The Discussion: Identity, Audiences and Avatars #21

In this episode of The Future View, we're going to get away from all this doomerism and take a positive view of AI. So we're going to look at some ways in which AI can potentially level the playing field. Now, I can't look at all the ways it could possibly do this, but I am going to focus on two: prompting and, well, what should we call it, avatareng. So in the first part, you'll hear from David Boyle, the author of Prompt. He's got a series of books that teach prompt engineering across many different sectors. David has worked in audience insight across entertainment brands for years, And as you'll hear, he's now turned his attentions to helping organizations get to grips with their audiences in order to understand how to better tailor AI for their own uses. You'll hear how he believes this technology can democratize creativity and strategy, no matter how marginalized people have been in the past. Links will be in the show notes to all of his prompt books and also to some of his writings more widely on the subject area. In the second part, I'll explore the creation of digital people, my favorite topic, with DID.com's Matthew Kershaw. He's the VP of commercial strategy at DID.com. Again, all info is in the show notes. We chat about what DID does and why and where it's all going in the future, the anthropomorphism of what are essentially bots, how avatars and multiple identities through avatars are going to change human-to-human communication and human-to-machine communication too. And we even ponder communication's world of performative AI politicians. Perhaps they're already here. Over to David and Matthew. Enjoy.

David, hello. Welcome to the Future of You podcast. Thank you for joining me today. I wonder if first you could introduce yourself to the audience and tell us a little bit about your background and how you got here and what you do.

Thanks. Yeah, I think the most important part of my background is that almost none of it involves AI. That almost everything I've done has been working with brands, and sometimes that's artists in the music industry like Snoop Dogg or David Guetta, and sometimes that's been TV shows at the BBC, or sometimes that's been startups like Masterclass. Or luxury goods like Harrods, but starting by working in brands and helping them to understand their audiences and help them to work out what to do differently. And that's what I brought to the AI conversation. So when ChatGPT came along in early December, a colleague and I, like, obsessed over it because it was practically useful in helping us do all the things we've been doing for 20 years. And so we obsessed over it. We thought we might be a little crazy, although we look less crazy now, and we wrote a book to try and help everybody else to work that out as well.

So tell me, what's the first thing you did with GPT? You know, be honest, what did you do with it or ask it or whatever?

I asked it, what are all the different ways that companies could segment audiences and what are the pros and cons of different methods? It's really nerdy. In the early days, it was absolutely explaining concepts to me in the style of Snoop Dogg, like explain quantum physics in the style of Snoop Dogg. Or another one I remember is a supplier repeatedly hadn't paid an invoice, and I was tired of writing emails, so I got it to write a poem, a really, really sad poem about how small businesses really struggle if they don't get their invoices paid on time. It's coming up to Christmas, you know, this is important. It wrote this heartbreaking poem for me. I sent that to the supplier and got my invoice paid. So being silly, but also practically useful, is some nice combination of the two.

So from sort of discovering it or discovering that it had a use for you, how did you get from that to thinking that you could advise other people or create like the program for successful prompt engineering or however you think about it? Maybe you can tell us what your philosophy is and how you got from that to this.

Yeah, it's a great question. And look, I think it comes down to having subject matter expertise and the thing that you're trying to do in the first place. This is a real lesson for all generative AI. Like, I'm really good at understanding audiences and knowing what brands should do differently. And so I'm qualified to know, is this helpful and how can I make it as helpful as possible? For example, I'm not a songwriter. I've never written a song and I shouldn't— I'm not a good judge of songs. So when I try and use it to write a song, I get garbage because I don't know what I'm doing. But songwriters, when I tell them how to use ChatGPT and when they use the language they have around how to write a song, it turns out they can find it very helpful to help them to write a song. So actually, all the lessons I've learned in the areas where I'm a subject matter expert can be applied to new areas. This is the journey we've been on working with songwriters and then scriptwriters in Hollywood and then, yeah, real estate marketers. I've never marketed real estate in my life, but I know how to use ChatGPT from my area. They know how to market real estate. And when the two forces are combined, it turns out to be very helpful in real estate marketing. So I'm on this constant journey to test the lessons that we've learned in our area in new domains and therefore like refine the lessons and help more people in more areas. It just seems like a fun thing to do to me.

So, um, in terms of audiences then, what have you found that the, the main areas are where you can use these tools to sort of help clients or customers improve? What kinds of questions or areas are they looking to better investigate or perform in?

Yeah, well, my experience working with so many different brands is that they're either not as clear as they could be about who their audience is, or they're somehow not able to translate that clarity into decisions. So in my experience, most brands could do with a bit of help on understanding audiences. And our philosophy has always been that to understand an audience, you have to look at their deep underlying needs, like what's really driving them driven to behave in a certain way or want a certain product or not want a certain product. And as it turns out, ChatGPT has a very, very good grasp of deep underlying human needs, and it's very able to apply concepts from one area to another. So if I ask about deep underlying needs around making phone calls, then it's very capable of teasing out all the reasons in novels that people have made phone calls over the years. All the academic studies on uses of phones, all the philosophical discussions on the nature of phone calls. It's very able to synthesize all of that into a very, very thoughtful list of the different fundamental needs that drive you to use a phone. Very good at thinking through different personas, and it's very good at thinking through the implications of those for how you might market that, or how you might advertise it, or how you might make that product more efficient or effective to meet their needs. It's like they're very good use cases, as it turns out.

So what's going to stop us from reaching a sort of homogeneity? If you take one sector, for example, and let's say every company or organization in that sector is using generative AI or GPT in the similar ways, or, or in fact asking similar questions, is it— are they all going to get the same output? Are we going to get a homogeneity over time of content?

