Episode 13 - Generative AI, with Adrian, Dave Wallace and Rob Reed

In this extended episode, the tables are turned, and Adrian interviews Dave Wallace and Rob Reed about Generative AI and why it is such an exciting technology. 

Dave provides a broad perspective, and Rob, a financial planner at Cavendish Ware, talks about how he has educated himself about the topic and why he believes it is such a transformative technology. And how he has become a world-renowned expert on Guinea Pigs thanks to Chat GPT. 

Transcript

Speaker: 0:01

Is AI good? Is it bad? Let's put it this way. Today is the worst AI is ever going to be, and that's pretty amazing. So talk about its potential in freeing up people to do and be their best and the pitfalls of the productivity track, Adrian Ware talks to Dave Wallace and Rob Reed here on Facebook.

Speaker 4: 0:26

So the students of D class on Facebook.

Adrian: 1:07

Hold on a minute, though. Hold on a minute. No, no, no, no, no, no, no. You didn't get the memo. This isn't actually today, Dave, about you hosting the podcast. It's actually about me hosting the podcast and turning the tables on you and having you in the hot seat. So sorry to spring this one on you, because the subject today is going to be all things artificial intelligence, and I know nobody better to talk about this stuff than you. So sorry for throwing this one on you.

Dave: 1:34

Well, my pleasure, but thank goodness we got Rob as well, who I'm going to be leaning on very heavily because yeah, as Rob and I were just discussing, there are no real experts in this field, other than maybe Sam Altman at this moment in time.

Adrian: 1:50

So let me introduce Rob. Rob works with me. Hi Rob, how are you doing? Very well, Adrian. How are you? Yeah, yeah, good stuff. Thank you. So Rob works with me. Um when he's not doing interesting things like flying helicopters or being the UK's leading guru in guinea pigs, which I'm sure we'll talk about later. Rob works as one of our financial planners in the team, but he's involved, very heavily involved in the AI project that we've got within the firm and is very knowledgeable on this subject. So I thought it'd be great to have Rob coming in from a sort of a broader perspective from CW. And Dave, of course, you've got a lot of a lot of history and background in the whole digital world and your original business and what you've done was always based around that. So, I mean, maybe that's the place to start. Maybe give everybody a bit of background. They're used to hearing you as the host.

Dave: 2:34

So, yeah, I've almost been 30 years working in digital. So I started my career many years ago in Yellow Pages, which I don't know if people remember was a big yellow book which used to thump down on people's doorsteps once a year with lots of adverts in. Through a series of bits of luck, I actually discovered the internet for Yellow Pages, and I remember having this moment of lucidity around the future where I could see that the internet was gonna be it, basically. And I remember saying to Yellow Pages, this is so pivotal for you. And you know, I was a young whipper snapper, no more than a T-boy, basically. But my manager said, Yeah, I think you might be onto something. So spent a few years with Yellow Pages really trying to point them in the direction of travel around, you know, an internet strategy. But I do remember I had this moment of insight that Yellow Pages was a data business, you know, they had lots of data on lots of businesses, but they didn't see that. They saw themselves as a book business, you know, they saw themselves as a publisher. And I mentioned that because I think you know what I've learned over the years, subsequent years, is actually what digital does is really strips organizations down to fundamentals where they have to really understand what they are all about. And as you see, when we talk about AI, that's particularly true now. So I then left Yellow Pages, joined an e-commerce startup, and did some of the first implementations of insurance and banking in the UK, and from that I then founded my own business, and that business was really essentially helping businesses or large financial institutions to deliver better customer experiences through digital channels, and what that means is websites and online banking and eventually mobile apps as well. So I work with people like HSBC and Royal Bank of Scotland to really actually help them along that digital transformation journey, but always from a customer experience point of view, and I built my agency up and eventually sold it to WPP. I grew it internationally as well, so ended up in Hong Kong and Australia, which gave me a really broad perspective on this world of digital. But I guess a few years ago, I sort of started getting frustrated because you know what I could see is that so many companies were essentially taking the existing processes and putting them online and digitizing those processes rather than thinking about well, how can we use technology to create real value for customers? So I spent the last few years thinking about that, and this is where I think AI is a real catalyst for proper transformation, and I think some of the things that we've seen over the last certainly six months have so profoundly impacted the world that now is a moment for companies to really think about how do we use technology not just to enhance existing processes, but really think about how do we add value to our customers, our suppliers, you know. So I think we're a very interesting moment in time. So, Rob, it'd be interesting to get your perspective.

