AI's Hand in the Tech Talent Shuffle: A Close-Up on Job Evolution | Interview with Oliver Sulley, Co-Founder of Edge Tech

Ryan Purvis 01:52:54
Hello and welcome to the digital workspace works Podcast. I'm Ryan Purvis, your host supported by producer Heather Bicknell. In this series, you'll hear stories and opinions from experts in the field story from the frontlines. The problems they face, how they solve them. The areas they're focused on from technology people and processes to the approaches they took that will help you to get to grips with a digital workspace inner workings.

Welcome, Oliver to the digital workspace works podcast, you want to tell us a bit about yourself?

Oliver Sulley 01:53:28
Yeah, thank you so much for having me. So my name is Oliver, Sulley, I run Edge Tech, we're a recruitment business focused within edge technologies, I guess. So anything emerging tech, whether that's artificial intelligence, automation, Intelligent Automation is a hybrid between them and also moving, I guess more into that generative AI side of things as well, which is really become popular over the past six months. And that's, that's our main focus. We work across the board from UK, Europe to US, Canada, and a little bit of Asia. And in terms of the types of areas that we're working with, and then we work with the whole spectrum from developers, project managers through two sales, guys marketing, and C suite. So like, think that we have quite a nice broad overview and oversight of of the industry as a whole at the moment.

Ryan Purvis 01:54:19
That's great. Thanks for Thanks for that. And you're right. I mean, I think AI has exploded, and someone was telling me yesterday, that thing is going to implode just as quickly. Because some of the stuff maybe looks better than it really is. I'm not an expert, really. So I don't know if that's true. We'll see. Maybe do you want tell us a bit about what the digital access means to you.

Oliver Sulley 01:54:36
Yeah, sure. And I think it's, it's one of those things that can mean a whole lot of things to different people, isn't it? I mean, we started the business in 2018. And I think if you look at the digital workspace from 2018, to now, it's shifted so dramatically with everything that happened with COVID. And this sort of thing, from in terms of the actual physical workspace, the majority of roles that we work on now, are hybrid or often fully remote. If you speak to someone now about being in the office five days a week, they'll, they'll gasp, they'll be surprised that we will then talk to you as if you were talking about working practices from 40 years ago. But if you said that to me four years ago, there, and you offered maybe one or two days a week from home, the black hole, they're fairly flexible. But it's the shifts been so, so big, and so sudden, because of necessity, so every company had to move to hybrid or remote, and people realize that it was possible and, and a lot of a lot of time in tech roles and this sort of stuff. They they're sort of like, well, why? Why do I need to go back? What what's the point and I think there's a whole lot of things that can be said on both sides of that of that fence. And I know that I, I can sit on either depending on the role, and depending on who it is, and this sort of stuff. But I think it's it's one of those things where it really depends on the company culture and this sort of thing. And then more to our side of automation. The work not being done necessarily by the humans, the work being done by bots, or overseen by humans, but the majority of the actual boring labor stuff side of it being done by by robots. And again, five, six years ago, it just wasn't a thing. So I think we're at a really exciting time, I think some people are seeing it as a scary time, potentially, for their jobs and this sort of thing. But I think that realistically, it's really exciting. Look, gives us the opportunity to really focus on the stuff that the humans are good at the stuff that the humans are there for, you don't want to you don't want to sit there and be doing winning and clicking and moving a mouse around to copy and paste something five times, you want to do that to actually be able to innovate, and be able to change and use your creativity. And so I think if people look at it as that it's it's all about the mindset, and how you frame it.

Ryan Purvis 01:56:50
Now, do you have the right I think the no one likes change. And especially if it's changed, and you don't really understand how it all works. Which one of the concerns I have with all this as we move very quickly away from something that you could pull apart and look at or dig into the code and look at and understand what's going on, you know, because of the load case of black box, which is that which is the tricky part. But I think I think we have to get away from doing the mundane. And, you know, we had lunch yesterday, and we're talking about how chat GDP has become part of our workflow and what would happen if you switched off now. You'd have to be the thing that you were doing in five minutes, because you got used to doing the writing prompts, which would have taken you maybe a day or two. Now you're gonna have to do that again and there too, because you don't have this thing. So I think we you know, it's not gonna be the toothpaste out of the tube cupboard. So I think we're on that path now.

