رونوشت کامل Creatives/Startups: The Old Way, The New Way, and the AI-Powered In-Between with Doug Ayres

00:06:52 It's interesting, isn't it? I mean, I'm standing in both camps because on one hand I think...which is not like me, usually I'm like black or white, right. But there's a reason why I'm in both camps. So I'm all for and always have been for the business that's future proofing itself and driving itself forward. So from that perspective, Dave, I agree. You want to see those people with those skills, like there's a massive tick in the box, right? But here's the other thing, and I've talked to other guests about this before as well, is that I think the challenge is how do we ensure that personal/interpersonal skills and a whole lot of other things are still being maintained. Cause we've already seen it with social media. Kids are not having conversations the way that we did or they're not out doing a lot of the things that we used to do because they're sitting on their phones or they're interacting in different ways than we originally did. And I think that's an issue. So that's why I've got a problem with it. All for skills being upgraded and looking for those skills when we're hiring, and them showing those skills, but what if there was part of it was done like that, and part of it was like there's still a bit where I need you to put your thing in there.That's where I guess we're in the middle of the interview. And hopefully it's a face to face interview. See, because I don't trust CVs for example, as the thing that's going to make me hire people. I've, I'm famous for hiring people going, "oh, I don't give a fuck about your CV...talk to me, answer me..." and the question that I've always asked, which has always been able to tell me usually whether I want to hire them or not, is like, "Who...who inspires you and why?" And there's a something in there. And I don't want a chatbot answer. No offense to chatbots, but I don't want an answer from AI, I want that person to give me their heart and soul and give me the 'who they are inside'. So that's kind of why I'm sitting on both fences, which is a bit weird because I'm not used to doing that. David Brown (Guest/Host of Startups WithAI) 00:09:00 But there's always, you know, there's, there's that, you know, famous expression which is 'you hire for fit and you train for skills'. And I 100% really believe that. And all of the most...I've hired hundreds of people through my career and the most successful ones were the people that I hired because I felt like they were the right person, even if they didn't have technically the skills or the background that I thought that they wanted. I'll do very quickly one example of that and then I want to hand over Lena for you to bring your story. But I interviewed this guy and as he was walking through the office when the interview was finished, there was some grapes sitting on the table on the desk of one of the people. And as he walked by, he literally took grapes off that person's desk. Didn't ask, just literally grabbed a grape and ate the grape as he was walking by and the guy that was sitting there was like, "what the...like, what, what are you doing?" And I was like, "that's balls. And I love that." I loved the fact that he did that and he just had the confidence to do that. And I was like, "I'm hiring that guy". And he worked for that company...he's, he's passed on, he actually, he's actually died of cancer eventually, but he worked at that company for nearly 20 years. Lena Robinson (Host of Creatives WithAI) 00:10:19 Oh, wow. David Brown (Guest/Host of Startups WithAI) 00:10:20 Through, you know, multiple sales, he grew his career. He ended up running an international team, like in New York and everything. And it was amazing. Lena Robinson (Host of Creatives WithAI) 00:10:30 Isn't that funny? I would have done the complete opposite. I would have gone "fucking rude. No, you're out." David Brown (Guest/Host of Startups WithAI) 00:10:35 But see, and I wasn't really gonna hire him. I wasn't that impressed until he did that. And then I was like, "I love this guy that he had the balls to do that". So, anyway, there you go, Lena, what's your story that you're bringing? Lena Robinson (Host of Creatives WithAI) 00:10:49 So I'm doing a little bit of a weird thing. Sorry. I'm opening my iPad so I can look at the article. Cause my brain is not working particularly well. 52 years old, brain fog, whatever. So mine is about something I read from New Scientist, which is an amazing publication: "Is Keir Starmer being advised by AI? The UK government won't tell us." So apparently that's the headline. They have something called...they've developed it inside of the governmental departments and it's called Redbox. And the New Scientist has asked for data and everything. Only two departments said that they were going to give any data. And I think some of them have said they delete it automatically. The prompts don't get saved. And I think half of it's probably bullshit. But anyway, what it made me think about...cause there was two areas I think that they were most concerned about when it sort of delved down into it, which was, and this is not new to us in this conversation, 'accuracy' of content and 'bias', and how is it being...AI used? How is AI being used by the different departments? But also, you know, and they did answer that 3,000 people within the group that looks after Keir Starmer specifically are using it. What they wouldn't get drilled down to is how specific it is. But it made me think. A question which I want to pose to the two of you is if it's advising Keir Starmer, what the fuck would it do if it advised Putin or Trump? Or is it already advising Trump and that's why he's acting like such a dick. I don't know. Throwing that out to you. David Brown (Guest/Host of Startups WithAI) 00:12:46 I'll jump in first. I think we'd probably be better off if it was advising Putin and Trump. Trump, frankly, because again, the way it works is it, it works on averages, right? And it's a predictive model, so

it's...it's constantly pushing everything, I think, towards the middle. And...and what it ends up doing is it's...it's presenting the thing that's the most likely or the kind of the average, and I'm sure Doug will probably correct me on this because he's...he's way more, he's way more technical than me. But basically that's what it's doing and it's predicting what's the most likely thing based on all the data. So if you put all the data of the entire world in there, it's averaging all of that out. So it's taking away all the extremes. So frankly, it's probably better. It would probably give better advice and it would keep the edges that, you know, the extreme edges on either side, out. So first of all, there's that. Second of all. I would argue that a lot of the tools that the government are probably using are not what strictly engineers would call 'AI'. It's more machine learning than AI. And those machine learning tools, they've been using for 30 years or 40 years or 50 years, ever since the invention of the computer, they've been doing sort of machine learning and statistical analysis. You know, I know certainly since the 90s, they've been using, a lot of organisations have been using machine learning, which today everybody's calling AI because it's the sexy thing that everybody wants to talk about. It's not actually AI. It's...it's a different...it's a different set of mathematics that work in the background. But the salespeople like to call it AI and the engineers will call it machine learning. And that's been in place for years and years and years. And not too long ago I worked in public sector, so I do have to be a little bit careful about what I say here, but I saw a presentation by the lady who ran the data analysis team at Number 10. Now, this was pre-Kier Starmer, but it has, that team has no bearing on who's the Prime Minister, it's the team that's there that supports whoever is the Prime Minister. But she was saying that some of the new, the newer tools and the newer machine learning that they had available meant that they could do analysis that used to take a team of four or five people three or four days, sometimes a week. They could literally set it up, they could run it over lunchtime. They could literally one person could set it up, set it off, go off and get some lunch and come back and they could have the answer by the time they came back to the office. And it just massively reduced the amount of time. And what they did is in those government tools, instead of it being a black box, like a lot of the, you know, if you just use ChatGPT, you have no idea what it's doing in the back.But the tools that the government used specifically export the instructions and the logic that they used to get the data that they, the answer that they gave you, so they can cross-check it if they need to...to make sure specifically that the maths and everything was correct in the background. So this is something that's not talked about publicly. And

