David Boyle on DisrupTV 3rd Feb 2022

Okay, talk about owning your own destiny, man. If you want to own your destiny for the next 10, 20, 30 years, you better know a little bit about generative AI. And. And our next guest wrote a book about it before it was a white hot topic. So talk about David being a true trailblazer. David Bore. Yes. Director at Audience Strategies, an agency that helps brands use deep understanding of their audiences to drive growth. Exactly what our past guests, our CEO said. Talk to your customers so you understand their needs. David has 20 plus years of experience. Looking at David, looks like he started when he was 8. Using technologies to develop audience intelligence and strategies capabilities in creative industries. His work has changed the culture and economics of world's most prestigious entertainment and luxury brands including EMI Music, Harper Collins Publishing, BBC Masterclass and Harrods, to name a few. David. David's an author of Prompt, A Practical Guide. You see it over my head. Over my head, A practical guide to brand growth using chat GPT. In the book you're going to find about 161 worked examples. So it's a really incredible instructive book. Prompt is the ultimate guide to growing your brand through generative AI powered marketing and product strategies. And it will provide you with the knowledge and skills you need to harness the power of again, generative AI to get ahead. You can follow David on Twitter. Welcome, David, to the Shrub tv.

Thank you so much. Great to be here.

Thank you for joining us during dinner time. You know you're dialing in from London, so appreciate it. It's very nice.

Pub time, actually.

Pub time, yes.

Well, hey, what better can we talk about pub time than chat GPT? It's in every single chat conversation everywhere, in every bar. But hey, the more important thing here is really I want to know about your background, right? It sounds like you've got right brain, left brain. We're talking analytics and creativity. Those are two things that we haven't always put together. It's great intersection of interesting skills. But how'd you get from here, like from music industry to talking about data driven insights? I mean it's a, it's an interesting journey.

So yeah, I think you get there by making an awful lot of mistakes along the way. I'm a data nerd at heart, mathematician by training. I have a weird passion in life which is like helping data to be very, very useful. And I think I ended up in the creative industries, not by choice, by the way. And then I realized how magical it is when you can help artists or labels or creative people to use data because they normally don't want to. They're normally quite skeptical. And so that brought me this extra kick. And the impact that the creative industries has on people's lives like that felt special to me. So I went, I left the creative industries and thought, I'll go do this again in brands. I realized it's just not as much fun, it's not as impactful. So, like, creativity is where my heart is. And those hard lessons of, like, messing it up because you think data's the answer. And it turns out that creativity is important and finding that balance of so much fun.

That's awesome. That's so, it's, it's. It is remarkable. So Open AI was founded December 2015. So it's seven years old and change, but it was really December of last year, actually November of last year, when the world, within five days, it reached a million users. The world in November 22, I believe, of 2022 is when most of us really, you know, saw the power in generative AI and in Dall? E and others. So 2022 in general was a good year in terms of exposing the masses at this technology, of creating new content. And so how did the idea of writing the book prompt come about? And what was it that allowed you to be a really, really early adopter of talking about this phenomena?

Well, I got to start by saying it was a bit of a surprise to me as well, because I've never been a particularly early adopter of technology. That said, a lot of businesses I've worked with think I am, but I'm like a fast follower.

Breaking news. David is a fast follower.

Wait for something to be proven, wait for it to be practically useful, wait for the hype to die away, and then help people to use it. So I've always pushed tech, always to scale, to democratize access to data, always. But I've never been at the forefront. But this new tech drops. Yeah, I played with the GPT stuff before chatgpt came out. It's fine. Chatgpt drops. I'm like, oh, this is different. Because right now, today, with no hype, it can change everything about how we work. So my co author and I, who've worked together for 15 years, we played with it. We were sending each other examples. You start by writing poems or explaining technical terms in the style of Snoop Dogg. And then eventually you're like, oh, this? We all do that, right?

Mine is Dr. Dread, but okay, yeah, there you go.

After a few days, we had this moment and we called each other. It's like, this is not just a joke.

Right.

This actually can change everything. We spent 20 years working really hard on.

Yeah.

In which case we should stop what we're doing and just work out how to do this properly, shouldn't we? Yep. And then our instinct is always to help other people to share or to, like, democratize. Like, we should write a book. Right. Years of expertise in craft and process in building brands, and we just, like, worked really hard to work out how ChatGPT could help. We wrote it all down and we published 23rd of December so that we could have a quiet Christmas with our families. I think there's a big lesson there, though, which is, like, if you know what you're doing, years of, like, messing it up and learning tough lessons, it's relatively easy to apply new tech to that. But if you don't, if you've got no guidance, no strategy, then you're going to be lost. And I think that's an important lesson for all technology, especially AI.