Oliver Sulley 01:57:46
Yeah, and I agree and I think is you mentioned sort of with it sort of exploding and imploding at the same time. I think it's people understand think that there are limitations to it. And the people who are hyping it up in this sort of stuff, make it make it seem like it's something we said at the moment make it seem like there's anything that you can imagine it can do, which isn't necessarily true. And I think there are things that aren't that aren't helping you. I was on LinkedIn yesterday, and I saw a video of someone playing a robot, a table tennis, and I looked at it and I was like, that's like, and if you scroll enough in the comments, it's actually be like, Oh, someone's just CGI the robot onto this. And it looks like they're really contributing evidence. But if you look at the other 100 comments, all of them are like, Oh, my God, this is how it's going to be in the future. And this is how it is now. And if we've already here, how is it going to be in five years time and, and realistically, you just have to look at like Boston Dynamics and this sort of stuff to know that actual physical robots are nowhere near nowhere near that. And there's still a long way off. And it's, it's one of those things where I think, Black Mirror all of these sorts of TV shows that the majority of people watch, make it or fantasizing it and making it think that it's a lot closer than it is. And we can stick in our little bubble, little bubble in tech and be like, Well, we know what it's good for. But 99% of people are there thinking, Oh, my God, if it can do this, how long is it going to be until it starts really thinking for yourself and taking over and all this sort of stuff? And I think you've still got to remember that it works for you. And it's only doing stuff that works for you. It's not, it's not going mad?

Ryan Purvis 01:59:15
Yeah, it's not it's not smart enough. I mean, it's, it's more efficient, more effective. Yeah. It's only good. It's got the processing power to follow the logical part of that is that's been determined, but it's not thinking for itself. And I think, yeah, for a lot of us, we kind of blur the movies with the stories with the real life because, you know, that's how our brain works. We everything's visual. So it's very easy dots from Terminator two. But, or even War of the Worlds kind of, at all. What was that movie back in? The 80s 90s? It's goint yo irritate me now,

Oliver Sulley 01:59:57
you're pushing me on Asia?

Ryan Purvis 01:59:59
The guy, the guy that? Oh, you probably didn't even know he hacks? Like wargames? That's the one. You know, where he's like, shall we play a game with this? Yeah. You're supposed to AI? I don't think we're there yet. Although there was there was some drone footage of an AI supposedly flying a drone and shooting a human but you don't know if it's real or not or.

Oliver Sulley 02:00:22
And that's that's the problem with having that. I think it's, it's a lot of stories. A lot of I've seen this, but as just as well as the AI can do that stuff. It can also make that stuff up. And I think that's, as you know, there's all of the stuff going around at the moment in terms of fabricating things in generation. So I think that's more of it. I did see, I did, I did have a see someone tweeting saying that the future that I was looking for wasn't where I'm going to be doing all the manual labor, and then the computer is going to be the ones writing the poetry and making the art, which is sort of a case in the moment. But yeah,

Ryan Purvis 02:00:51
No, you're right. And so the thing that I was talking about yesterday, which I think is as long as same lines, is we still have that point where it still takes a lot of effort to get the value out of an important narrow AI or a very, very narrow AI. At some point, yes, it'll become the other way around. But like you said, you know, the technology, people know what's coming in there, and they're working on it. But it's not JGBs, probably the first one is actually crossed the chasm to say this is something the end user could play with you, then copilot, which is coming through the macro stack can definitely see that adding value, and we were talking yesterday about how great it would be to just not have to go and create their PowerPoint deck. Because you can tell Copilot to go create the PowerPoint deck.

Oliver Sulley 02:01:38
I'm gonna be like a next generation Clippy type thing.