Yeah, great question. It's a big concern of a lot of people. Um, let me tell you a story which illustrates why that's not the case. So we had one client, I really loved them. They read the book very early and tried it out for themselves, were quite excited about the results. That's great. And then I did a project for them for one of their brands. And I came up with these audiences, these audience needs, and these kind of marketing suggestions. They were a lot better than the ones that they'd come up with. And even when I gave them my prompts, they still weren't able to get output that they thought was as good as what I'd done. And I think that therein lies the lesson and the kind of answer to your question, which is if you really deeply have expertise in the area, you know what you want. There's a lot of very subtle things you do differently, subtle wording changes to prompt, subtle changes in the sequencing of the prompts, giving ChatGPT feedback at every step of the process. And it should be lots of steps in the process. You shouldn't ask for the final answer in just one prompt, which means by the end of it, although the text was literally written by ChatGPT, it's very, very much created by you and is unique to you. And so there's 100 different ways you would, you would guide it towards the answer that's yours that you want. Then at the end of that as well, I would also say you never should just take ChatGPT's output and just use it. You always should then reword it to make it yours and truly yours.

See, the way you were describing it there, I don't see that there's a massive substitution there for, you know, certain job roles. I mean, I know some tasks will swap in and out, but there's still an awful lot of human interaction with these tools, right, from what you've just described there.

Yeah, and I think the challenge is that it's most substitutive or most helpful for very experienced, quite senior people. Lots of other technologies have replaced like workers because at the low end, this actually is challenging to the most senior, most creatives. But challenging in that they should learn to adopt it because it will help them to be better, not challenging because it will replace them. So you're absolutely right there. The only replacement will be that people who don't use it will be replaced by people who do. But all things being even, no, that's really the real worry. I see it as an accelerator of your inherent abilities. So like I say, I've got no songwriting abilities whatsoever, so it's accelerating from nothing to nothing. But I'm pretty good at audiences. And so it's accelerated me from being pretty good to really very good. I would say my simple phrase here is like any thinking or communicating task. That's really vague and sounds really extreme, but I'm yet to find good examples where that's false. I'm really looking for them. Like I say, in the books, we've worked on 5 different industries with experienced professionals in those industries now, from screenwriters to songwriters to marketing professionals and chief executives of large corporations. You know, we're yet to find an example of a thinking or communicating task that it can't be helpful to. So I really think it's that broad and that vague, which makes it very hard for people to know what to do. And so that's why I think the books are really helpful, which is a step-by-step walkthrough, specific examples, which might be useful to you.

Yeah, because you've got some great— I've got the one for brands, I think, which is As you break down the chapters, obviously there's exploring audiences and markets, but you're then into sort of go-to-market strategies, innovation, having fun with ChatGPT. That's great. Why did you put that in there?

Well, a couple of reasons, I think. One is that it's not restricted to your professional life. This is helpful in having fun with the kids. So I write bedtime stories with the kids every night, and they're accustomed to that night. I have a standard prompt, and every night I say to the kids, pick 3 things that you want in today's story. They each pick 3, and then we get this totally custom bedtime story, which includes all these elements from their day, but woven together in like really wacky fun ways. I'm not creative enough to do that myself. I wish I was, but now I've got that ability to have like this weird fun with my kids at nighttime. So it's not just about professional use, personal use is good as well. So that's one reason. The other reason is, and this is pointed out to me by a friend and an ex-colleague I used to work on the BBC Earth TV shows with. I actually made a joke on Twitter about why is everyone being silly with ChatGPT? Don't you know, there's really serious applications to it as well. And she said, you know, that's— don't forget, David, that's how we all learn and that's how animals learn. So chimpanzees play with tools and then eventually work out they're practically useful. Children play with knives and forks and eventually work out they're practically useful, you know. And so that's how we learn new things. Have fun with it. Be proud of that, actually. And then eventually you'll work out how to use it professionally.

It's a great point. I remember when Apple first came out with some of its best innovations, people were describing them as toys, which I didn't think was a bad thing actually, right? And turned out not to be, of course. You've got also in the appendix how AI helped write this book, which is fascinating, and I wonder if you can tell us a little bit more about that. But also brings me on to ask you a question about authorship, you know, in this new emergent world Who or what is the author?

Yeah, well, although I think our book was the first book on ChatGPT published late last year, there are now hundreds and hundreds and hundreds, and it does feel like, it looks like, it seems like most of them were actually written by ChatGPT. We kind of guessed this was going to be the case, and we wanted to just be really, really clear with people, partly because we wrote the book, ChatGPT helped, but partly because the ways in which it can and can't help are real important lessons for people. So every time ChatGPT wrote a paragraph, and it's the odd paragraph here and there, we write a little footnote saying, yep, ChatGPT wrote this paragraph and here's why. Like, here's why that paragraph was better from ChatGPT than it was from us. Like, if I'm just trying to define a basic concept, that's much easier to just ask ChatGPT to find this concept for the reader than it is for me to do. It's much easier and it frees me up to be able to do the really difficult stuff which is deciding what goes in and what doesn't go in. When we run ChatGPT examples, deciding if it's a good output or it's a bad output and deciding like how it needs to be improved. So using ChatGPT for the things that it's helpful for, but, but making sure you reserve human creativity for the things that it's not helpful for. I would say that like there's a lot of other things other than writing that it was very helpful for. So when we first had the idea to write the book, we, we wrote 4 or 5 bullet points about what we thought it might include, and And then we pasted in the section from our website about who we are and what we care about. And we said, "Hey, write an outline of a book that covers roughly these topics, maybe, but picks on our interests." And it wrote a really thoughtful suggested outline and vision and purpose for the book, which we then edited heavily, of course. But it really accelerated our process of thinking through what the book was going to be. But yeah, the core of the book is just real worked examples of practical problems that we're actually working on and our judgment about whether it's useful or not, and then our advice on how to make it even more useful in the future. So yeah, very, very helpful, but don't get it to do your work for you. We didn't, and it's not how to use it.