Literally within seven years, one week. OpenAI was December 11th and you December 23rd, launch your book. I want you to share with us how you approach being on Disrupt TV and your experiment in terms of our podcast and how you could best add value to our listeners, because I thought it was super interesting.

I think of AI as being like an electric bike for the brain, so it makes it quicker and you can tackle bigger problems more easily than you otherwise would have done, but you still got to pedal.

So.

So what I did was to say, all right, I need to work out what these guys are going to ask, what types of topics they talk about. How can I do that in the most leveraged way? So I pointed an AI at the videos of your previous shows, and then I asked it. What type of questions do they ask? Great. I put that into ChatGPT, along with a synopsis from the book and my career, and I said, hey, what questions might they ask me? And then also what answers should I give? And I know you haven't used any of them, so thanks for that.

We want you back on the show, so we're going to save that for your second.

No, no, no, no, no.

We're improving your LLMs by adding new data. That's what's going on here.

That's right. That's right.

I think one lesson I learned like that, for example, one lesson is like, you don't need to recreate the wheel on this stuff. I don't need to listen to all the shows to work it out. I can get that in my head and I can, then I don't need to work out how it applies to me, because an LLM can do that for me. What I do need to do is work out which questions make sense, which ones need rewording and which ones need enhancing. And that's a much better use of my time than like the first draft.

I like your. You still need to pedal. So even if it's an E bike, you still need to pedal. Paul Graham talked about just AI in general as like a calculator for thinking and writing. But I like your bicycle and pedal analogy more. Go ahead, Ray. I know you're dying to ask a question.

I'm going to actually take one of the questions that was generated just so that we can actually make the LLM feel good about itself. But there's a lot of talk really about is AI taking away human creativity or is it actually improving human creativity? And what is that in terms of human? When do we insert a human in the process? Because a lot of times if you don't insert a human in the process, we end up with Skynet and it's like, oh, wait, what happened here? And so the human in the loop is intentional friction, an intentional set of controls. And so what do you suggest for that, given all the experimentation you've done, the fail fast you've done, and of course the lessons you've learned.

I'm going to talk just about my area of expertise, which is building brands, understanding audiences, and then working out how to do things differently to build brands. And I can tell you after 20 years of doing that first ChatGPT or LLMs or AIs like, help immensely at every step. That's point number one. Immensely at every step, Full stop. That's remarkable. But it doesn't know where to start. You've got to start it. You've got to tell it what the problem is. You've got to define the category or the audience or the business problem in your own words. Not the client's words, by the way, because they're not in your words as a strategist. And then you get a good answer sometimes, usually maybe. But you need to nudge it iterate, update your prompts, and then you need to know where to go next. Where do I elaborate? Where do I reject the answer? Where do I enhance it with my own perspective, add special knowledge I have about the brand or the marketplace? And so there's quite a lot of human involvement there. At the start, it's 100% human. At the end, I think it should always be 100% human. I own that document I sent to you in the middle. That's where the LLM can help you quite a lot. And what it does that for me is a very firm lesson in my.

We're still pedaling on this E bike. Vala.

Yeah, yeah. But, but, but what will the E bike look like a decade from now? You know, this company that we're talking about or this technology, you know, it's the last five to 10 years that we saw incredible processing, storage, access to hyperscalers that allowed us to create models at a fraction of a cost. So take US David, to 2033 and, and let's take us to a typical marketing organization who is, who was an early adopter of AI. And for the 10, for 10 years, like does PR function exist? You know, I don't mean to, you know, I always think that you're more.

Does anyone need to write press releases? You know.

Well, yeah, yeah. Or web copy press release. I mean the, the content that still creative but, but you know, ultimately over time looks a lot very similar in nature. It seems like very deterministic and it's in its, in its construct. But, but your thoughts 10 years from now, marketing and use of, use of technology technologies like chat, GPT.

Yeah, I have a very clear view on this which is that it just enables everybody to do more, take on more, build better and more brands, full stop. So I think, yeah, many tasks that today need me to start it, manage it and then end it won't need me in the same way in future. They won't. So I'll be freed up, right. What do I do with that free time? Do I go sit on a beach or do I do it, take on extra clients and more brands and personalize it for more countries because the. There's infinite possible complexity in the world. I could do 10 more brands or personalize it for different countries or different audiences or different campaigns or strategies. And at the moment there's so much like slightly lazy thinking, like well, we'll just do Europe and treat Europe like Europe, right? The future. We can personalize it for every single country so there's infinite possible complexity and optimization and products that could be launched. We'll just do more. So that's.