Ryan Purvis 02:01:43
Let's open it up. But let's so useful. But even you know, if you think about code development, as much as people said arts and the developers, I don't think that's gonna be the case, I think you're still gonna have you're gonna it's going to be helpful to generate some code, but someone's gonna have to make sure that code actually works. And makes sense. And, you know,

Oliver Sulley 02:02:01
I think it's, it's one of those things. I there was a stat that I saw a few years ago. And it was something along the lines of 80% of primary elementary school kids, when they have jobs, those jobs don't exist now. So when they start, so how can you teach? How can you train them up for it? If you look at what looks like a viable career at the moment, another viable job is a prompt engineer. Someone who can who can write the prompts, who can who can knows how to work out GPT Bard, whatever it is, but knows how to work it really accurately. So it comes out with the right thing every time. If I mentioned that to you 12 months ago, you'd be like, I don't know what you're talking about. It's moving so quickly at the moment, but still, it's still a type of tech role. And it's a role that people within tech are going to move into another role within tech. So within that, it just means that there's no real way that you can understand what's going to happen and we can all prophesize about it. We can all try and make our predictions but whatever or anyone says they're still just shooting into the dark, I think.

Ryan Purvis 02:02:57
Yeah, and it's funny, you mentioned prop engineers as well, because we were talking about that as well, yesterday. And I don't know what I was doing, I think I was doing something where I needed to generate a job description or something. And I found an example of a prompt engineer actually writing that check GBP. And I actually hadn't even thought about some of the stuff that they put in there. Things like to obviously get the instruction like pretend you're going to be xy and z to write this piece of text, like we all know that one. But then there was things like, make sure this is no longer than 260 characters. If this is longer than 260 characters, rewrite the text, make sure this is unbiased, make sure this is unbiased. So I'm biased from a racial point of view, make sure this is generic, you know, there was all these little things that I was like, hmm. Now, now I understand what probably engineer would do. Now, I think that would be one of those roles, that would probably be quickly become a role. And it quickly be absorbed into a skill set you just have? Yeah, it's one of those weird things like who would exist for a period of time and then just be absorbed and be gone?

Oliver Sulley 02:04:01
Sure. No, no, I get what you mean, there, I think it's gonna be super valuable, for a period of time, but also, it's gonna be one of those things where people pretty quickly gonna be like, boy, I don't want to pay for someone to just do this. Like it like Why Why am I doing that it was in a, it's gonna be a real useful skill, but a business isn't gonna want to pay people specifically just to do that. Because then they're going to add a certain amount of value, but they're, they're basically just adding any value that chat GPT can. They're not necessarily adding their own, and they have to get the other expertise from someone else. So whether it's that or whether it's just the AI becomes more advanced, so you don't need them. I think it's one of those situations where it's going to be a flash in the pan, but it will be a real big flash.

Ryan Purvis 02:04:39
Yeah, yeah. It's a skill set. And I think, you know, maybe your typical business analyst or your system analyst would probably be doing that stuff or your data analyst. Because they will understand the context. Because that's, I think that pick the part that AI will never understand is what's the context question? Because you can build your own prompt. I mean, I was doing it with this sort of notion where I'd like a few columns in a row. And then I would just input the one that the input, which would be my actual question, and then the prompts would just be me selecting options to build the product. I don't think it's that far away from your automated. Yeah,

Oliver Sulley 02:05:14
I agree in. And it's one of those situations. So we work really heavily within robotic process automation, which is, I guess, a lower level of that. So nothing to do with manual, manual labor. I know people get confused with the term robotic, but automating invoices, automating anything that you can point and click, it will do that. And I remember I was speaking to a head of within it, and they were talking about their training academy. And one of the first things they train people on is how you define that process. And they did it by asking people, how would you make a cup of tea. And when they say, Well, they'd like, I put water in a kettle, or boil the kettle, I put the water in the tea, I put the tea bag in there. And then when it's done to my level, I take the tea bag out if you need sugar, or milk would do that. And it's like, say seven steps, and then they pour out of a cup, pour out the bag, 45 steps. And then like this is how many steps are in that process of me of boiling the kettle? Do you check if the power is on if the power is not on? What do you do all of these sorts of things that you don't think of? Because it's just human nature? You're just like, well, I know all of those things already. And it's realizing that there's so many steps to everything, that you really need to start from ground zero to work towards that. And it's exactly the same with this. It's working out. What sort of things do I need to know, which I don't think I need to know? Because in my head, they're obvious.