I don't think that's any, you know, that's not any government. I didn't have to sign like, you know, it's not classified information or anything, but I do think it's an interesting little wrinkle that most people don't realise. So I suspect that there are a lot more controls in place than people realise. So it's probably fine. Lena Robinson (Host of Creatives WithAI) 00:16:22 Doug, want to wade in on this? I can see you scribbling notes. Doug Ayres (Guest) 00:16:27 I've been scribbling away. I call BS on the advising line. I don't, I don't think that's happening. What I like to think, although you can...know you have to take these things with a pinch of salt with...with any government, is that they've got, as David alluded to, the controls and the security around what it's doing and when. So that, like at the foundations of it, when we're talking about machine learning, natural language processing, etc. etc. You've got a data set that is its corpus, it's knowledge base and you know, garbage in, garbage out. So let's hope the data set that it's querying is competent, and right, and truthful, and all those good things, whatever the topic is. And that therefore becomes a knowledge retrieval job and how well it can do that on a very large data set. I think, yeah, things have moved a lot in the AI space. I think a really interesting area might talk about a bit more later is around prompt engineering. And this calls into question who's, you know, what is that prompt? Is Keri sitting there? You know, what's he asking? Like that's just on a really naughty basic level. But these systems get better and better if they have a really sophisticated prompt, or an engineer approach to prompts, prompting the system. And it's much more than "Can you tell me what the weather's like in Mallorca today?".It's like these system prompts are long, you know, and recently ChatGPT, for example, I think Anthropic and you know, Gemini and the rest of them are all sort of, you know, containing. This is the control around the reasoning. And so, you know, you can prompt it to say "look, don't make shit up, just give me facts back". I mean, give me real data. And the model will adhere to that. So they're trying to absolve themselves of hallucination. But the knowledge retrieval and the prompt and the data set is really, really important. And yeah, that's my view on it. I call BS on advising, but I think that it could be, you know, the government could be using it for, you know, knowledge retrieval, and at serious speed. David Brown (Guest/Host of Startups WithAI) 00:18:52 Hello, everyone. David here. Just a quick note to say that we had some technical issues with the software we use to record remotely, so we ended up stopping the conversation and switching over to Zoom. So I apologise for the difference in the quality if it sounds any different, and it definitely looks different if you're watching it. So, yeah, just wanted to jump in and let you know what happened, but we picked up the conversation where we left off and we carry on. Lena Robinson (Host of Creatives WithAI) 00:19:18 Okay, well, that was an interesting area of conversation, wasn't it? So, Doug, what's your article? Doug Ayres (Guest) 00:19:26 Really quite geeky, to be honest. I'm a bit embarrassed. There's something being talked about a lot in tech news, but also quite a lot on YouTube. If you...if the algorithm is checking what you look at around AI and it's called the Model Context Protocol, which is 'MCP' for short, there's been quite a lot of fanfare about it. It's basically how you can...it's, well, it's been called a USB for models and tools, and functions, and how you can intertwine LLMs into actions on...mostly in the examples I've seen are around on your own device, your own laptop. The importance of this is that you can really get Agentic AI working. You know, it's not like go to ChatGPT.com ask a question, get a response, or go to something like Replit and create a website and push it to a database and yada yada yada. If anyone hasn't heard of Replit can check out, we'll stick it in the... Lena Robinson (Host of Creatives WithAI) 00:20:34 Show notes for you. Doug Ayres (Guest) 00:20:35 No coding. David Brown (Guest/Host of Startups WithAI) 00:20:35 You really are in the way of stuff. Doug Ayres (Guest) 00:20:39 So an example, a really naughty example of this would be you set up an MCP server on your laptop. I mean, caveat here, that's slightly technical, but I think in the next three, six months there'll be wrappers around that protocol so that anyone can do it. Just like downloading WhatsApp. So if you've got an MTP server running on your laptop, you can give it access to your local credentials and also sort of your online credentials. So an example for me might be every so often I go into London, so I don't have a season ticket. I buy a ticket on the day and it's, I mean it's a very low level first world problem faff of going into the app, you know, choosing the time, knowing that I want to go off peak and choose a time, you know, again on the way back and yada yada, yada, you know, it's booking a train ticket, and then I have to go and do the Apple Pay or whatever with, with this protocol you could basically give it access to your, you know, your card details and it's all secure because it's the same process as Paypal by, you know, I say also all secure. If anyone looking at this going hang on a moment. But anyway, for the sake of this conversation it's secure and you know, it's, it's encrypted at rest and yada yada. So I could go to the agent within MCP and go, "I'm going to London tomorrow." It already knows because I've used it before that it's Paddington. Even if I say London because that's where I go into, it knows who my local, what my local station is because it's got memory and context about the previous things I've done. So you can literally, you might even go, "well last time you (Doug), we booked a ticket...ticket for

you at about 10:30. Do you want to go the same time?" You know, whatever. You can just ask it and it'll do, it'll perform it, it won't just give you ideas for when and where it'll go. "Do you want me to transact", you know, just give it permission to do it. It's the same with lots of transactional things. I'm not just talking about money transaction, but when you need to send something to somewhere or do something with data. I think that, that understanding of how you can bring an LLM to do, to run, I go, you know, interconnections between their purpose. So the purpose of that model is to go and look at this, go and do this, fulfill this and come back. Another part of that whole movement of MCP is kind of being supported by what the likes of ChatGPT are doing of their latest models like O3 and O4 mini, where they're really thinking about, you know, how you can use dynamic feedback loops and, and reasoning and not just go, "there's the output". That's right. It'll...it'll question itself and you know. Lena Robinson (Host of Creatives WithAI) 00:23:33 This is what it needed, isn't it? It needed that. Yeah. I mean I, on one hand, I quite like the fact...because adulting is fucking hard, and I hate admin...and I...and admin's annoying, and I'm a creative, and when I get too much of that stuff, it blocks my brain. Right! Doug Ayres (Guest) 00:23:51 Yeah. Lena Robinson (Host of Creatives WithAI) 00:23:52 So on one hand, I'm like, oh, yay, let's. Yeah, let's let it. And also when I'm using things like ChatGPT, I have to then go and think, like, I need to grab this document and I need to grab that one. I mean, if it had access to everything that's on my Dropbox, around my methodologies and tools and everything, that would make it a lot easier. It would be quicker. So from that perspective, I think it's a really cool thing. And I'm probably 99% like, "yeah, let it just do its thing". The one thing, and this is a generational thing and an age thing, letting it have access to my money...Nah, fuck off. There's something like...but I know it's silly because it's not logical. It's an emotional thing. It's like, I don't think friends have access to my bank cards. Like, lots of people just hand their bank cards. I don't do that. Yeah, anyway. Bit of control freak. Doug Ayres (Guest) 00:24:43 But you're handing, like, my thought just on that point, though, you're handing like, just because they're trusted, like Visa, MasterCard, PayPal, Stripe. David Brown (Guest/Host of Startups WithAI) 00:24:52 I know. Lena Robinson (Host of Creatives WithAI) 00:24:53 Stupid, isn't it? Doug Ayres (Guest) 00:24:54 You've given them access to your card details, your bank details for Direct Debit. It's the same thing. Lena Robinson (Host of Creatives WithAI) 00:24:59 You're not...giving brand power, I trust the brands. David Brown (Guest/Host of Startups WithAI) 00:25:07 But they're the ones giving you the card. Sort of, some of it, so it's a little...it is a little bit