Rob: 5:57

Yeah, so for any non-techies, essentially in the past, you have to define an input, define an output. So you say if this comes in, this is what you need to respond with. So if you are having some sort of chat bot, if someone types hello, how are you, you say back hello, I'm fine, thank you, how are you? Whereas now artificial intelligence is the understanding piece to allow the machine to create responses similar to what it understands that language to be. So it's taken out that I need to find the output. And as you rightly say, it's not thinking for itself, but it looks like it's thinking for itself, and so it's very clever, very intelligent in that sense. So I think that's where AI is. It's in the programming, it's taken away the input-output that we have always had, and now it's think for yourself almost.

Adrian: 6:57

It was fascinating. So yesterday, Rob did a training session internally for some of the staff at Cavendishware, and one of the things you did, Rob, was you'd asked me and some of the other people in the team a question. I didn't know what the question was. Can you answer this question in a hundred words? Do you want to explain what happened next? Because it was fascinating.

Rob: 7:15

Yeah, so we played a bit of a game just to get people interested, and the game was human or AI. So I asked about five questions to different people in the team and asked exactly the same question, no iterative prompting at all, the exact same question to ChatGPT. And I used 3.5, so the free version, so anyone can do it. And without context, without tone, without anything else, the results were so similar, including the one that I asked Adrian, which is what's the best part of owning your own IFA firm? And there were even some of them formatted in the same points in the same way, and I think it just proved the point of actually that's instant access. I mean, one of the questions I asked to a member of staff, he was on holiday, he's come back from holiday since I still not got a response, but I've got a response from AI instantly, and there's ways to input, which is what we went through yesterday. The output's only as good as the input, really, and it's 80% there if you don't know how to input properly. So we trained on how to input properly and get a much more refined, better output. But even without that refined input, the power of it is incredible, really.

Dave: 8:28

It's also worth noting that it's evolving and learning. So the more inputs it gets, the more it's learning in the background. And I think this is where it starts getting quite like proper intelligence. So, you know, we are heading to that moment where again, I'm not sure it will be proper sentience, but it's gonna get cleverer and cleverer as time goes on. So ChatGPT has a number of variations which have been launched over the last few years, and the original versions were trained on very small amounts of data, and at each time there's a release, it's been trained on more data, and ChatGPT-4 is basically trawling the entire internet and learning from the internet. So I even talked to someone recently from News UK, and he said some of these large language models are actually coming behind their paywall and scooping their data out. So even things like paywalls aren't safe from these AIs scooping content out and basically learning from that content, and the more it has, the more it can actually be like you. So there's a number of things to think about, but going back to outputs, fairly soon, or even now, you can say, Well, write me an article in the style of Dave Wallace, and it will go and have a look at my articles and come back with an article in my style. So not only is it coming back with interesting content, it's then able to apply nuance and tone to that content as well. Just a tiny example is I wrote an article which was all about green issues and software, and I put it into ChatGPT and I said, turn this into a children's book. And it came back with an incredible little children's book. And then, you know, you've got a piece of technology called Midjourney, which using similar prompts, you can actually create images. So you ChatGPT, as well as providing the content, provided prompts to go to Midjourney. So within about half an hour, I had an illustrated children's book are all about green software. Not that anyone's going to be buying that in a hurry.

Adrian: 10:49

I mentioned at the beginning some of Rog's other accolades, including his being a leading light in the world of guinea pigs in the UK. I mean, it's just quite a niche field. I think we all have to agree that one, but this story is just brilliant how you did this. And it reiterates what you're just saying, though.