Ryan Purvis 02:06:31
Yeah, yeah, you're right. And to be honest, my impression of rpa, to some extent, is it just moving the mess from one place to another place? Because I mean, there's there's value in rpa, they're gonna be wrong, you know, to go and have a whole bunch of agents that are pulling in people filling in stuff in the same manner that you can automate. If you do that, my background is more integration. So I would be like, Why do they want to integrate it? Or, you know, you've got two separate systems that do kind of the same thing between them together and have a golden source. But those things take time. So my bias, position perception of RPA in the value has always been, you can train the RPA agent with using, you know, what's the access anyway, I think he's the one or one of those companies to do that stuff in a shorter period of time, they will take to do the integration. So you know, some products don't didn't have API's for a long time, especially if it's like a mainframe base. And then while you get that running, you could then do the integration if it made sense, or whatever it is. So I think there's there's, you know, sort of horses for courses, conversation there. But that that failure logic that you mentioned, I think people don't think about that at all. I mean, I'm trying to get my house fairly smart at the moment. And so it's like smart light switches and all that stuff. And one of the things that I have which sounds really stupid, is I want when I turn off some certain plugs at night, because they know they they've created a hammer and the hammer irritates me when I when we were in the room, so they every morning at 06:30, they turn back on again. And the first thing I want to have happen when that thing turns on that play because the dice machine was stopped. And I've been trying to find a little button to press the ice machine. And the problem I'm having is for some reason, when we turn that switch on, that button doesn't switch on. I cannot figure out what with this little push up in Vegas. And I cannot figure out what that is. And then I've talked to support it or not, it's very interesting on the Wi Fi, but it's almost like it resets itself. Now that would be where you'd want to so it sort of RPA bit of instruction set to go okay. In this case, you need to do the reset, to reload the instructions. So you know, to press the button at 631. Yeah, I think it's the buggy. I got I got the Monday morning, built one after the big hangover. But that, you know, it's going to the trying to do these simple, simple simple things, you start realizing how like bad Wi Fi, bad power, you know, the human interaction with like, my son knocked off the button, so didn't work. There's there's no way to know that. So like, if we're, if I'm like, downstairs, I wouldn't have known that, you know, these silly things. But I think it's that whole process that we take for granted. Because we ignore the things that we do often become because it's routine becomes conscious, unconscious competence, which is where I think AIs are great to call those things out as well. So they can almost deconstruct that stuff. You know, so like, if you go write a scent operating procedure, and you ask chat GPT to give it to you, it'll give you a lot of steps. Yeah, because it's analyzed so much content that tells you all these steps. And you might go okay, well, there's 29 steps that we only need 15. But you probably wouldn't have come up with 15, you'd probably come up with seven, because that's what you were thinking about at the time. And it only after experimenting and iterating, would you come up with the other five or, or eight to get to 15?

Oliver Sulley 02:09:57
Exactly. Now, I really mean, when you mentioned that RPA side of things. I was speaking to a guy last week. And he put it fairly clearly, which was people are saying oh RPA is just a bandage and this sort of thing. But if I've got a cut, I want a bandage. Yeah, I want a band aid, I want a band aid, I don't need a big old, I don't need a big old huge monitor on my arm, I just want like a little band aid to send this going to fix it, I just need something that's going to sort of for the time being, it doesn't need to be a whole big thing. But it needs to be something. And it's something that can can fix it in that in between. And as you say that works. And then whether people move on whether people use integration, whether people maybe use more of a whole Intelligent Automation is a solution, something that's got a bit more more to it. So that's doing the process mining the process, analyzing, and then bringing it in board, maybe got some intelligent document processing, and then moves on to chatbots or anything like that. So you can interact with it. And I think that's it's the virtual assistant side of things, which is within business especially, that's useful, where it's like, oh, this is flagged up this needs a human's attention. That's not the times. And that's when it pings up to you. And you can look it up like oh, yeah, that is right, and you send it back to it. Or if you even have like internal training, when you're messing me like, how do I do this. And then it can bring all of these things up. I think those sorts of areas. They're not other people's jobs, necessarily. They're just ways that make your life easier.