different. But, yes, I agree, Doug. I...I totally understand what you're saying and...and Lena understand what you're saying as well. It's interesting, this whole idea of agents. I was having a discussion with somebody else on another show the other day...so you have no idea how many podcasts I go on...but one of the things they were talking about, the Perplexity Voice model, which is essentially the same thing that you're talking about, they were saying that it's basically reached the point that it's better than Siri at doing a lot of stuff. And we were sort of thinking...and I threw the idea out there, so I'm throwing it out here in case anybody hears it, because you heard it here first or second...which is, I wouldn't be surprised if Apple went out and just bought them, because Perplexity, they've already tied it in so it has sort of more access, particularly on an iPhone, than most apps are allowed to have, so they've obviously got some sort of a special relationship with Apple already. And if it's already more functional than Siri and can do a lot of stuff that Siri can't do, you almost have to think that, that would just be the play, right? Like Apple has some ridiculous...like a trillion dollars of cash in the bank, so they can literally just walk up and, you know, they can just buy Perplexity for whatever. Like they don't even care at this point because they can't do their own AI. Anyway, their Apple intelligence thing was ridiculous to begin with, and it's terrible. Lena Robinson (Host of Creatives WithAI) 00:26:46 And you're going to be buying...next thing? David Brown (Guest/Host of Startups WithAI) 00:26:49 Yeah, well, I should have bought them 20 years ago. I wouldn't. Doug Ayres (Guest) 00:26:54 Apple or Plex team before they existed. David Brown (Guest/Host of Startups WithAI) 00:26:57 Well, yeah, yeah, so yeah, so anyway and on a personal level, I'm all for it. There's a book and the reason I was looking away is I was looking up the name of the book. So Dan Brown, the guy who wrote 'DaVinci Code' and all that sort of stuff, his last book that was published was called 'Origins'. And if you haven't read it. Have you read it, Doug? Doug Ayres (Guest) 00:27:23 I haven't finished it. I started it. I've got it somewhere. Lena Robinson (Host of Creatives WithAI) 00:27:26 I'm going to add that to my long list of books I haven't not read yet. David Brown (Guest/Host of Startups WithAI) 00:27:30 I can't talk about it then. Damn. Okay, read it. Let's...I'll just say, just read it. I think that's, and we'll come back and have an offline discussion about it. Lena Robinson (Host of Creatives WithAI) 00:27:41 That means we're already setting up round two of this discussion, by the sounds of it David Brown (Guest/Host of Startups WithAI) 00:27:44 ...or drinks at the pub or something. Doug Ayres (Guest) 00:27:48 I mean, just on your point about Perplexity, I mean, I've watched a few interviews with the founder and I'm not going to try and butcher the pronunciation of his name, but I, like, anything can happen. But I...I feel like he's not the kind of guy who wants to sell out to Apple. He was talking like literally this week about the conversation around OpenAI wanting to buy Chrome, you know, if they were going to dismantle part of Google. And his viewpoint was, no, we, like, obviously we discussed it, but not as the discussion should we do it? It was more like, oh, that's happening. That's a conversation. They're looking into building, you know, their own browser. They're thinking about the future. Like, there's another podcast in its own right around 'The future of search' and around, and that, that kind of links into MCP territory. It's like all those browser based things you do in the multi-modality and just, you know, you know, it's watching, you know, locally what you do, going back to the security point, which will be brand recognition. I mean if, if Microsoft or Google come out with something like that, you'll be like, yeah, they' fine. But who knows? I mean maybe, maybe they will. But my hunch is that he wouldn't sell out. David Brown (Guest/Host of Startups WithAI) 00:29:04 I, I think everybody has a price. Yeah, I mean you know, you look at, look at Barbara Broccoli and the, you know, the family that controlled the Bond IP, and you know they, the...the...the studio bought all the Studio IP, but the, you know, but Barbara still and...and her son I think had control of the...you know...who did what. And she basically said "we're never going to sell", and then Amazon gave her a billion dollars and she said "okay, fine". So you know, again, it's happening all. Doug Ayres (Guest) 00:29:33 The time, isn't it? It's like whiz getting when you have. David Brown (Guest/Host of Startups WithAI) 00:29:36 An unlimited amount of cash on hand. I think that, so anyway, that could, that could be interesting. Lena Robinson (Host of Creatives WithAI) 00:29:43 Yeah, it could, it could. David Brown (Guest/Host of Startups WithAI) 00:29:45 Well that, so while we're talking about agents. That's a good segue. Lena Robinson (Host of Creatives WithAI) 00:29:49 It is a very good segue. So that is the first true trial of the news area, which was actually really fun and really interesting. So I'm definitely looking forward to more of those. So Doug, before we jump into talking about specifically what your views are of what's going on in your industry at the moment, there's one question I always really like to ask. We are Creatives WithAI after all...is your view... is about how you define creativity. How do you define it from your... Doug Ayres (Guest) 00:30:25 I did write some notes on this as a precursor, because obviously we saw the question beforehand and then I just ended up writing reams and reams of drivel. Top line...creativity is like symbolically, it's like new, like a new...new thought, a new approach to something that might exist, or something totally new in...in terms of

thinking, or how you present an idea and a concept. I...I feel like it's...it's difficult to articulate in one sentence what it, what it is. I think that if that's like pertain...as it pertains to AI, and I'm probably jumping the gun here, is that the creativity that people are talking about from AI is beholden to human creativity. Like at its core everything that it's leveraging, what it's making stuff up from, reasoning from is...is human knowledge and human creativity. Whether that's written knowledge, written creativity, visual sense, whatever. So yeah, creativity is not, you know again as it relates to AI or human. Creativity is not dead. Lena Robinson (Host of Creatives WithAI) 00:31:48 No. Doug Ayres (Guest) 00:31:49 Like I think it's the one shining light around everything. If you know, around everything becoming so you know, on a level plane as to the points about bias and about what these models do, what these things are doing. How you know, it's...it's making everything probably it'll get more and more normalised and boring. So creativity is becoming, going to come more and more important. So is human empathy. So the things that we...what makes us human, versus being able to spout out knowledge and science. It's those things that's creativity. From my...to my mind I've been talking about this for about like 8 years since I really got into this was around...that's just a really important facet to what makes us human. Anyway, let's not get too deep. Lena Robinson (Host of Creatives WithAI) 00:32:37 No, we can go deep at another time maybe on 'round two'. What I think is really interesting about that. So you've taken a journey, and having done my stalking on LinkedIn thing and researching you and talking to you as well a month or so ago when we first had a chat about this was you know, you've been an architect, you've been a designer, so you are truly, in my mind, a creative. But you've ended up sort of owning several businesses, one of which I think you, you sold prior, that are leading the way in a...in the...in the AI evolution. So you've got EBM which is your sort of end to end consulting and engineering support business and then you've got...and I'm trying to read my own writing here Chat2Impact, which is...I find I wanna dig more into this...I'm sure Dave does as well... is AI powered Chatbot solution for nonprofits and charities. So based on that what I'd like to ask you is like how is AI, through that lens, impacting what you are doing? What you are doing as businesses, impact across how it's impacting the humans as well. And that is really what I'm trying to get. Doug Ayres (Guest) 00:34:00 Really good question. I mean probably a bit of a weird one for me because that's what our business gives as consultancies to other businesses. So. Lena Robinson (Host of Creatives WithAI) 00:34:11 Well, how is it impacting those businesses then is probably the better question? Doug Ayres (Guest) 00:34:14 At the end of the day they come to us because they're either looking to understand the implications of doing something with it obviously to the benefit not, not to the detriment of the company. But that can be taken many ways. But I mean we've been to our customers, we've been building out different kinds of solutions. A lot of them have...the low hanging fruit back in the day was...were Chatbots, and it's quite funny, but obvious to me, that the biggest shout, the biggest implementation, the most obvious implementation that's going on everywhere is essentially ostensibly a Chatbot for the masses. So I always, everyone was going, "oh chatbots, crap"...Yeah, they were crap because they would, you know, they were decision trees and you know, more basic forms of lateral language processing and very, you know, turn based. So lots of those things were doing, but in terms of our other projects, clients starting to understand how they can use, I don't want to say generative AI, but LLMs for knowledge retrieval, which is you kind of the most important one from, from my, to my mind around how you know, business can take their, their own IP, their own knowledge and commoditise it internally or make it more openly available. So conversations where I've had a few conversations this year around SOPs sort of function. So standard operating procedures in quite complicated businesses. Whether you're in marine biology, or marine manufacturing, or you know, northern gas networks or something like that, there's these really smart folk there who know how these things work and there's, they're getting, they're getting to retirement age and now is the best opportunity to bottle their IP into these, you know, into these processes. So you have all these, you get them to write loads of documentation and ways of working everything else and then use an LLM to query that so that when people are, you're onboarding people and they're out and about, they can do that, they can get that knowledge back really, really quickly and you know, summarise and reasoning and all those things. We've been talking about how it's affecting us even in the way we work, just in our day to day process as a business other than the fact we build, we basically stand on the shoulders of giants. So we use, we're a Google Cloud partner, we'll take their technology, we'll plumb stuff together, we'll, I don't want to talk about wrappers, but we basically plumb these APIs and these technologies together for a given task. But even internally like we, we use it as much as any company probably to write, you know, blog posts as a first draft based on a topic, and then we go through it. We don't just shove it out there but we use it as a basis for what we're doing. We'll use it for all sorts of kind of knowledge retrieval. We'll do it for ideation as well, just as a kickstarter. We're thinking of a name for this, this thing we're doing. It's related to this, this and this. "Can you think of any words that are seven letters that sound this or

that?" and it'll produce a list. It's great, low hanging...low hanging stuff. Like we even use it for image generation for some of our decks. You know, I mean I'm a...you know, I've been using photoshop for about 20 years and I still need to play around with that. But you know, off the cuff image generation is getting unbelievable and it's going to, going to get better and, I don't know, there'll be a point where you just can't make it any better. But there's a long way to go. But it's getting pretty darn good. Lena Robinson (Host of Creatives WithAI) 00:38:19 Yeah, I've found that...probably using the wrong tool. Actually I haven't used MidJourney or anything. I use ChatGPT to do some image things. Oh my God. It was horrendous. And I'm used to briefing creatives so I do brief it like I'm briefing a creative but oh my God, the things that came out I was trying to do...it's gonna make it sound really funny. I was trying to just basically do a visual representation of a process that I...onboarding process for something for my artists actually, and it came out with something that had been...that had been designed for Russian propaganda in like the 1930s. It was hilarious. And I just kept trying to...and it just kept getting it wrong, So I, I was using the wrong tool, but I do think it's interesting that you're using it for ideation and things like that. I think that's where a lot of people are starting to use it. And what you're doing though, you're talking about the sort of, the Chatbots and things for what you're doing with the NGOs. I'm really interested in digging into what you're doing with that because I think that's quite an interesting...Charities and NGOs. Doug Ayres (Guest) 00:39:41 The use case, low hanging fruit is for process-automation...automation. A lot of these non-profits we work with are short staffed, so they want to be there for their supporters and their beneficiaries and they're looking at ways to be there and be, you know, present in a strange way, but to answer questions when the answer to whatever that person's looking for isn't quickly receivable from their website. Researchable. Also automating processes that they usually spend a bit of time doing. Like if you're a...most of these medium-sized charities have some sort of CRM and when someone's changing their address or wants to do an event or do this and do that, you know, change their donation. It's...a lot of these processes are, you know, you can configure it...these systems like a Chat Interface to, to push and pull data from an API and do those things for them with authentication. So we're doing some of that stuff. The one that I talk about a lot, because it's a really, I think, a really cool use of the technology in a really sort of harsh area is for NSPCC's Childline. Lena Robinson (Host of Creatives WithAI) 00:41:10 Right, okay. Yep. Doug Ayres (Guest) 00:41:11 So we started working with them in 2019, still working with them now, and it was quite well-timed ahead of COVID So if you don't know what Childline is, it's basically a support line that mostly youths, you know, from, you know, say 8 to 18 or whatever, will call in if you know they're in, they're in tough circumstances, whatever that might be, whether it's domestic abuse, or peer to peer abuse, or just having, you know, dark thoughts about the world as it is. The problem, the NSPCC, the...the problem it was solving with what we built was a...and it was a Chat interface sort of in front of a child...the counsellor on Childline...is that there wasn't always a counsellor or enough counsellors there to be immediately taking a call from, you know, many hundreds or thousands of children at any given time, often in the middle of the night. They do have counsellors on, but not enough to deal with the volume. So the idea here was "how can we feel like their needs are even minutely met whilst they're waiting to get on a call?". So we created a dialogue that helped with that. That was also obviously there's one