a lot of our listeners, they wanted a bit more variety. So the business talks about picking we get picked guests that talk about new businesses, they’ve started business ideas, we discuss things like that. And then finally, we wrap up the week with like something for the weekend on a Friday where we talked about new things. So we review it’s been it’s pretty can be pretty diverse. It could be like new films TV, it could even be we’ve got the episode going out this week. Well, I think we talked about upcoming what’s been or it’s not so much a review, there’s been what’s been happening in the world of professional darts. So its first range of topics. So that’s where that the really interesting podcast kind of bit ironic, but it’s, uh, we touched upon anything of interest, really,

39:15
I mean, that whole market, the whole podcast market has been a super competitive at the moment of so many, there’s so much content out there and trying to keep it fresh is always the challenge, it seems. And you’re right. There’s a lot of like, large podcasts that are out there that are really using more like traditional media and traditional, like big names, really, as you know, it’s more of a, you know, large media type approach, which you got a headline name that then basically draws people in, right. So and it’s it’s difficult. It’s difficult these days, right? Yeah,

39:43
exactly. You it. The way I look at it is you need to create content, unless you’re, as you say, unless you’re some media star already, and you’ll bring your following into it. You need to have content that ticks the box or gets people to doubt Load. Because otherwise, there’s no point really doing it if no one’s going to listen to it, and you can’t have to follow the trends, but I suppose it’s quite difficult because you’d go to would be like, what are the most popular podcasts in the UK at the moment? Most of them are from people that are famous for something else. And they’re doing something else. So I suppose like the rest is politics. In the it’s one of the largest podcasts in the UK or EU? Would you really be listening to that sort of thing? If these were complete? Nobodies that have never been involved in it? I don’t know. It’s a harder one. And as you say, if you don’t have sorry, if you don’t have that luxury of a media spotlight already, you do have to look at different ways to get people to get bums on seats, shall we call it?

40:43
Yeah, I think also, if you if you’re always looking at the common topic, though, or the news, you end up being quite my favourites ends up being quite you become the lowest common denominator. So you kind of the same as everyone else. So yes, it’s mathematically or looking at statistics is probably the right things to focus on. But you’re not necessarily doing something different. That’s actually interesting. So actually, in actually, the ones I quite enjoy, tend to be the ones that are I read are the ones where people are doing something really well. And doing something that is probably a little bit unique, maybe even things that I would like to do, that I don’t have the time to do, I don’t have the money to do or those kinds of those are the ones I watch, or the ones I’ve listened to, at least anyway. And it’s like, how do you and that’s what a nice niche, isn’t it to be in because we can probably all got something that we can probably offer that’s probably a little bit unique. And it’s maybe the audience’s is big, but it’s a good place to stop. Oh, massively.

41:32
And I suppose that the thing is like, there’s a podcast I listened to it’s called CHART Music. I don’t if you’ve ever come across it. But what they do is an they troll YouTube for old episodes of Top of the Pops. And then they bring in old journalists that work for like enemy, Melody Maker and stuff of that special. Yeah, exactly. And they review the episode, they talked about the artists that were on paper sounds like it’s not going to be very good. But the episodes run for three or four hours long. For how long was Top of the Pops in its heyday probably like 30

42:11
minutes. Exactly.

42:13
So they talk about so what they do is they say, and bear in mind, they go, mainly, most of the episodes are from the 70s 80s. And then you get the old one from like when it was when its death spiral dawn. But you’ll get in but you’ll get like a journalist that works or Melody Maker that’s on there that will say like the last one I listened to was from the mid 90s. And they were talking about like, interview what his experience was interviewing like Oasis because they were one of the guest. But the guy that writes that presents it he writes background knowledge on every artist, they also do things like what was on TV that night. So if you think it must take him, like, where

42:57
do you find that stuff out? When do we what was on TV on the third of March 1975? Maybe there is an archive somewhere? I don’t know.

43:07
So he tells you like what was on? I suppose it was a little bit easier then because you had only four channels or three channels. And some of them are even two channels. I don’t remember but but they tell you like and it brought back memories of things like they mentioned Open University, which then I think anyone under the age of about probably 30 April has remembered on TV Yeah, just remember like guys with beards in like brown jumpers on Saturday mornings on BBC Two. And then the other bit was like TV, like channels closing down for the night. And also channel some of the I don’t remember this as much as you because I was probably at school but like TV channels shutting down in the day like for, which seems killer?

43:50
Yeah, channel 14 used to come on in the evening. Didn’t do the five o’clock news to start with Countdown, didn’t it? Yeah. Yeah, yeah.

43:58
It’s just to forget about this. It’s just like that. It’s like the amount of work that person puts into it. Like, I don’t have a month to write for, like solid days to write an episode for podcasts. It’s nuts. But like you say, when you find those unique sort of stuff, that you you’ll keep listing, that’s a thing, but what I tend to find is that the ones that are done by like larger names, they lose a bit steam, like I suppose they start with a big bang and then it tails off a bit. But I suppose like the rest of politics of that we’re still ticking along a bit, but I think that just might be because they’re making that much money out of it. But it’s kind of Yeah, it’s a weird one.

44:40
I do think that YouTube and podcasting to a lesser extent, but YouTube in particular, it’s just changing TV to a certain extent. I get most of my my news probably from social media. That was a discussion the other day with as a young people I’m not old person, but it’s just like it’s so much quicker. But then also then TV you know outside of Netflix So some of those are things like YouTube and some of the online it’s just so easy to consume. And you get these nice kind of content. Yeah, that you know that you wouldn’t get elsewhere. Right? So nice science content on things. And maybe it doesn’t have the big production values, but it’s pretty good these days is

45:14
brilliant, but also isn’t just the nice providers. If you think like as well, I was looking at when was the was it last week, SpaceX did a launch. And obviously, actually, to be fair to they’re pumping out on I don’t know if they still put it on YouTube, but it’s all on on Twitter, or x or whatever. But like the production values for that is obviously through the roof. But that isn’t going out on terrestrial or subscription TV. But it just gives you a reason to you to watch that sort of content because yeah, it’s just, I suppose the Why bring up the SpaceX stuff is really just, it’s reinvigorated people watching space launches, I know that it wasn’t massively happening that much for a while and then it’s come back with a vengeance. But I remember you only really ever heard about a space launch if it went wrong, which is a sad thing to say. Whereas now on these other media platforms, it seems to invigorate especially watching

46:11
that land is just pretty impressive, isn’t it? Pretty cool. David, I know we’re almost out of time. Anyway, we’ll put links into your podcasts really in the description but David thanks very much for making the time and I’m also a timer we could have chatted it for even longer anyway. I’m gonna go and get go down to the Deathstar canteen now and get get some food for lunch. So

46:31
just watch out there might be some Darth Vader types in there. So thanks very much.

#ArtificiallySpeaking


RO-AR insider newsletter

Receive notifications of new RO-AR content notifications: Also subscribe here - unsubscribe anytime