I wanted to ask you about how, if we were to extrapolate that out into the future, because one of the things I'm very interested in, particularly because this podcast is about the future of identity in a digital world, how GPT and the like can really be harnessed by an individual to become almost part of their persona, their personality. I don't know, you're obviously at the forefront of these, so you're probably thinking ahead in terms of the ways in which this is going to go over the next 5 to 10 years. Is that feasible and desirable, do you think, that we take these tools and we take generative AI, etc., and we apply them to ourselves so they're almost an extension to the self and express us when we are in communication or information transfer?

That's a much better way of saying what I've been trying to say. Absolutely, yeah. Thank you. I'm going to write that down. And that's all about both prompt and method and judgment. And so, for example, when I said we pasted in some of the website, yeah, we did. But what's on the website? What's on the website is an honest and heartfelt overview of who we are and what we value. We talk about the things that we value. We talk about our philosophy. We talk about, you know, our experiences and the things we love doing. It's a really heartfelt expression of who we are, which happens to be well-written to train an AI to know who we are and therefore give us the kind of things we want to, that are going to work for us, that are going to make sense for us. So if I paste my website in and say, please write a proposal for a client, then it will uniquely tailor it to the kind of projects I like doing, the kind of ways I think. It'll be very different from anybody else's. My prompt for my bedtime stories for the kids reflects my values and the right amount of humor and the right amount of sarcasm and the right amount of, you know, science.

That's what I was going to say, the tone of voice. It's the tone of voice, isn't it, that people engage with? So how personal can one get, do you think?

Oh, incredibly, incredibly personal. So let's take me as an example. Yeah, maybe slightly sarcastic, overly honest, mathematician at heart, really cares about helping people, deeply believes in the value of evidence. I mentioned deep underlying human needs. If I just string that together into a sentence, that means ChatGPT behaves very, very differently to me than it would for you. Maybe you've got 4 out of those 5 and you've got 2 extra ones. That's, that's very, very different. And so part of this is understanding who you are and what tone of voice you value, but also what things you value. You know, I value evidence and truth and logic. Other people value creativity and innovation, and that's fine, but I want my responses to be grounded in evidence and truth and logic. They don't. And that's OK. Yeah, we have the ability to guide it in all these ways. We have to be clear who we are, what we want in order to do that.

And I'm really interested in these personality-aligned outputs, if you like, with profiling on different platforms. You know, the way in which we represent the self or re-present the self on Facebook is different to LinkedIn. It's different to Discord, if anybody can work that out. And I'm interested in all of these different profiles or selves or identities. We've got these multiple selves or identities, to stretch the point. And where do those reside? Do they still reside within us, or do they reside in the data that is, you know, kept somewhere on a— I don't know, is it on a server? It's, it's GPT. Is it in, you know, OpenAI's, you know, data banks? I'm not sure where my personality maybe resides now. Maybe it's in all those places. I don't know, I would love your sort of point of view on this because one of the things I've been talking about a lot is this move from authenticity to profilicity, a phrase coined by Hans-Georg Müller. And this idea of profilicity, and that's the main sort of mode that we're in now because we're communicating in digital rather than a physical environment, is fascinating to me. And it feels to me like it's going to be absolutely fueled by these generative AI tools.

Yeah, and I'd say it's, it's easier and easier to have multiple profiles. So I have something going on in my life right now which involves some legal issues, and I would say my emails on that topic are firmer and clearer and more thoughtfully laid out. My legal profile is clearer than it would have been if I didn't have ChatGPT to help me, and so it accentuates my ability to player profile I wanted to play but wasn't very good at. I posted a tweet earlier, which I was in quite a grumpy mood, and I wrote a little grumpy tweet. And before I sent it, I said, that's not really me, be positive. And I thought, I don't know how. So I put it into ChatGPT and said, reword this to be more positive, reframe it positively instead of being negative. And it did that for me, you know. So it just makes it easier for you to express yourself in the way in which you want to express yourself, because you can focus more on the idea that you want to communicate. Those legal notes I've very much my ideas and the things I want to communicate, but it helps you to stylistically present it in the right tone, the right, you know, framing. So just that alone means you've got more ability to communicate in more ways to more people on more platforms.

I think I've heard you talk before about how that might open up greater access for people who maybe struggled to express themselves, or they found themselves, I don't know, underestimated or looked down on in certain work environments. This could be a huge positive that nobody's really latched on to yet. You're the only one I've really heard talking about this.

Thanks. That really means a lot. Yeah, our books are dedicated to people who've traditionally been marginalized or disempowered in some way, and they're free to anybody who feels like they're a benefit to them and they don't have the cash right now or don't want to spend the cash. So, and they're $1 in all of Asia, Africa, and South America, for example. So we really want to help people with these books. And I think, yeah, In theory, this means that a student in Africa has the same ability to think through, let's say, audiences and strategies as the CEO of a major corporation in the US. In theory, that's absolutely possible by reading our books. In theory. Yeah. Or if you have bad English or you've not been trained to think as clearly as you might otherwise have been because you're not at that ability, then bad English and bad jumbled thoughts can be reorganized into beautifully coherent and thoughtful emails or reports or documents by ChatGPT very, very easily. So I think this is incredibly empowering to people if they have guidance for how to do that. Incredibly empowering. I hope that it really shifts the balance of power and brings much more diverse opinions into companies and boardrooms and entertainment and everything else.