of, you know, benefit of that was just trying to hold them online and reassure them, and we worked with the NSPCC to make sure that what we were, the information we're giving out and it was all, you know, safe, hygienic for the use case. And the other side of it was, it was helping counsellors prepare for the call before they got onto it. Okay, yeah, because they're looking at the, the chat that the individuals already had, so there's a little bit of context of really obvious stuff. When you think about it. But they, they were, they had the...the balls or whatever the term is to do it. Yeah, the impetus to go and do this. Right. And so that's one really interesting use case. But our other areas around again, around knowledge retrieval, is like working with another charity called Turn to Us and looking at how we can implement a better grant search for you know, some 3000 grants that they... Lena Robinson (Host of Creatives WithAI) 00:43:29 It's tough if you. I've had to do it manually when I've worked with charities before. It is horrendous. Takes days and days and days and even then you don't know what you're doing. Doug Ayres (Guest) 00:43:37 Well this is even like this isn't food. Like when people hear about grants they go well as businesses when you hear about grants go oh that'll be that. You know, three grand, five grand, ten grand for something. This is for really small grants, like some white goods, like a shopping coupon, like things like that. So it's a one to one like grant where can I get things that are, you know, for that individual? And therefore the search becomes quite nuanced and you know, using it you can see how an LLM approach or...and this kind of approach to that...all of that knowledge...of that information and using this. It would have been a...taken a long, long time to do it through just like machine learning etc. But now you can do a bit of rag retrieval, augmented generation, use a question, take out the intents in

that and the entities and bring an answer back. Sorry, getting a bit geeky there. Lena Robinson (Host of Creatives WithAI) 00:44:37 No, I like that. Doug Ayres (Guest) 00:44:39 We wouldn't have approached it like two years ago like this. Yeah and it would have been a longer project quite frankly. Lena Robinson (Host of Creatives WithAI) 00:44:45 I'd imagine it's quite a game changer for the charities and the not-for-profits to have somebody like you guiding them, because having worked in several, it's a really tough place to work. The time constraints, and to your point, the resource constraints, are just so massive so that if you can help with any of that, that's extraordinary. Doug Ayres (Guest) 00:45:10 Hopefully. I mean we're still, we're still small like you know, we're a boutique. There's like ten of us, so we can't do everything. That's why we built this Chat2Impact solution like try and make it super, super cheap, but start to try and get charities on...on the AI train to. To do things in a slightly different way. We've got a long way to go and what center and front of mind for me is keeping it accessible but also safe. I think charities are...oh I know charities from what I've heard anecdotally, they're often the last to adopt new technologies not because they don't want to, but because they're charities, and they've got so many other, you know, bulls they're juggling. So we're trying to lead the charge a little bit in that area on a very obvious use case for for them, but there's so many other use cases outside of chatbots like Grant Search like actually on a one to one doing the like this MCP conversation we had earlier that could be really useful for individuals at charities to go off and use these different systems. But to your point Lena, about image generation, you made a funny comment about the AI chose the wrong tool and I was thinking actually no the question, you should... Lena Robinson (Host of Creatives WithAI) 00:46:30 No, no, no...I chose the wrong tool. Doug Ayres (Guest) 00:46:34 So what a really of I'd say obvious but a hopefully useful sort of way of thinking about it is actually before you ask something like ChatGPT to do that thing for you go, and ask them how they'd approach it or what are the best tools. It'll...because it's got a better way of doing. Yeah I mean and then it'll go well actually you could do it this way and da da da da. But you might want to go and use this over here to do that particular because there's a lot of talk around wrappers around LLMs about rappers or solutions. They do one part of GenAI really, really well. So you'll go off and find a slide creation tool that can take diagrams, and there's a few out there. Another one that's just come out from China is called Genspark and they, they do lots of stuff but they're...they're...their shining agent is this...this slide-creation model. So go and check that out. But there's...That's what's happening, like it is getting splintered because people are creating wrappers to do certain things really, really well and that, that will then become down to user interface design and UX because you probably could spend hours going back and forth to ChatGPT to try and get...and you thought "oh I might as well have just gone into PowerPoint or Google Slides and gone and done it myself". David Brown (Guest/Host of Startups WithAI) 00:47:49 Well the challenges at the minute I think, from someone who uses it, is it you...you now end up with, if you want to use all those tools you could end up with 20 subscriptions. Yeah right. And that's the, that's the big challenge is it then becomes not economical to have to do that, and I've said from the very beginning and, and from some of my experience working with machine learning and stuff like that in the past, I think what we're going to end up with is a federated model, which means there will be one AI that's your LLM, that's your conversational AI that will sit on top and that's what everybody will interact with. But underneath it, there will be specially trained AIs or machine learning tools that will do different tasks in different specific areas. So you'll have a physics engine, you'll have a marketing engine, you'll have...Do you know what I mean? You'll have an image engine, you'll have a video engine, you'll have a chess engine, you'll have a...and that top layer will abstract all of that and you'll just ask it a question and then it will go away to that specialist AI that sits in its network and it will go and bring back the data and just act as a conduit. So I think that's where we'll end up eventually. And then what we'll see is also that abstraction layer on top, that's where you'll get the different versions. So there'll be, you know, the culture and everything in Western Europe is very different than, say, North America, which is very different than, you know, places in the Global South. You know, you look at Brazil and South America is very different than Australia and New Zealand. So, you know, I think each kind of region is going to end up having its own top layer, but I think you'll end up with these core services that will be underneath it. I did want to go back to something that you mentioned earlier. You know, one of the things with the very original Chatbots, I think was, and you, you mentioned decision trees and, you know, people, as soon as the new LLMs came out, they went, oh, this is great, they can talk about stuff. And then what they found is it was giving inaccurate information. What I want to ask you, Doug, is from, from your perspective, when the, when companies now are doing chatbots and you're saying, you know, or...or developing an AI product and they're training it on their own data, are they turning down the randomness quotient or randomness factor? Because that's a part of why AI hallucinates as much as it does I think, is because what the researchers found is that they have to force randomness into the answers,