Ryan Purvis 02:11:18
But you set in other people's jobs. But this is the irony is that becomes someone else's job, because every time someone else needs help, and I'm not saying there's anything wrong with helping people, but every time someone else reaches out to get help for you for something that is simple. Yeah. You taking away someone else from the work that we're doing. And is there's actually a cost to that we've got a report. Now, when I, when I started my first career a long time ago, we had a thing, Google it. Like if you don't know how to do it, good luck. I'm asked me you Google, you know, anything Google that, then you can't find out what to do, then you come talk to me. And I get frustrated with people now even today, they come to me, they say I don't know how to do this thing. And as you say, Well, have you figured it out at Google? Have you asked if you know that none of those things? Well, they know come to me, come, go. Go. Because you need to try otherwise, you'll never learn. And I think this is the the concern that I have. And I was talking about this yesterday as well, where you're gonna go and have people generating code, then go and build spreadsheets that are coming out from AI and have no idea how it was put together. And they're just gonna go Oh, that was generated, here's the answer. You don't know. Now you have nanogenerator you actually have to be double critical of it. Because you don't know where it came from. So you have to ask questions of it so that you are comfortable with it. And then you also have to give it that last sort of sense check because you're you're closer to it now. So you have to think so. I think there's there's there are risks to it in the sense that people are going to forget quickly how deeply it already has to be something before they pass it on to somebody else.

Oliver Sulley 02:12:39
Yeah. And I think on the first thing, people in my office hate me, because there's a website called Let me Google that for you. And I put in, and then I'll send them the link, I got this answered for you. And then they open it and get some every time. But yeah, I think I think you're right. And I think that's why as you say, there still needs to be those technical skills and technically, expertise overseeing it. I think the main concern isn't in the next 5 10 15 20 years, I think that the main concern is in 30 years time when those people aren't working anymore. When when you sit down with your kids and this sort of stuff, and you point at the save icon on word. Yeah. And they look at it now. Like, that's the Save icon. I don't know what else that is. It's a similar sort of thing. They don't know what a floppy disk is, like, these people won't know what life is like without track GPT without any of these things that they can rely on. And then that's when you've got to hope that they've got some guidelines or safeguarding in place, because otherwise, you're just gonna get to a point where you've got the Prime Minister asking charge up what to do with the budget. I mean, it probably couldn't really wise, probably won't be a bad thing.

Ryan Purvis 02:13:57
But well, you know, one of the guys said, yes, they used to be to work at that monetization on his mortgage. Which, you know, and he said, It wasn't so much that it worked out for him, it gave him the, the macro or macro, the VBA script to put into Excel to do it for him. And I would have done the same thing. But I would have still probably gone to the bank website, or the you know, where it normally has gone and put the numbers and just check that they were kind of the same. Yeah, but you need to have that validation, I think, at least at least now. I mean, I think we're a long way from sort of Star Trek world where you say computer do this, and it does it and you're like, that's 100% accurate?

Oliver Sulley 02:14:37
Yeah, and I think it's, it's always, the next advancement always seems so alien to us. But it also did 30 years ago, yeah. When, when a advancement came around 20 years ago, when it was the first iPhone, you're like, well, it's only gonna be a matter of time until, like, actual full augmented reality is a thing. And that's still not a thing. But you could have said that 20 years ago, and people were like, oh, it's only a few years off with like Google lens and this sort of stuff, but nothing's ever really come of it, because it still doesn't really work. And there's all of the concepts of it. And they're all these concepts 50 years ago, and they're like, Oh, well, it was back to the future. 2008 2013, something like that the future one. And that people like that's probably realistic, that's probably not gonna happen. And we're a lot closer to how we still were then than we are. It's the rate of change is, is going exponentially. I'm not saying it's not, but I still think people are expecting a lot more than is have a tendency of years to come.