Of course, there are some communities who don't really use the written words as much as they do sort of oral storytelling or whatnot. Are we on a journey towards a sort of a verbal version of this rather than the written version, do you think?

Yeah, absolutely. And there's a lot of technology that's progressing in that area, so I strongly expect that's next. But let me sell those communities on the benefits of ChatGPT, because I said earlier, any communicating or thinking task If you have a story that's oral, or the nub of a story that you're interested in, or a loose thought you've got in your head, if you put it into ChatGPT, which you can do using voice on the app these days, if you put it into ChatGPT and ask it to elaborate, ask it to challenge your thinking, ask it to bring it to life or come up with some analogies that would help you to tell that story, it really, really helps. So you don't have to give it much in order for it to massively amplify or accelerate the way you were going, or challenge you. I mean, it's a great use case as well. Like, here's a story I think is great, critique it for me. And it might say, well, that character's a bit weak, and that was a bit unexpected. Do you really mean that to happen? And we might have expected this. Have you thought about these options instead? You might say no, but at least it's helped you think it through. It's a great thought partner.

So if somebody picks up your book, and we should remind people of all the details in a minute, can they follow it simply themselves, or do they need like the David Boyle tutorial on YouTube or Discord or somewhere where you literally walk talked through it, how difficult is it to become aficionado in this space?

I would say it's very, very easy to get going, and that's all that's important with any new tool or technology, is get started, preferably have fun with it, and then keep going. Know that there's more to it, know that if you push yourself, you'll get better and better, and it'll help you in more and more ways. So I think very, very easy is the answer. Books are all full of like very simple, very practical examples. We describe them as cookbooks in the introduction, which is like maybe you want a dessert, go straight to the dessert section, or maybe you want a main course with veg, like go to that section. But you don't have to read it all the way through, although you can. Pick out examples, maybe in the Have Fun with ChatGPT section if you like, pick out examples. And every one not only is a practical example you can apply, but will teach you a lesson you can then apply in other areas. So it doesn't matter where you start, get started. And by flipping through and reading examples, even if you don't do the things in the examples, you'll learn things that will help you to apply it in brand new areas. I was working with a very senior entertainment economist who I love, and he told me the story the other day of how he applied ChatGPT, and he just kept going on about how It was thanks to the book, thanks to the book. And the example he gave was absolutely not in the book at all. It was a brilliant example, but it wasn't in the book. But he said, I learned this lesson in this section, this lesson in this section that helped me do this thing. And I was like, that is such a perfect example of this in action. Yeah, absolutely. I would say as well to anybody, this is an open offer to anybody. If you have a use and you're trying it out and it's not giving you good answers, it's something you're not happy with, first it's probably you, not ChatGPT. And my challenge is like, I will help you get to a good answer. So message me, the email's on the website, and I will tell you how I think you could improve it or like work with it better to get a good answer. I'd love that. I'd learn a lot from it as well.

So one final question. Is this the end of Google Search, or is Bard going to make a comeback and actually fight off ChatGPT?

My guess is Google's going nowhere. It's It's such a, such a great product and such a great company at heart that all the things they do are so, so beautiful to me and my values. So I don't, I'm not worried about that at all. I think it will change the nature of search because if I'm looking for help doing something, then probably ChatGPT could explain that to me in my language more clearly than any other method. So it will take away some search use cases, but I don't think it will fundamentally replace Most of search, no.

You are working in my absolute favorite arena to pontificate about at the minute, which is digital people. But you're an expert, so I would love you to just tell people what you're doing at DID and how you got there.

Sure. So, uh, DID is a platform the number one platform really for creating digital people. Um, and we allow people to, uh, you know, take that image that they've created through Midjourney or Stable Diffusion and the script they've made using GPT-3, or perhaps audio they've made using ElevenLabs, and bring that all together. And so we're like a— we think of ourselves as one of these core platforms in this sort of generative AI space and allow you to bring all those things together and, and create a digital person from get that face talking in the language, or, you know, get them saying what you want them saying. The thing that's becoming really interesting though, apart from that, is we can do it really fast. So we operate at kind of above 60 frames per second, 60 to 100 frames a second. So what that's leading to is now kind of an interactive person, not just some content that you're creating, but something you can interact with. And that's, I think, where we're sort of super excited about the future of of this space that you suddenly brands and all of us will have digital assistants who are helping us navigate the world and, you know, and get things done. So that's the space we operate in.

Amazing. So where did you do your PhD in interacting digital people?

I learnt on the job over a number of years. I was head of interactive at MTV back in the day. So I've sort of seen, you know, you've had glimpses of this coming through, you know, through the social media revolution. I think that was a place where we were all starting to first play with our identity, that you could have multiple identities online. Yeah, I've definitely witnessed a lot of this space without ever having done a PhD in it.

I know I'm finding a lot of people from sort of content audience backgrounds I spoke to David Boyle, and he has a similar sort of entertainment background. I think it's really interesting the way that people with all of this wisdom and experience, like yourself and him, are coming back, um, with this, you know, amazingly innovative technology. But actually, there's some real fundamentals that they understand, that you understand, about audiences and about content. It must, must be incredibly helpful because you could do anything potentially with these digital people, right? The question is, what are you doing the right thing them.