because otherwise it gives exactly the same answer all the time. And it doesn't feel human, because humans, if I ask you the same question five times, you will give me a slightly different answer every time. And so there is a number that they put in to force it to be more or less random. And I'm assuming that for things like, you know, company information or internal tools. And I, I don't...I was looking for it while you guys were talking. I was trying to find the article. But it's, it's one of the big consultancies like Ernst & Young(EY) or PwC, or something, has trained a tool on a hundred years worth of their documentation, and work that they've given to all of their customers. And it's a totally enclosed system, so they've trained it on...only on their own data, and they now give that to their consultants to use to help them generate more information. But I would imagine that the randomness is turned down on that so that it gives a...a little bit of randomness, but it will give a more consistent answer. And is that, I guess my question is...is...is that how it works? Because you're building them. So do you do that or is there another way that you try to ensure that you get consistent answers? Doug Ayres (Guest) 00:51:53 Really good question. So Gemini, for example, calls it 'creativity' versus 'randomness'. I think a lot of them start to use creativity and you can set the creativity down to like literally, not literally zero. I think it's like 0.01 or whatever. So there's that. I think that the big thing that was missing from what you were just saying was around the prompt, was actually what you're asking the model to do and how you can also control it in that sense by saying, "do not make shit up". Like you can, you can. It's a really good example, actually, Dave, that you mentioned around, you know, Jackanory-time, when you're telling a story down the pub and you'll...you'll tell the story down the pub one day and, or, you know, bar, whichever the audience is, and then two weeks later we'll tell the same story and it'll be the same story, but you'll, you'll just say it slightly different really, because it's not a creative license. Lena Robinson (Host of Creatives WithAI) 00:53:02 Yeah, creative license. Doug Ayres (Guest) 00:53:05 Right. Maybe that time it gets funnier with age as you make more stuff up. I mean, I know mine do. It's probably because we just forget...it's called 'embellish'. But the, the interesting point around that, if you could say it's interesting, is what you were saying about how information is reproduced. Now if you can make sure the...the...the inform...the factual information is always correct, like you're doing. You're looking at an annual report and you're going, "well, EBITDA was 'X'. Yeah, that has to be said". It can't make that stuff up and that's in the...the knowledge base, but it can fluff it around. It, it goes "I found this", "Oh, interesting you asked that. You can do that." And that's not hallucinating, that's being. That's kind of where they're meshing human interaction with factual stuff and that...that's doable in terms of what we like with Chat2Impact. What we first launched with this, which is the current iteration, but we're out adding more LLM smarts into it, is...one part of it is called the FAQ Genie, which is basically, we go...we...we get a URL from the charity's website. They put it, one in at a time. At the moment, we're not doing the whole lot, one page at a time. It goes and scrapes the page, it indexes and it basically creates questions and answers from that page to get them going. And then they, the charity can view the questions and answers and they can annotate them. I'm not annotating, sorry, edit them. And what's sort of controlled is that we're using an LLM to, to look at the question the user asks and match it to a question in their version Chat2Impact, but not the answer. The answer is always what they've defined as pre-canned. No hallucination. It's just going, "the user's asking this, this is the closest we've got to that. Their question to our question. And here's the answer", if you get what I mean. And that's therefore controlled. Right. You're not...it's not making stuff up and if it doesn't have an answer, it doesn't. It doesn't give an answer. What we're about to do is actually go, well, look, you've got the FAQ genie, but you've also got an approach where you can tell the user, we found this information on the website, so we scrape the whole of their website, store it, and if...if there's no answer in the genie, there's a fallback to the LLM to look at and do the rag piece and it'll tell them, here's the answer. We've done a prompt that says don't make stuff up. And also reference where you got the answer from. And that for charities and everyone, really, quite frankly, it feels like it's quite safe. It's not...but this is low level controlled AI versus the bigger conversations around generative AI like in ChatGPT are doing some things really well, and other things really badly. Anyway, I've gone off on a tangent there. Sorry. David Brown (Guest/Host of Startups WithAI) 00:56:08 No, no, no. I led you down the path. Doug Ayres (Guest) 00:56:12 Thank you. Lena Robinson (Host of Creatives WithAI) 00:56:13 The reality is to a lot of big organisations this is just sort of reasonably basic, but to a smaller company or to a small charity, that is so...I don't keep wanting to use the word "game changing", but it will be game changing for them to be able to, you know, like you said, I've worked for charities where, you know, you don't have people that are on all the time, but the ability to be able to answer questions of...I've worked for young people at risk charity in the past and I know that, you know, they didn't always have people on, on board being able to answer questions. If they can do that through that process, then that's quite a game changer for the, the young people as well or whoever is the

beneficiaries of any kind of charity or what have you. Doug Ayres (Guest) 00:57:03 Yeah, I think so. Yeah. David Brown (Guest/Host of Startups WithAI) 00:57:05 Yeah. Lena Robinson (Host of Creatives WithAI) 00:57:06 I mean it, it kind of makes me then think about, we were going to talk about the future. So I thought that said grace nicely into what does the future start to look like? You know, I was gonna sort of do sort of three, five years and then all the way out. But let's just sort of mash it all together and look like what does the future look like from your perspective? Because you're obviously having to look at it on behalf of your clients for both businesses. Look from that lens. What do you think the future sort of looks like for future of work, the people that you're working with, but just sort of humans in general when it comes to AI through the lens that, that you understand what's coming. Doug Ayres (Guest) 00:57:48 I...Armageddon? David Brown (Guest/Host of Startups WithAI) 00:57:51 No, I was gonna say a tsunami. Lena Robinson (Host of Creatives WithAI) 00:57:56 That's hilarious. Doug Ayres (Guest) 00:58:00 It does. I'm not intelligent enough to really figure out...well, find a happy place for myself and what's about, what's happening because I can't quite configure it in my head, to be honest. And I'm not trying to be like a fantasist or like you know, churning the, the wheel of doom, but I think everything's gonna change in the next five, 10 years. I don't know how quick. I'm not talking about AGI yet. I...you know, I do worry about bigger things that are happening in the world of, you know, quantum computing and about how that's going to disrupt every, everything from a security point of view. Lena Robinson (Host of Creatives WithAI) 00:58:39 David, I have to talk to you about that. Doug Ayres (Guest) 00:58:41 I'll try and bring this back. I will bring this in my own mind back down to where we are and what's happening is it's all very well bring it back to what we're doing now. It's all very well bringing these chat interfaces onto these websites etc, but what is the future of how people interact with the web? Like how are they finding out what they need and what are they doing? You know, talking about MCP or about how they transact or do things and where they go and that in just as humans how we interact with knowledge and the Internet, I haven't, I think just think I can't even begin to articulate what that is. The other side of it for businesses is like what is, what is the role of so many people? What's their purpose and what's the game? What's the end game here like when businesses are all about...Well, not all of them but a lot of them are about profit above purpose, about you know, shareholder value and you know this technology is disruptive to the nth degree in that regard because it's going to make a lot of roles redundant like it, it will because you will be able to use. I think David, what when you're talking