Rob: 11:04

I'll clarify, I'm absolutely not a leading expert in anything guinea pigs. I know nothing about guinea pigs. However, I spend my evenings because I'm part of the core kid gang, trawling different businesses that haven't maybe adopted this approach or are maybe a little bit timid about the AI bubble and doing things for them. So, for example, and I gave the example in the presentation I gave yesterday, which was all, you know, the structure, the presentation, the prompts, and everything was generated by ChatGPT and all of the images was generated by Midjourney. So I'm fully on board with those. But one evening I sat there and there's an amazing magazine called the Guinea Pig Magazine, and it's about everything guinea pigs. And quite frankly, I wish I had one for when I got a puppy, because it just goes through everything. So I wrote an article in the style of their tone of voice. I wrote three articles, in fact, within five minutes, and submitted them to the magazine, and I think I'm being featured in a magazine as some sort of know-how expert in guinea pigs. And I honestly don't have a clue, but I'm subscribed to the magazine, so I used some of their magazine articles, copy and pasted that into ChatGPT for tone of voice, and it wrote a brilliant article, and I asked ChatGPT what should I write about? So it gave me the article to write, it then structured it, I then just copy and pasted some text and it can copy the style.

Adrian: 12:33

It's a brilliant story, and it's representative of what this stuff can do. But the flip side of that is there's something that's quite dark and scary about that, isn't there? And I know there's been lots of column inches written about you know, shouldn't we be scared of where our AI is going? And if not scared, what sort of impact is it going to have on our lives as human beings generally, and certainly on the business community in terms of where businesses are going? I mean, what's your take on that, Dave?

Dave: 12:59

The genie's out of the bottle, so it's not going back in. My view is that you have to look for the positives. Now, Rob's story is a brilliant one because it shows that the media industry is going to have to change. When I talk to News UK, they're like, yeah, this is a challenge for us, but we believe that our journalists are top quality and we believe in the human power to write articles and investigate, and they absolutely see that this could be competition, but they see it actually re-establishing themselves as centres for excellence for the news. But there will be lots of other magazines who can't afford journalists who want to get their messages out, and they will embrace this technology in exactly the way Rob's talking about, and that's brilliant, to be honest with you. It means that, for example, minority groups or small communities can have a voice and publish really good quality content. I think the other thing which I find fascinating is we've just had the A-level results, haven't we? I'm not sure how many school kids there are in the UK or university kids who haven't used Chat GPT at some point. Now, you could say, Well, we'll just ban them from using it. That's never gonna start. Or you could start saying, Well, how do you change education to incorporate the fact that kids will be using Chat GPT, you know? So I think there's an imperative around actually, we're gonna have to change the way we think about things. Going down to businesses, I think you've got Rob who's seen the light, he's sort of seen the opportunity. Rob represents the future, I think, of people who go, ah, actually, I could do my job so much better if I use these tools. I mean, I don't know, I think it may have even been you, but within the relationship management IFA community, 70% of time is just spent on doing admin, which must be a frustration for everybody because you want to be talking to clients. You trust it. If you can get machines to do the admin, that leaves you 70% more time to be talking to people.

Adrian: 15:23

It's very time-consuming and it's not valued at the end of the day, it's a necessity, but it's not a valuable necessity, it's just something you've got to do. The value comes from the relationship and the understanding and the sounding board and all those other things, which is where I think as human beings we still very much have the edge over a robot, and I think always will. I'd like to think that. So we're very much embracing it within CW in terms of a whole project running now, which Rob's involved in with myself and a few others, to how can we make it positive impact? Internally, we're doing a number of things at the moment. We're kind of scratching the surface and looking at what we can do right now, but we don't know what we're going to be able to do in sort of six months, nine months, twelve months time because this is a very fast-moving world. But things like how do I do such and such? One of our members of staff needs to know what's our internal process on this, how do I enter this on the screen? You've got an immediate system on your computer where you just tap in the question or verbalize the question. That's not far away, is it? And immediately you get the right answer. It takes you to the right page, it explains what you're doing. That saves a huge amount of time and effort, and it means we can centralize what we're doing, we can make sure that the consistencies that any business needs to have are there, and that makes those iterative small changes will actually make in the bit long run make a massive change to what we can do and the time we've got to spend doing the bit that quite frankly we love doing, which is talking and advising people and helping people.