Yeah, and I think that one of the things that's really interesting in this space, just generally, not outside of just us, is the way that it's leveled the playing field. Suddenly anyone can create great images, great faces, you know, talking people. And what that does actually is it puts even more onus on the creativity side of it, like your idea how you're going to do it, what you're going to do it, how you're going to bring it to the table. We had a kind of recent bit of a hit with kind of meme online with, uh, someone was fusing together kind of the style of Balenciaga with Harry Potter and various other IP and kind of smashed them together in a really funny, interesting, quirky way. And, you know, tens of millions of views online. But what happened really quickly is there were loads of imitators. But literally by, you know, by version 3, you're like, oh, seen this. Oh, it's another Balenciaga mashup with an with another thing, with the Hulk. So for me, that tells me that it was the idea that was really the prime thing for that. It wasn't— the execution is becoming easier and easier, and that puts the onus on us to sort of be more inventive and more creative ourselves. So I think that's my— one of my key takeouts from all of this.

It's very interesting you're saying that because it resonates with something else that David was saying. You're both expressing a similar belief in that there's a positivity around innovation and democratisation of this new technology, that actually it's not so much that it's going to replace some of these creative jobs or tasks even, it's going to push us to be even more creative. We've just got more tools at our disposal. And actually it's about really thinking these things through creatively. So actually creative thinking could be the benefit of a lot of this automation.

Yeah, absolutely. I mean, craft skills will remain, but because those craft skills are becoming less important, it's almost like anyone who's a creative is now a creative director, essentially. You don't need people to do stuff for you. The machines will do the production for you. So where does that leave the human? Well, the bit the human does is, you know, coming up with the ideas. Selecting ideas is actually really important. You know, an AI will spit out hundreds of options. Knowing and understanding human culture and therefore understanding which of those 100 versions is the really interesting one, that itself is an incredible skill. That's part of the job that the creative directors do is selection and understanding where this fits with the, you know, everything that's come before and why it's interesting. So those skills will absolutely remain. In fact, they'll become paramount. For example, my daughter is studying animation at Bournemouth at the moment. And obviously she's studying a lot of craft skills around creating 3D objects and bringing them to life. But I've always told her, keep focused on storytelling, character creation, these bigger, more high-end things, because in the end, a lot of that craft stuff you're going to be learning will be redundant in 5, 10 years' time. But the stuff that's sort of uniquely human will become even more important.

Do you think that people are going to need some digital watermark or some kind of signal that a lot of the content has been created with these tools? Or do you think we're going to be in a position where, you know, it's for us to pass through it and work out what's authentic, original, what's fake, what's real? I don't really like the terms fake and real because it's all, it's all a blend of both now, and therefore it's all real in the end. But how do you think we're going to sort of work our way through understanding this?

Yeah, and this is a really, really important issue for us as a business. We don't want— you know, there are so many potential hazards out there. It's a very new, very powerful technology, and any powerful technology will have impacts. And the biggest one probably is around transparency and understanding what you're looking at. And that's one of the things we do is we put a little watermark, a little logo on every single video that gets produced. Boost because the minute you look at it and it says, you know, it says AI in the corner, you're like, oh, okay, this isn't really President Biden or, you know, whatever the thing is you think it is. And it's a bit like, like you said, nothing's real, nothing's— everything's fake. I mean, to a degree that's right, right? We go into a cinema and we're seeing flickering images, still images projected at a certain speed and it gives us the impression of people and we weirdly relate to it and cry and laugh at basically some still images strung together. Nothing's— it's all fake, but the context is critical, right? So you go into cinema, you don't think that, you know, Tom Hanks has suddenly de-aged by 50 years because you know it's, you know, it's cinema. You're in a cinema. I think where it gets tricky is where some— you're in an environment, let's say a news environment, and then you're seeing something brought in that's been sort of synthesized. So I think For us, transparency is the critical thing. And we, we all really, really want there to be legislation or regulation in this space, and particularly the industry to come together and create like invisible watermarks that aren't, aren't going to ruin the kind of visual experience, but allow the consumer of that content to understand. Bit like when you go into a supermarket, you turn over the packet, you see what the ingredients are. Every consumer should have the right to understand what went into a video. And one of the reasons we want regulation is we're already seeing some of our rivals out there boasting that they don't put watermarks on the content. So effectively, we're being penalized at this point for doing the right thing. We want there to be a level playing field. And this sense of transparency that would be created as soon as you do that, it really gets rid of a lot. I'd say 90% of the potential pitfalls. Once everyone just knows where this came from, what it is, how it was created, so many of those issues fall away.

Who is the creator now? Is it the human? Is it the AI? Is it a com— is there still authorship and, you know, sort of the genius creator now, or are we changing the way we think about this?

I mean, it's really interesting you say that. I I don't know if you've ever heard Brian Eno talk about how he makes his music, but since the '70s or '80s, he's been letting computers, as it were, make the music. But it's always him that selects the moment. He's the one— he says it's almost like when you grow flowers and you kind of hybridize plants, you know, and most of those hybrids are quite boring, but every now and again, there'll be an interesting one. It's called a sport. In the, in the jargon, you know, it has nice frilly edges, I don't know, does something interesting. And it's for the, the flower grower to, to spot that and understand that. And he says it's a similar thing with music. He lets the music procedurally make lots of things, and he'll be writing documents in the background or doing something else, listening out. And then when he hears a good bit, he goes, ah, that's the bit, I'm going to take that bit and use that. And, but, and I think it's a similar process here. It's like, like I said earlier, you're creating The computer is, quote, creating lots and lots of options, but the reality is that what creation is and what art and culture is, it's gatekept by humans, right? You have to go and sell that idea to ultimately someone else, whether it's a client, a creative director, the, you know, the world itself, other consumers online. So that process is a really, really important part of culture, and people forget about it. They just think 'Oh well, if a computer can spit out, you know, this amazing stuff, therefore job's done.' The reality is the job isn't done because this process of gatekeeping, you know, is part of the culture and part of the process.