about the layers of AI systems and ChatGPT, they'll be one to rule them but and then these others which is basically Agentic AI or AI systems. I feel like, you know those things getting built out, whether OpenAI and owns all the underlying stuff as well is those things are going to get built and they're going to do things really, really well and it will need less people to do those things. And that's. I'm talking about in the online world what happens in manufacturing and engineering like maybe because of this technology it's going to help us create stuff that's going to need more people to go and build be more building stuff and doing things versus very much my environment is the online environment and comms and information augmentation versus actual physical world. What is getting done out there? I don't, I don't know. My answer is I don't know what the future brings. David Brown (Guest/Host of Startups WithAI) 01:00:58 Right. Doug Ayres (Guest) 01:00:58 I'm slightly nervous, I'm slightly nervous. I'm not gonna lie about. And I've got three kids and I don't know what when they get into work and later down in their career or even their kids, hopefully like what's... Lena Robinson (Host of Creatives WithAI) 01:01:12 I think they'll automatically, because I think the Generation coming up, I think they'll flex because that's the world they automatically know. Like, we...we've got better. Like our generation compared to our parents generation, whatever, we've all got better at flexing. We no longer expect that we're going to have a job for 20, 30 years and then retire. Like, you know. And I think as the generations are going through, you know, Millennials flexed a bit more, Gen Z's are flexing a bit more and whatever. The next one, I can't remember, it's called Alpha or whatever, you know, they're. David Brown (Guest/Host of Startups WithAI) 01:01:43 Calling it Gen AI now. Lena Robinson (Host of Creatives WithAI) 01:01:45 They're called what? David Brown (Guest/Host of Startups WithAI) 01:01:46 Gen AI. Lena Robinson (Host of Creatives WithAI) 01:01:47 Oh, okay. Gen. Well, there. Right. I think they'll be fine because they'll be born into a world of knowing that they need to pivot, pivot, pivot. Hate that word. But anyway, you know what I mean? That's my thought. Doug Ayres (Guest) 01:02:01 My question, though is what are they going to do? Lena Robinson (Host of Creatives WithAI) 01:02:04 Who knows. Well, make coffee. David Brown (Guest/Host of Startups WithAI) 01:02:07 Well, it's interesting, Doug, how old are your kids, if you don't mind...and... Doug Ayres (Guest) 01:02:11 At all about to turn 16, 14, and about to turn 12 in the next month or so. David Brown (Guest/Host of Startups WithAI) 01:02:17 So. So you've got someone doing maybe GCSEs or just to study for A levels coming along. My son's going to be 18 next month. He's doing his A levels now. Doug Ayres (Guest) 01:02:28 Right. David Brown (Guest/Host of Startups WithAI) 01:02:28 A lot of the kids, not all, but a significant number of the kids in his school basically have already sacked off the idea of going to uni because they've recognised the fact that anything that they would study for at uni in 10 years isn't going to probably exist and is going to be extremely difficult to work in. So a lot of them, particularly the boys, now, this is a whole different gender discussion that we could veer off and crash into if we wanted to, but maybe we'll do that another time or on my other podcast. But a lot of the boys are looking at trades and they're going. They're going to go down the apprenticeship route. They're going to get back into doing. I mean, I keep telling my son, I keep saying, look, be an electrician. I'm like, seriously, you can get an apprenticeship. They're desperate for people. You could get a good job, you get paid and you get training. At the end of it, you've got your license to, you know, to be an electrician and you can go off. And once you're fully qualified, you're looking at between £50/and/£80 grand a year straight away. And I'm like. And you put it down at the end of the day, right? Yeah. Okay, you work hard during the day, but when you're finished with your shift, you go home and that's it. You don't have to worry about it till the next day. And then you go back and you do your thing. And you know, and so I think that for the midterm, I think a lot of the jobs were that people are going to have to go back into doing physical jobs, jobs. So it's going to be service jobs, it's going to be, you know, doing trades, it's going to be those kinds of things. Because what we're seeing is they're, they're calling it 'quiet layoffs'. And essentially what's happening in the marketplace is...is...is twofold. What we're seeing one is...is companies aren't laying off people because that's bad PR, but what they're doing is they're saying you have to come back to the office. The people don't want to come to the office, so they quit. And then those companies just don't hire anyone to replace them because they don't need to, because their employees using AI can do everything. And this is what we're seeing. So this is, and I talked about this in March of 2023, not to thing, but, you know, this is what we're seeing. So they're just not filling in those junior roles anymore. And so we're going to see over the next kind of decade a very slow erosion of, of the lower levels working its way into the mid level positions because they're just not going to be needed. And that's going to be the problem. And it's going to be five or ten years down the road, you know, when kids that are in primary school now are coming out of doing their A levels and they're going out and there's literally, there's not going to be anywhere for them to go other than them, for them to do things like jobs that require physical, you know, a physical human to do it. Robotics is 30, 40 years away, no matter what anyone says. I know people who

build robots for a living and they're like, they're way too expensive, they're way too sensitive, they break so easily and they're, you know, they're just not economical to even maintain. So yeah, okay, robotic arms in a factory is one thing, but like a human kind of shaped thing that you think, go into your house and do your electrics. Nah, we are miles...miles away. Doug Ayres (Guest) 01:05:43 Yeah, I...I can, I totally agree. I mean, I'm...I'm pleased there's another Dad out there, he's of that mindset. I mean, I haven't haven't had this conversation down the pub very often, but I...I...I'm of exactly the same thought and I probably said it in like 2011 or something just to get that in there. That's a joke. But I've been saying that to my... Person 4 01:06:07 We're not worthy! Doug Ayres (Guest) 01:06:08 I've been saying that to my eldest because he's obviously the one who's getting into that space faster than the other two, is that those, like, things that AI can't replicate are the things you need to go and look at. And I think there's some areas of engineering that are quite, sort of, a little bit protected. I can't substantiate that at all. I've just got this weird idea that there are things, but I think the things where you need to be there in person, out and about, or doing stuff that isn't going to get automated anytime soon is where they should be going. And I've also said to, you know, my eldest, like, I'm not sure university is a, is necessary either. I mean, I was at university 30...yeah, 30 years ago, and I even remember at that time thinking I shouldn't have bothered going to Uni. Yeah, I studied architecture, but I never became an architect. But I just remembered thinking, "well, that was a waste of three years. I should have just gone into work". Honestly, I met some wonderful people. But it...it was a bit of a joke really. But yeah, look, going back, bringing it back on track, I think that's, you're absolutely spot on. I think in terms of the future of work and what people are going to be doing, and the knowledge workers are a bit screwed. David Brown (Guest/Host of Startups WithAI) 01:07:30 Yeah, it's coming for the smart people, which it's never done before and we don't know what to do with it. Doug Ayres (Guest) 01:07:35 I think I'm, I'd be screwed in like five years, 10 years time. David Brown (Guest/Host of Startups WithAI) 01:07:39 Yeah, yeah. I like a bit of physical labor though, so I don't mind. And interestingly, you know, in my own career, I'm now looking at setting up and running a studio which is a physical space for people to use because again, the