Dave: 16:48

You know, I'm actually involved in a small startup. Conceptually, we see that every business will have a digital twin, which is actually the artificial intelligence which sits at the heart of their business and is able to hold the data, make sense of unstructured data. So one of the beauties of ChatGPT is it can respond in a way that you would understand about very complex data, which at the moment, how does that get delivered back to staff and employees? It gets delivered back through turgit presentations, which nobody ever listens to or opens. If you think about efficiencies in businesses, like we spend a lot of time doing things that nobody ever really cares about. And actually, if you can find an interface to that information, for example, like we're actually talking to a bank about a staff survey where you know, religiously every year they do the staff survey and it sits there, they do a presentation, people fall asleep, you know what it's like. Whereas now, what we're gonna have is an interface, so managers can go and say, Well, what are my three biggest problems? And actually, how could I solve those problems? And give me case studies and give me a podcast so I can go to my start. So the the possibilities of all of these things are endless. So I think actually what AI offers is the opportunity to create massive efficiencies, but also help people fall in love with going to work again. I mean, we've just been through the what's it called, the Great Resign or whatever, where you know, what's the message? People hate going to work, you know, and I know that's not the case with Cavendishware because you've just got an award.

Adrian: 18:36

Thank you for plugging that one day, I appreciate it.

Dave: 18:39

Yeah, well, I thought I'd say it. You know, what better way than getting people back doing the things that they want to do? You know, so a lot of people are worried about the impact of AI on jobs, and actually, my view is it's a potential for a real renaissance in people doing the things they want to do. And I think if you're really smart, this isn't just about Cavendishware, all of your clients with businesses. If people are smart, they need to think about well, actually, how can I use this to supercharge my workforce, to get people doing the things I want to do, get rid of the crap. We have a productivity crisis in the UK. Maybe this is one way of solving that productivity crisis.

Adrian: 19:25

Yeah, one of the things that I heard quite a few times from people is yeah, I don't want to be in the vanguard of this, it's all a bit new. I won't do anything yet, I'll just wait and see how it pans out, and I'll look at it later. I personally think there's a real danger in that, that people will just get left behind and it's moving so quickly. I'm lucky in that I've got a Rob and I've got Amy and Matt and other people in the phone who are very focused on AI and understand it and embrace it thoroughly. But where do you start? For me, as that Luddite who'd never really heard of it until ChatGPT suddenly hits in the news channels in January. How do you get over that fear that probably sits inside because I don't know what to do, I'm gonna put my head in the sand and not do anything? Because I think that's a really dangerous place.

Dave: 20:05

I literally would start with ChatGPT. Go to ChatGPT and ask it questions. I did a presentation to the board of a company the other day, and I asked ChatGPT in the presentation to tell the board about its competitors, and you know, it came back with a summary, and I was like, and by the end of it, they were like, Oh my god, you know, we'd have paid a consultancy thousands of pounds to get that level of detail. But you know, there's things to watch out for. For instance, at the moment, ChatGPT, its data set only goes back to 21, but there are increasing tools where you can point it at specific data and it will make sense of that data, but just get your hands dirty. I was lucky enough to go to Money 2020 in Amsterdam, and NVIDIA were there, and NVIDIA are one of the big winners because they provide the chips, but their background is in actual graphics cards that people's children will be using in their computers, but those chips are what is being used by the large language model-based businesses. So they had various people like Abian Amro and Swift along. Everybody's advice is just go and start playing because as soon as you start playing, the penny will drop. I'm sure Rob would be very happy to do his presentation. You know, why don't you put on a webinar for clients where it sounds like Rob's presentation is a brilliant one? I can't wait to see it myself, to be honest with you.

Adrian: 21:31

As we were sitting here talking about this, I was thinking exactly the same thing. So there we are. There's an action point, Rob. As though you don't have anything better to do. No, I think that would be an extremely good idea, actually. I think people would like that. We found it fascinating yesterday. It was really, really interesting.