I think people are finding that with ChatGPT, aren't they? The more they engage and have a conversation and share, the more iteration that happens, the better the output is eventually. You can't just ask it one question and then it spits something out. You actually have to guide it and advise it almost. And the output's malleable in that way. And I think people who spend the time on it are getting the best outcomes, right?

Yeah, absolutely. Or I quite often just take it and edit it. I mean, you know, it's doing the job of a junior writer for me, but I'm the editor and I'm going to find that, oh, that bit's interesting. Get rid of that bit, rephrase this bit, bingo. And it saves me a huge amount of time for sure, but I'm still— yeah, it's what we talked about with creative director. It's like my role has now gone up a level. I'm not the writer anymore, I'm the editor or I'm the commissioner, you know. And I think it's doing that for all of us. You talked about democratization. I think that's exactly what it's doing. It's putting writing powers, you know, illustration powers, photography powers in everyone's But the reality is not all of it's going to be good, right? Only the really good bits shine out because they have great ideas behind them and they're well executed. And, you know, so I think all it's done is just the bar has just been massively raised, but our expectations just grow with that bar. So it's kind of a weird arms race almost.

So somebody wants to develop a digital person,— or maybe more than one digital person. I mean, you tell me what people come to use the service for, but where would they start and how long does it take? And explain to people a little bit about, you know, how does this get done?

So our process is incredibly simple. You literally, you can upload one image. It could be some face you've generated, it could be a picture of you now, it could be a picture of you when you were 20 years younger, maybe a little bit slimmer, um, in my case. And then you either type in text, you tell it the language you want it to speak, could be 120 different languages, or you can upload audio. So it could be audio you've recorded or audio you've created somewhere else, you know, spoken word audio. And that's it, you press a button and, you know, literally a few seconds later, a video is generated of, you know, that face talking those words. So what it sort of seems to be replacing isn't yet, you know, full Hollywood, you know, film production. But what it is doing is allowing organizations particularly to create internal content around training, learning and development, marketing content at a fraction of the cost, you know, and much, much faster. And that allows you to put video where maybe video wasn't before. What were boring documents now turn into interesting videos and, playing out different scenarios maybe. So it's kind of, again, it's sort of putting video production into places where it wouldn't have been affordable before because, you know, the cost and effort of getting people and a studio and lights and a director and costume or whatever wouldn't be feasible. So that's where actually have a massive influx of corporate enterprise level clients at the moment. We're sort of signing a deal every sort of 3 or 4 days. And that's, that's, I'd say where it's, where it's going. It's going to be, you know, that's where our interest is, is how can this be used in business to, to make better content, make it more effective, make training content better understood, more completed, make sales content more clicked on, you know, all this kind of stuff.

What about the difference between a digital person as a presenter and an interactive persona, if you like. I mean, say you wanted to move into customer service and you've got your digital person, a bit like the neon people from Samsung or something. You've got to have a real-time conversation, like with a bot, haven't you? How do, or can you yet do, real-time question and answer sort of conversation rather than just presentation of pre-created material?

Yeah, so this is the next frontier really for this technology because we can generate our content at, you know, 3 times real time. What we've done is connected it now to large language models. So if you go to chat.d-id.com, anyone can just go and have a chat with ChatGPT, but they're just getting text. You're getting a real person, or what looks like a real person, actually talking to you kind of a more or less real time. So that's super interesting because as humans, we're just naturally attuned to faces. You know, text is a relatively new part of human evolution. So it stands to reason that a friendly face talking to you in the language you want, you know, you can select your own avatar. So you can, you know, people tend to receive information better from an avatar that kind of looks a bit like them. You know, that's been shown in a number of studies. So suddenly that kind of really awkward chatbot experience you have on websites becomes a fully fledged conversation with a digital person. And that's where we think this is, is headed. And it's a super exciting space where it's going to take a massive leap forward. You— businesses are going to have that— the knowledge that an LLM would have, you know, especially if you train it on your domain. It would know everything about your business. It's seen all your customer experience, uh, you know, scripts and so on. It's ingested all that, so it knows exactly the right answer to give in the way that the customer wants it, in real time, and in a much more friendly way. And, and also potentially, you know, more two-way. So it starts to understand and read from your face, you know, are you, are you happy with this response? Are you frustrated? You know, it can hear your tone of voice. So suddenly you're into a much more dynamic relationship, which we think will kind of massively increase customer satisfaction, loyalty, all those kinds of measures.

Are we going to get to a place where it's avatar-to-avatar communication and actually the human beings have gone off to, I don't know, play tennis or something?

I mean, yeah, I do sometimes think about that. Like many people, when this is being recorded, We haven't yet got to the end of Succession, but everyone's eagerly awaiting this final episode. But I do look at the lifestyle of the Roys, or these sort of super-rich people. They don't hold on the line when they need to renew their insurance. They have a person for that. They have people who go out into the world and do stuff for them. Their PR, their finances, their buying houses, you know, all the practical, boring stuff of life. Is done for them. And I feel like, well, that would be great to democratise that. I want to have digital people who, like you say, can go out in the world and interact possibly with other digital agents and negotiate on my behalf and, you know, get me the stuff I want without me having to, "Please hold, your call is important to us." Yeah, you can already foresee whole new ancillary industries and service sectors setting up around that, can't you?

You know, testing things, double-checking things, or passing things for you, or aggregating. You know, you've got several digital agents going out and doing— somebody might need to aggregate it, come back to you. Or what— one can see the whole industry exploding with lots of what might seem quite niche helpful elements to this now becoming everyday. I don't know, in like 20 years' time.