knowledge work, the desk work, you know, I'm already looking at it and going, more and more people are going online, they're doing more creative stuff. They all want to be influencers and they want to do this and that and having a physical space where people can go and they can do that. It's something that AI can't do. And so I'm already heading that direction myself. Lena Robinson (Host of Creatives WithAI) 01:08:15 And me too. I mean, I made a decision that I told Dave about a couple of weeks ago that although I will take on consultancy work, I am no longer going to be marketing the consultancy side of the business. Because where I see myself, I've got multiple elements of content creation and what have you...and my biggest strength is me, my personality, my background, my stories, my everything else and my ability to do this. Like, I've not been trained to be a podcaster. You know, I do seem to have a...this is me not shooting my own...blowing my own trumpet too much...but seem to have a natural capability of it, you know, And I think I'm having to, at 52 years of age, I've got to start thinking about what does the future look like. I'm the kind of person that doesn't want to probably truly retire ever, because that's too bloody boring. So, like, what do you want to do? But how do I be that person that steps from where we are at the moment into what the future is going to be? And a...I'm quite excited. I'm quite lucky, though, that I'm a person that can do physical things. You know, my dad taught me how to build things and create things and what have you. And that's the creativity as well. I'm a creative. I'm an artist. But I'm also a strategist and a thinker and what have you. And it's like...but I'm a communicator, so I'm not worried about my future. Cause I think it'll be fine. But also, I'm one of those rare people that I'm really good at flex. I do not get frightened. I've intentionally...I've done talks on this. I have intentionally, over my life, well, sometimes intentionally, sometimes not. But I throw...I throw grenades into my life when I think it's just like, oh, for fuck's sake, this is just ridiculous. And I will make a...I'll be going like this. And then I'll go whoop. And go completely in there sideways. Now, whether that's my bipolar or not, I don't know. But, you know, I'll throw it through my...yhrow the grenade because I know something's not right and it needs to move in a different direction. I'm very lucky that I'm not afraid of that. I think what will...people will find difficult, and I do, I understand where it comes from. Fear makes people make bad decisions. My first business that I launched over here, it's underlying philosophy was "bring on the brave", because my business partner and I had realised that the flip side of that was the reason why businesses make bad decisions or people make bad decisions in business is out of fear. And I think we need...ne of the things I want to be focusing on to help people move into the future is how do we help them deal with the fear? Cause I think that's gonna be a big one. Doug Ayres (Guest) 01:11:07 Yeah. I mean, a person could talk for hours on this topic. Also, like, you don't wanna sound like you're too sort of unaware of what else is out there in the world. Like we're talking about...you know, I think we're all based in the UK,

but we're talking about a very sort of developed world, Western kind of mindset and. Lena Robinson (Host of Creatives WithAI) 01:11:26 Yeah. Doug Ayres (Guest) 01:11:26 Worry and fear and problem. Whereas a lot of the world isn't developed and a lot of those things are. You know, they're industrialising in...in a different way. So they...they won't...that isn't a pain point for them because they're already doing...there's still so many of the population doing these things, David, that you're talking about, that were, I guess, in a weird way getting maybe regressed back to, by being more physical in our presence and more...yeah...doing things the old way, in a new way. Lena Robinson (Host of Creatives WithAI) 01:12:02 I kind of like that. That's a really good place to end. Doing things. I like that line. Doing things the new way based on the old way or vice versa. Doug Ayres (Guest) 01:12:11 I think whatever it is. Yeah. David Brown (Guest/Host of Startups WithAI) 01:12:13 The old way and the new way. Lena Robinson (Host of Creatives WithAI) 01:12:14 The old way and the new way. It's been a really interesting conversation today. I've loved having you on, Doug. It's been quite fascinating to hear your background and what's going on, but also really nice to co-host with Dave Brown. It's been lovely. I have one more question. I know we're heading towards the end of the show, but it's a question we always ask and it's a very simple one. I haven't prepped you on this one, which is...if you were to recommend or request that we would have one particular guest on the show, and it can be anybody that's out there, whether you know them or not. Who would you recommend, is that as a guest that you'd like to see on Creatives WithAI? Doug Ayres (Guest) 01:12:58 Can they be really famous? Lena Robinson (Host of Creatives WithAI) 01:12:59 They can be really famous. We've had famous people on already. Doug Ayres (Guest) 01:13:07 God, there's just so many interesting people out there in the world. Lena Robinson (Host of Creatives WithAI) 01:13:12 They could be an artist. They could be a musician. They could be an architect. Tey could be anyone. Agency person. Could be a film person. An actor. I don't know. David Brown (Guest/Host of Startups WithAI) 01:13:24 Do you want me to give a name while Doug's thinking? Sure, I'll fill some time for you, Doug. Well, I want to have Jordan Peterson on. Lena Robinson (Host of Creatives WithAI) 01:13:31 Oh, yeah. David Brown (Guest/Host of Startups WithAI) 01:13:31 I want to have Jordan Peterson on. Lena Robinson (Host of Creatives WithAI) 01:13:33 Oh, yeah. David Brown (Guest/Host of Startups WithAI) 01:13:33 Every time anybody asks me, I'm always like, I want JP on because I want to talk to him about. Lena Robinson (Host of Creatives WithAI) 01:13:38 We'll get him. David Brown (Guest/Host of Startups WithAI) 01:13:39 At one point... Doug Ayres (Guest) 01:13:41 It's really weird you said that because I was...I was thinking around those areas and I was thinking the...and I think a lot of people also share this, this opinion, but Stephen Fry is one of the most interesting people to listen to on any topic, even when he's really reading Harry Potter. I...I would probably, and I know that he's obviously chatted to Jordan Peterson quite a few times and they have a pretty intellectual debate and, well, it's always bound to be, isn't it? But I'd love it if you got Stephen Fry on. David Brown (Guest/Host of Startups WithAI) 01:14:12 That would be amazing. Lena Robinson (Host of Creatives WithAI) 01:14:13 Yeah, but he is already on my list. Don't worry. David Brown (Guest/Host of Startups WithAI) 01:14:16 He talks about AI quite a lot. I saw him at...at...I saw him...God, what was it called? There was a big AI conference a few years ago with the O2 and he came and did a presentation. Doug Ayres (Guest) 01:14:30 CogX. David Brown (Guest/Host of Startups WithAI) 01:14:31 CogX, yeah, yeah, I saw him at CogX. That was it. And yeah, no, that's a great name. He'd be amazing. Lena Robinson (Host of Creatives WithAI) 01:14:37 Well, we will give it a go and I'm sure at some point there's probably people I know or people that Dave knows somehow that we could find him. I know he's not particularly cheap because I've tried to cost him before. David Brown (Guest/Host of Startups WithAI) 01:14:49 So if anybody knows Jordan Peterson or Steven Fry, make an introduction for us. Lena Robinson (Host of Creatives WithAI) 01:14:53 Exactly. Doug Ayres (Guest) 01:14:54 Well, just get on the case. David Brown (Guest/Host of Startups WithAI) 01:14:56 Exactly. Lena Robinson (Host of Creatives WithAI) 01:14:57 I just want to say a massive thank you to Doug to come on. It's been a very enlightening and interesting conversation. So thank you very much. Dave, thank you for being my partner in crime. It's been amazing and thank you to the listeners. You guys are always out there supporting us and giving feedback on what you find interesting. So thank. Thank you very much. So stay curious, everyone. See you later. David Brown (Guest/Host of Startups WithAI) 01:15:24 Bye bye. Cheers.