Ryan Purvis 02:15:33
Yeah, and then expect exponential change. What's interesting about that is, for those that are in the staff, it's going really quickly. But those are not in it, get us talking about, you know, afterward muscle bag will relax the retirement capital of South Africa. You know, these are people that are still struggling with using an iPhone and WhatsApp. You know, the concept of a cat GDP, you can't even we can't even explain that to a lot of them. You know, there's people that understand no follow up. But there's lots. And when you look at some of the other things like using a CRM, or, you know, doing online banking, they haven't got that stuff. So now this concept of an AI potentially getting involved in going through your transactions, and, you know, doing your investing for you, and all that kind of stuff. I mean, I think that's a fear factor that that's magnitude too much for a lot of people, I think the adoption, but that's what's happening, I kind of agreed that there'll be a fizzle out at some point, because it'll become people, like people get used to it, like very much talking on teams and zoom, everyone's gotten used to it down FaceTime and WhatsApp. And then there'll be this suddenly, like, Oh, my God, we've got this copilot thing. And then we'll use it. And then there'll be another thing they look for in the background like AR and VR has not has not materialized. Yeah. You know, it's we go, that's become technically viable for a long time, but it just hasn't materialized.

Oliver Sulley 02:16:58
Yeah. And it, it works in those niche areas. I've seen it working in engineering and this sort of stuff where I can spot parts for you and help you build things and this sort of stuff where it's great. Yeah, but it's not day to day use. And it's not to say use for the majority of people. I think you're I mean, I was in the bank yesterday, and there was a guy who he was an elderly gentleman, and he was trying to get cash out. And they were like, Oh, you can use cash receipt. And he's like, Oh, no, I can't because my card doesn't have a pin. Just refuse to believe this card had a pin was I know, mine doesn't because I don't know. And it's one of those things. It's like, well, I know every card has a PIN, but he doesn't. And it's not. It's just because he's never been taught. It's never grown up with it. He's just happened so late in his life for him, that he hasn't felt the need to take it. I guess the question is, with more and more people being digital natives, is it going to are people going to be able to pick this up a lot quicker in 20 years time when realistically the majority of the population are going to have been in a relative internet that era, to the point that especially in the workforce, most people in the workforce within 20 years, are all going to have grown up with the internet, they're going to have grown up, I still remember what it was like, when I was younger, and Google wasn't a thing. And there was text message service, that you could text and you'd ask your questions, you pay like two pounds a text, and it would respond back to you, and this sort of thing. And that felt that felt amazing at the time. And then Google came out to answer all of these things. And that was unbelievable. But for people really any younger than me, that's not a thing. They've always had Google. They've always had any answer to that sort of their fingertips. And so does this feel just more like an extension of that to them so that it's going to integrate fairly easily? I don't know. You're probably have to ask them.

Ryan Purvis 02:18:36
No sure. I mean, you know, I've got two young kids and we see it, but then, I mean, my daughter's two and a half. And she walks up to the TV, she presses the screen, because that's what she expects. Get on my son's five. And I give him the two TV remote to change the channel. Doesn't want to use a TV remote. He wants to ask Alexa to change the TV. Yeah, that's that's their expectation. They don't have any reason not to think it's gonna make their own car works with Alexa. So they want music they shot in the back. But Alexa, play the song. I don't even try the radio anymore. There we go.

Oliver Sulley 02:19:12
But you don't want to hear Baby shark plan.

Ryan Purvis 02:19:15
Oh, no, no, no, please, no here. And this is the thing that is always made me and maybe this is when you get older, you look back at the younger generations. And you think, Oh, they don't know, like we used to know it. But I think people need to, like I don't understand how large language model works. Personally, right. So I've been I every Friday, it's been two hours researching something. So my Friday this week will be on large language models. Because in and the way I've figured these things out that they need to be done as I talk to a whole lot of people in a week. And as a caregiver concept comes up a few times, and that's what the research is about. So the people have spoken, as we've mentioned, building LMs. So now I'm gonna understand how LM works. I mean, conceptually understand that we're sure but I actually want to now understand what if you wanted to build one? How would you build one because it should be something now, that our day say, you could probably go to Azure, and press a button and get a service that is your annual in? What it needs is data? Yeah, I

Oliver Sulley 02:20:12
think there was, obviously chat GPT first started is that open as open source, and then it got taken away from it. And I think one of those things, I don't think it's necessarily hard to do. But I think it's just the cost of it. Obviously, the cost of all of the processing power and server space, and all this sort of stuff is just now so out of the hands and out of bounds for open source. And the hobbyists and this sort of stuff that is, it's just going to end up being a world of cooperation trying to fight for each other. But but it's odd, isn't it, because everyone talks about charging up now No, chat GPT is very much the noun for it. Like, like jacuzzi, and like vacuum and Hoover. Like you just use the company name to cover all of it. But if you think if you can name me one company that did the first one that is still around and the most successful in whatever technology, I'd be surprised, because what friends reunited MySpace, those sorts of things. I mean, Nokia, Nokia, for iPhone, for phones, and even Blackberry, these sorts of things. All of the first movers in every single one, are all now gone. Pretty much. There's not really many that are still around. And it might be the one that defies the rule, because it's so well known. And because it's so but you'd have said that about every single one, wouldn't you? I think when for the first couple of years when they've come out?