No, 100%. Donotpay.com is a great example of that. Helps you negotiate with companies and, you know, get you out of your subscription to your cable provider or something. Yeah, and I think it'll become emotional as well. I think we've talked a lot about the functional stuff of it. I think it'll become quite emotional. That digital assistant, you know, could operate more like a companion. It knows you, it knows your tone of voice, it knows— you know, we're starting to see there was that digital influencer who made $72,000 like in a day because she put a kind of a version of herself online that you could talk to for a dollar a minute. I think, and having a much more kind of emotional relationship with these, these agents, I think will become a thing as well. And I think there are potential pitfalls and dangers around that as well. But yeah, I mean, certainly we've seen for older people there, again, there are a lot of studies showing that even if they know it's not a real person Older people who tend to suffer from, you know, isolation do better and they thrive more having someone to talk to, even if it's a virtual person. We're just at the start of this really interesting journey.

Yeah, I know, I agree with you totally. I think there is that thing of, you might say it's telekly possible. So when I've talked to people before and researched sort of like resurrection versions of the the self, resurrection avatars, whatever. There is exactly that feeling amongst people, and I can feel it in myself anyway, that, you know, it isn't like real real, but it's really enough for you to get value out of it in some way. And actually, to suggest that it has to be as real as the physical world of somebody really existing next to you is too high a bar. And so these are all complementary things, right? So we can have all these additional experiences to the physical experiences. They don't have to be substitutes. I think people always expect them to be substitutes, and I think people recognize the telekinesis possible where they're real enough, they're valuable or meaningful enough for us to kind of buy into it. And a whole host of new kind of relationships will come about. And as you say, there's, you know, there's pitfalls as well, but there are with any, anything, right? There's pros and cons, advantages and disadvantages. And this is the interesting thing we're trying to work our way through it as humans.

Yes, yes. I mean, you know, we invented cars and quite shortly after we invented zebra crossings and traffic lights and airbags. And what we didn't do was just go back to being on horseback. Yeah, I think it's a very similar process that, or, you know, the printing press, actually following the printing press, there was a lot of what we would call fake news now, but very quickly it became regulated and, you know, trustworthy. I think we'll, I think we'll be in a very similar space with this kind of technology quite quickly. And I'm glad that people like Sam Altman are going to, you know, Congress and petitioning and saying, yeah, we want regulation in this space. We need it. And you're not being quick enough to make it happen.

What about privacy, Matthew, in terms of the amount of data that's being collected? Where does it actually reside and who has control of it? Is privacy even a thing going forward in this sort of world of synthetic selves and digital people?

Yeah, I think so. And I think— I don't think the regulation needs to change very much on that. We already have quite strong privacy regulation about where your data is stored, your access to it, your ability to, to view it and, and if necessary delete it. So I don't think it's— that's going to change very much. But you're right, as we talk to these assistants and digital people more and more, they will get to know us and they will understand us. But, you know, I mean, I have a huge amount of data with Google, all my emails for 20 years, 15 years. I don't think this is a new problem, but it is a concern, obviously. And, you know, bad actors coming in and using that in a manipulative way and, you know, such and such. But these are problems that we've solved before and we'll solve again.

Do you think a politician might ever— I was thinking about DeSantis launching his presidential candidacy on Twitter the other night, and, uh, yeah, not very, not very well. Yeah, no, exactly. That was a rather interesting experiment. That wouldn't have been the time to run an experiment in my mind. But anyway, it just made me think, you know, do you think in the future a politician will announce their candidacy through their digital persona, or, you know, create slightly different versions of their persona that go out to do different roles in the life of a politician?

I don't think so. I think humans are different to digital people. No matter, you know, even if we get to some kind of level of intelligence, there'll be, it'll be a different kind of thing to being a human. So, I actually don't think so. I think people would see that as very, very inauthentic. And authenticity is everything in politics. You ultimately, you believe in a person making decisions, not them delegating those decisions to some kind of AI.

See, I think they will. I think this is what will happen. You know, it's interesting discussion in a technocratic kind of future. Politicians are the ultimate performers, aren't they? And so they have these personas. It's going to be very, very interesting. Because, I mean, one of the things to think about is that, is authenticity still the benchmark? Because once one is profiled, and indeed profiling oneself in many, many different ways, and has multiple presentations of oneself, kind of authenticity, it's kind of been left behind really. And it's almost like the old world benchmark for a new world of communication.

I mean, if you were saying that about an actor, I would agree. If you were saying about many types of people, but I think in the end, I think you're voting for a person, not what they might, an AI might or might not twist their words to be. I mean, do I think a politician would get advice from an AI? Yeah, absolutely. I think there will be AI advisors or information that comes from an AI and that spots patterns early in voter behavior and can have a view over the political world greater than an individual, yeah, absolutely. But I think in the end we want to connect to human beings in terms of our politics, not machines. But I mean, who knows, maybe, you know, maybe at some point in the future, maybe they'll be much cleverer.

It's interesting to probe it because you, you, I think what you were saying to me is that it's very much in the role of assistance. It's not going to take over in terms in terms of like personality and—

Well, this is where you touch on the sentience issue, right? And everyone has a slightly different take on when will we get to artificial general intelligence, AGI. And I saw Sam Altman slightly move the goalposts a little bit towards, well, it's just very, very smart at doing very set things. But for me, AGI is about ultimate human flexibility, that one minute I can be digging my garden next minute I can be making, doing a Zoom call, and then I can be reading a book, and then making some food. There's something about the flexibility of human, the human mind, human intelligence, which I think is a long, long, long way off for computers to do. So I don't think we're going to get to that point, but I'm a big fan of science fiction author Iain M. Banks. And, you know, he had a lot of the characters in his books are sentient AIs. And they have all the foibles and issues that, you know, we have as people. They have personalities, and I'm sure that will be where we will end up. But I think that probably won't be in our lifetime.