Dave: 21:47

Absolutely.

Adrian: 21:48

And it is it's learning how to ask the right questions, isn't it?

Rob: 21:51

I've written a lot of notes down, and because I know that we haven't got four days to record a podcast, I will probably skip over a lot of them. There's huge amounts to unpack, and everyone's going to have a huge amount of questions, and everyone's got their own opinion. And as we said, beginning this, no one's an expert. Um, there are a couple of people at an expert in certain things, but AI as a topic is like saying healthcare is a topic or technology as a whole is a topic. Um, you can't be an expert in all of it, but we've all got our own opinions. I could quite easily play devil's advocate, and I think some of my opinions have actually recently changed, having done some deep thinking about this, including having actually an opposing view to Adrian on something that I'll mention in a minute. But when you're talking about, you know, asking the right questions, a lot of people are concerned about well, this is taking away creativity or intellectual thinking and critical thinking. And actually it isn't. It gets you to think in a very different way and it makes you ask the right questions because you can get huge amounts of value, information, and insights out of any AI, whether that's audio, visual, or text-based. But you need to know how to do it. And the very last slide on the presentation I gave yesterday was humans or AI in terms of you know, where's the future? Are our jobs at risk? And I've got my own views on all of that, including how the future of org structures is going to look and everything else. But actually, we will be replaced not by AI, but by AI-assisted humans. There will be pushback from some people, and they will adopt this late, but they will adopt it late. I think the change in the landscape of businesses and jobs, people will struggle to keep up with, but that's change anywhere. That's when you know the internet came about or computers became widely available. That is just natural change, and we need to adapt to that environment. I think what we do need to be careful of is this whole productivity crisis thing. There was an example that I gave yesterday, which was when a washing machine was invented, it was to take away the amount of time that a household individual was washing clothes by hand. So a freed-up time for them to enjoy leisurely activities. That's how it was originally sold. What it did is create opportunity and expectation for people to be more productive, and with that comes more stress. So as we see this productivity curve or this technology curve, this capability curve go up, we're going to see stress go up with it because the expectation of productivity is going to go up with it. Well, you can do your job in half the time now, so therefore you can do double the amount. Not we can now sit down with clients and actually enjoy that conversation. And I think, you know, where we've got leaders like Adrian who are so adept to this and will lead the charge on doing this right, I think there might be a lot of people that can mistake that for, okay, well, I can halve my workforce and double my business turnover, and all that's gonna do is just put so much stress on people that we need to be careful of a healthcare thing here as well.

Adrian: 25:11

I agree entirely. You look back at when the internet came out and everyone was using emails and all that sort of stuff, same thing, you know, always going to create lots more leisure time or whatever, and it didn't. What it ended up doing, the positive is it actually made goods and services cheaper, it drove the price down, which meant that people could buy more stuff because people always need stuff, don't they? And I have a house full of stuff. Um, your standard of living increases, your quality of life increases, but it doesn't because actually you're now a slave to your phone and your emails and the work that goes along with it, and you have the horrible boss who puts all that pressure on you, you don't have the time to enjoy it because you don't have the time. So I think you're right, and I think that's where there is a moral and an ethical part to how this progresses. I don't know how that's going to work in practice because, as Dave said earlier, the genie is out of the bottle, and I think there will be people who won't do that, and it's an issue, but there we are.

Rob: 26:06

If you see the GVA, the gross value added per employee of large AI enterprises, which are very small org structures in the main part, there's government-published AI guidance on this, and there's a quote in there that says large dedicated AI companies make a huge contribution to the UK economy per employee GVA estimated to be £400,000. The economy benefit to this, the economic benefit to this, is huge. And it means that the GVA per employee will go up massively so. And you know, that is happening. And if you even just look into this and play with it, you know, I think we all started out by typing into ChatGPT, how do I make banana bread, or anything silly at home. If you don't know where to start, go on ChatGPT and say what questions can I ask you? How do I prompt you better? It's a person, and I think people expect so much of AI because there's this huge hype around it, but people don't understand that it's not this intelligent thing that knows your thoughts yet. So it's as if you meet someone on the street and you just then ask them, What do I need? They're just gonna look at you and just walk past you, quite frankly. You've got to provide it context, you've got to, you know, give it some actual additional information. You've got to imagine you're talking to a person, okay, because it needs the information, it can't just know what you're thinking.