Ryan Purvis 02:21:34
Well, I mean, I think to be fair, on the Chat one. And I think I think because I mean, it's been on my screen next to me, it's so ubiquitous. I don't consider it bad, which is bad. That's on my list as well. But the thing is, because of the partnerships, I mean, the Microsoft, you know, 14 billion in their Incorporated. So I think what will happen is the chance that you'll be brandable will be like Coke, it'll always be there. Yeah, but but you'll just find it in everything to be white labeled it Yeah, you just don't have you. And, you know, I saw your friend yesterday he's got a specific company and he runs and they do a specific LLM. They actually might be unique for one, and then it won't be unique anymore. Yeah. You know, it'll be eaten by the ones. But then I think also in some respects, it'll be good to have different algorithms that are specialized and useless computer base and Oh, got capacity to the point that you can get her on a Raspberry Pi, you know, the key Bolton, you know, that's that's the interesting thing for me is you sort of sort of see how Nvidia and whatever and TTM I think it is, what the financial reports are talking about the kind of chips they're building. That's the that also are exponentially making. The ability to process AI will deliver AI narrow AI capabilities devilish more efficient. that you're even though even the iPhones that are coming have already got some AI capability built into them this sort of marinara stuff, you know that the aliens might go away in some respects because it becomes a distributed lamb. Or, again, I don't know how to work. So I'm kind of guessing by I'm guessing are kind of. But that's that's the exciting thing about I think we all will see massive improvements, and then I think it'll lull down and then we'll see another massive thing and I think one of those things is going to be the deep fake thing. I think that's going to become so prevalent, then it's going to be a huge problem. And then next we'll need AI fighting AI sort of to authenticate and qualified and confirm that actually, that was Joe Bloggs walking down the road of the machine.

Oliver Sulley 02:23:38
Yeah, I agree with you. And I think one positive that could potentially come from it, if you can call it a positive is maybe just people being a bit more skeptical about everything they read, and everything they see because people take, he will take everything just as truth. Even as I said that that video of that robot that was clearly CGI, everyone just sees it once and like what are some video or some pictures, that must be true. So maybe at least because you've been in tech long enough, I've been in tech long enough to know that photo shops existed for ages, and all these sorts of stuff. But if you as you say, if you're showing this to the majority of people, something that's a half decent Photoshop, no one's gonna know. And so at least maybe this will the positive will be it will just make people a bit more critical about things they see online, whether it's articles, whether it's videos, whether it's pictures, it'll mean that people will have to use a bit more skepticism and use their own critical thinking a bit more to actually work out if things are real. And maybe we can see that as a positive. I guess that comes out of this.

Ryan Purvis 02:24:33
Yeah, no, sure. Sure. I mean, I think there's it's kind of a wait and see, it's a response to see what what actually materializes. What do you find in the market? I mean, are you getting? Are you seeing roles that are very AI specific, like a chief AI role or, or something like that?