No, no, we probably won't. And of course, then that takes you on to whether they are deserving of any rights, you know, and not necessarily the same rights that maybe a human intelligence has, but rights rights that a non-biological intelligence might have, in the same way that animals have certain rights. You know, they're not the same as humans' rights, but it really opens up, again, a whole new, you know, aspect and adjacent industry really around these new creations. How's it going to affect us interfacing with machines, like the human-machine interactions?

Yeah, so I think it's really interesting when you look at software at the moment, it's we've kind of reached the end of the graphical user interface. You know, that's been a— we've had it for 20 years or maybe a bit longer, and it's starting to reach its end. The capability of our software is so great, and yet we're still navigating on Salesforce, you know, dropdowns and menus. And yeah, it feels awkward. It feels like we're just about to reach a point where you just tell the computer and maybe tell your assistant, you know, instead of using Salesforce, I just I want you to tell them what you want, not to have to speak the computer language effectively. You know, okay, yeah, we're not putting in line-by-line code, but we are still navigating interfaces and programs in a very old-fashioned way. So I think that will be one of the impacts of these digital people, will be to allow us to command and get what we need from our software. And I think that will have a big, big, big impact on the world of tech as well.

You know, that answer is a good answer to a question I've written down here. What does the synthetic media revolution mean? You know, how, how is it going to change us? Because it's a point well made about the old-fashionedness of the, the user interface.

Yeah, we, we've had graphical user interface for 20 years. We kind of updated it a bit when we got our touch phones, but fundamentally we're still doing things and getting the value from our software in quite an old-fashioned way. And I think that will be one of the things these digital people do. They'll change the way we interface with everything digital and that, you know, everything from computer programs to how we interface with our bank and so on and so on. So, you know, you just want to get what you need from the program. You want to know what the sales figures are for the last quarter or what are Bob's latest deals. You don't have to go into Salesforce and navigate 10 submenus. So I think that will have a radical shaping effect on the technology industry specifically as well.

And what happens when we haven't got our smartphone devices and we're in the world of spatial computing and ambient communication? Presumably these assistants, it's their realm really. Surely they'll come into their own where they're not constrained by, you know, like a screen of a mobile phone or something.

Yeah, I mean, it's— I'm dying to see what Apple release in terms of their kind of VR headset type thing. Yeah, I mean, well, you're looking quite far in the future. I think we're going to be screen-based for quite a while, personally. But yeah, maybe I'm, I'm just a bit old-fashioned like that.

No, it's interesting to see what you say about that.

What do you think? How do you imagine it?

I imagine it being much more gesture control and ambient computing environment, because there's something about the phone device, looking at mine now, which is quite distracting, actually. I'm drawn to that, and I'm distracted from everything else that's going on around me. And all the things you've just been talking about with the assistant services, they're doing things around me. They're like in my world, giving me new perspectives and information. And it just feels like suddenly you have to go, shh, and it's all like on the phone. I just think that can't hold up. And so I think there'll be much more gesture, voice control, and it'll be a more ambient communication.

I mean, that's interesting. I was talking to a woman yesterday who's developing haptic gestural— she refused to call them gloves, but as far as I can make out, there was sort of something, something wearable on your hands, so I'm calling it a glove. But she was talking about particularly blind people, it could give them you know, a form of Braille so they can suddenly read things or be given information. So yeah, I think all this ambient computing is, is going to be there, but I just feel like it's quite far ahead. I feel like we've been longing for this for a while and been willing it into being, but somehow the computing power and the battery power hasn't unleashed. I think, I think the batteries might be the thing that's really holding it back. We our battery technology hasn't kept up anywhere. The pace of battery technology is very slow compared to, you know, development of chips and microprocessors.

Or it could just be vested interests that are going to hold us back. The timeframe might be dictated by the profits from certain devices, and maybe it's not until they think they've eked that out that we'll move into different formats. No, I think that's going to be really interesting because, like I think I said to you before, I was really— years ago when I looked at Edward Fable's Lucy character, I knew that. I was like, it's not about social robots and robots going into the home. It's about these characters. They're so engaging. And whether you think they're like an assistant or an extension of you or whatever the role of the digital persona or the interactive AI, just once you anthropomorphize it, it's just so interesting and engaging.

Well, this is one of our themes actually, is about anthropomorphism. It's a really interesting thing because you know, we see faces in clouds and we, we, we see intention in things that are just random. You know, there's a really famous study in, I think, the '30s or '40s. They just created a random animation of shapes moving around. And when you watch it, people will say, oh yeah, you know, the triangle's bullying the circle. And, you know, we, we, we seem to be able to project onto objects, inanimate objects, things. So imagine how much you're going to project onto something that's not only animate but is talking back to you in a smart or slash intelligent way. You know, of course you're going to start to anthropomorphize. I mean, I say thank you quite regularly to GPT. I mean, I know, and I say please. I don't know why.

I think most people do.

Yeah, you will feel emotional about it. You will connect to them. You will. Even, even the most logical person, you know, I know ChatGPT is not a human. I say please and thank you. Why do I do that? You know, I'm, yeah, it's inevitable.

Thank you for listening to The Future Review hosted by me, Tracey Follows. Check out the show notes for more info about the topics covered in this episode. Do like and subscribe wherever you listen to podcasts. And if you know someone you think will enjoy this episode, please do share it with them. Visit thefutureofyou.co.uk for more on the future of identity in a digital world and futuremade.consulting for the future of everything else. The Future of You podcast is produced by Big Tent Media.