Dave: 27:39

The difference between the latest version and the previous version is enormous as well, which just shows you the progression of these things. I just want to go back to your point around productivity, because I think that's such a good one. Going back to my Yellow Pages days, I remember seeing the internet, and it was like St. Paul on the road to Damascus. I could see the future, and I get very similar feeling about this, which is in a way, the genie is out of the bottle. This is something that everybody's got to get on board with. But now's the moment to start saying we've got to stop the downsides from happening. So the internet was just allowed to run absolute rife through economies because it was driven by thoughts of making more and more money, efficiencies, and actually became a way of just bad actors being able to be worse, to be honest with you. So now is the moment to put some of the structures in place. But in order to do that, you need to be educated. I really feel if anyone's sitting there going, I'm gonna let this one go, I'd just be very careful. Not only is it out, it's speeding up in terms of its capabilities.

Adrian: 28:49

I love the phrase, and I think I've said this before, which isn't mine, I've got it from somebody else, but I thought it's just so relevant. You know, today is the worst that AI will ever be. So the speed is incredible.

Rob: 29:02

The ball's rolling down the hill now, and people are chasing it to catch up. Yeah, so true. We can't learn about the capabilities of AI as quick as it's evolving. But what we can do is pick our niche, specialise in it, and understand how it's going to benefit us. So Adrian mentioned earlier about how we're gonna implement some of the parts in the IFA in Cavendishware, but there are things that we're starting to do, which all businesses and you know, especially our clients, should consider doing first, and that's just quite frankly, getting the staff on board to just have a play within their personal time and understanding the limits. So we mentioned yesterday about what at the moment we don't want staff to do whilst we're getting it put through legal and compliance and everything else. Um, so that understanding the limitations, but otherwise, just having a play, you know. Just personally, I put some just fun examples in there from director level all the way through the company, um, administrators, advisors, and planners, but also just relevant for anyone that just wants to have a play. That's what everyone needs to start doing, is just pick one, and Chat GPT is a very good one to start with, and then as you understand that a little bit more, you can then maybe play with some of the image generator ones and then understand this is my role or this is my interest. So, for example, we've got someone in the company that loves photography. Actually, the AI that's in the beta Photoshop is extremely good already, and it's good for him actually, once he understands a bit about you know the basis of text AI, to have a play with that. So, following what your interest is and that, starting with the base of ChatGPT, I think is a really good way to go for people.

Dave: 30:50

Yeah, fantastic. I'm very conscious of time. I think we can ramble on all day about this, but it sounds like Rob, you'd be very happy to do a webinar. I mean, I'd certainly appreciate it. I mean, I don't think you can learn enough about this. I'm fascinated to read your article or Chat GPT's article about guinea pigs as well.

Adrian: 31:11

Having eaten a few in my time, I mean, having had a few in my time, and I think uh the societal issues and change that we sort of touched upon, I think they are big issues. We individually can't influence that, but it does need thought at government level and regulation level to try and get that right. But what we can do, and exactly what we talk about, is look for the positives, look for how we can make it improve our lives, our businesses, and what we're doing, and focus on the good bits. So I'm actually very excited, but I think you've got to have an eye on the risks and the dangers that sit within it, but embrace it, guys. Listen, thank you very much indeed. We've probably rattled on far too long, but that was really interesting, lots of great insights, and really appreciate your input there, guys. So thank you very much indeed.

Rob: 31:58

Thank you. Yeah, pleasure, thank you.

Speaker 5: 32:02

Thanks for tuning in to Baseline, a monthly podcast series dedicated to topics that matter and wealth management. Be sure to check out the podcast on the next month on the marvelous week.

Speaker 4: 32:17

You have been listening to Baseline from Cavendishware, an NMD plus production.