Oliver Sulley 02:24:50
It's I think, people like to hire what they know. Yeah. So I think also, if you naturally if you put a word for hype word, in front of any job title, you suddenly added 20% on to the salary when it's the same thing. So we've seen it over the past years with AI with rpa, with chatbots, all this sort of stuff. As soon as you put that word in, you're like, Well, I want 20 grand more now. So people are trying to avoid it. Data science, we're seeing a lot of data science, hiring more and more data science, hiring, machine learning this sort of stuff, I think is is what people are sticking to and but you're seeing people on that on job descriptions being like you need experience in chat GPT. And it's like what experience you always have. It's been out, it's been out for six months. It's what November last year was when 3.5 first came out for the public. It's been out for not even a year, you're asking me for experience with it. No company has got any guidelines about what's being used. So at the moment, everyone's just flying by the seat of the pants, doing whatever they want, completely unregulated. So yeah, I can I can show you what I've done. But there's no framework around it. It's just been me playing around with it. And when people like they must have experience with it's like, but what what part of it, especially if they're not necessarily doing the product side of things, they're just doing other stuff, but they want to integrate it with it. And I think people need to realize that, while the technology is great, and the technology is mainly new, the way you're interacting with it probably isn't that new than the way that you're incorporated into the business still has to be fairly safeguarded, still has to be fairly ring fenced in terms of how you're laying it access everything, and what is accessing, and, and all of this sort of stuff. And then at that point is, as you say, it's just an integration job.

Ryan Purvis 02:26:24
Ya know, and, you know, all of these things, I mean, always laughing, it's like you need experience in this thing. It's like, you know, I was an entry level role, but you'd have five years experience in the role before you entered. It doesn't make sense. I mean, I think what I'm hearing in the market is very much people are putting in policies, procedures to limit the use of these tools. And they are prescribing that if you do use the tools, it has to be via the API, not to the sort of chatter, open AI, intercom or elation, or PBM contract, and you're definitely turning off the ability for chat GPT to use your data to learn. So I think those are good things. And I've heard anecdotally of sort of AI roles being created, and then disappear and then just not created and then basically giving the responsibility to somebody else because of that inflationary aspect to it. And I don't think there's many people that can really say that are experts on AI because it's such a big broad topic. I'll be worried if someone said they were. Yeah, yeah, no, exactly. It because I mean, just being for example, computer vision. Skilled is very different to say, NLP skilled. So I think it'd be interesting to see how people how the market plays out in some respects.

Oliver Sulley 02:27:45
Yeah, I think it's, I think it's almost a situation with a lot of these big companies of Keeping Up with the Joneses effect. They see that their competitors done something with AI. It's got a big Spread the newspaper that they're competitors doing something with open AI or or the pinky partnership. And then they're like whether their share price has gone up. So what can we do? Like, what, what you want to do is I don't care. As long as it's got AI and I don't care, we just need to do something with it. And then that's when you say these sorts of roles come out, and then the person sits there, like what you want to do. And they're like, we don't really know. That's why That's why it fizzles.

Ryan Purvis 02:28:15
Yeah. Is that valid? Or is that come driven thing to? And this is the thing. I mean, we were building a piece of functionality and actually asked the customer, would you be interested in this with that, like, check GPT flavor to it in the sense of you're saying in the situation, we could repurpose it, you know, make it more professional and nicer. And you actually said don't, because the minute you mentioned that, that thing you'll you'll be at, we won't be interested because you don't want that thing. You know, because it's, it's we're not at that stage yet. So I think there's a for those that don't have the regulatory requirements, you can go and use these tools. But those that do is a lot that has to be written before they'll even touch it. So you don't need to be cutting edge. You just need to be comfortable with that, I guess,

Oliver Sulley 02:28:58
just needs to work to make sound that works and is what they want. Exactly.

Ryan Purvis 02:29:02
Exactly. Great. I mean, it's a great conversation. Is there anything you want to do with anybody else to get in contact with you?

Oliver Sulley 02:29:08
Yeah, yeah, sure. Obviously, my contact details will be below. But if you're ever looking to find out more about the emerging tech space or anything like that, we've got a few events throughout the year that we always post about, feel free to connect with myself, Ollie Sulley on on LinkedIn or follow reg tech, and we'll keep you up to date on anything you do. But it's great having you on it's been going on. Thanks so much.

Ryan Purvis 02:29:27
Thank you for listening to today's episode. Heather Bicknell as our producer and editor. Thank you, Heather. For your hard work on this episode. Please subscribe to the series and rate us on iTunes or the Google Play Store. Follow us on Twitter at the DWW podcast. The show notes and transcripts will be available on the website www.digital.workspace.works. Please also visit our website www.digital.workspace.works and subscribe to our newsletter. And lastly, if you found this episode useful, please share with your friends or colleagues.

Transcribed by https://otter.ai