Yeah. And small businesses, large enterprises, I mean they're all going to be empowered. Right. So I've got a realtor friend in Austin named Philip, right. He's showing houses to a bunch of clients. Like he can just basically generate the whole realtor flyer like on the fly. Doesn't have to do that anymore. Run it in through, you know, a GPT to do it with a large model and basically know exactly what words to do to target each individual. Right. I mean, it's personalization at scale.

Right. He can take on more categories of property or more towns or other businesses at the same time because he can handle more now. Absolutely. I think the big problem there is if you don't buy an electric bike or adopt it or work out how to use it, you're in big trouble.

Right.

Because the guy next to you is suddenly like pedaling much faster.

Oh, the arms race effect.

Yes, you're in big trouble. And I think this is a generational challenge, which is that older people are less likely to want to do this. Younger people are. That's incredibly empowering. Which means that the young intern who never had a voice at the table can now take a strategy paper to the chief executive, which is very, very good. I had a business I was working with and the CEO gave a talk about how we all need to use AI. Now this is a multi billion dollar business. And somebody messaged me like, hey, the CEO's just given a talk about AI. You wrote that book, right? How can I use it? I was like, do this. She's like, I got so many brownie points for that, you would not believe.

That's awesome. That's awesome. That's awesome. You know, I have a daughter who's a sophomore in a very good business school, very expensive business school. I wonder how many. Yeah, you know, I won't go there, but yeah, very. Let's just say I get a 5 Series BMW every year, but I don't get to keep it. What do you, what do you have insights in terms of curriculum that's being taught to these interns that are now joining the workforce in their early 20s. Are they learning about or do they have the mathematic critical thinking and even exposure to these technologies at an early. At an early or not, I suppose early, but undergraduate level, college, university?

Well, that's a good question. So I think you've got colleges who are like rejecting this thinking. It's like plagiarism. That feels wonky. Most professors I talk to, like, no surprise, are saying, how can we use it? Like, how can we integrate it? One of the one big college I'm talking to is like, can you teach a class on this? I'm like, yes, I can. But you know what, actually it needs to be integrated into every other class. That's the big impact. Because if I don't know how to write a press release. I certainly can't use ChatGPT. If I don't know how to do a marketing strategy. I certainly can't use ChatGPT. So it should be like learn how to do the thing and then learn how to do it quicker and better than everybody else.

That's great advice. I love the fact that what you're, you're, you're, you're going to be a best selling author if you're not already on a white hot technology. So invitations for you to teach and speak is going to be, is going to be pouring down on you. Any, any lessons learned going through the process of documenting 161 some odd cases and like did you have blind spots at the beginning of the book that you were able to identify and address in the book? What were some of the big takeaways?

Yeah, I think yes and no. So I think one interesting thing is we stand by everything in the book. So totally expected that not to be the case.

Absolutely.

Expected to push an update out after.

A week and then after a month.

And that hasn't happened.

Right.

Because the fundamentals haven't changed, which is how do I build a brand and how can it help me other nuances to what the best prompt is. Yeah, definitely. And other things that it can do that I didn't think it could that aren't in the book. Yeah, definitely. So we're going to push an update out, but it hasn't changed the fundamentals. I think the lesson there is start with a process in the real world, know the job to be done, look at how it can help you and then move on to the next one. But work out how it can help. Get really good at doing that job with the tool, then move on to the next one. So master something and then expand and expand through different jobs.

That makes a lot of sense. And one of the other things that we're seeing is now we're operating in machine scale. As humans, are there certain things we have to do differently so that we humanize this operation and, and we don't feel overwhelmed, which we'll talk to our next guest about later. But so we don't feel overwhelmed and completely like, you know, taken aback by like the immense pressure of trying to keep up with a machine?

Yeah, and I think I said it earlier, but I just think we're all going to be busier. We're going to have more topics on our desk for more brands in more countries, more products. That's really stressful actually. That's been what's happened over the last decade or two, I think.

So the stress levels are going to go up. You think?

Yeah, every individual task will be easier and more. And like I'm happier doing my work now. It's less stressful for me. Like it's easier. Right. But I've got more. Am I happier overall? I don't know. I'm having fun.

What a delicious paradox. Yeah, it's easier but the volume is higher. So you're not maybe shifting the source of stress from one area to another, but it still exists. There's no question that you know, as in a more, in a digital first, more decentralized, hyper connected, knowledge sharing economy, every stakeholder employee, customer, partner, community is asked to do more because you have, frankly, you've had a supercomputer in your pocket for well over 15 years now, 2007. So you know, it's democratizing access to information to improve decision. Velocity comes with. Was it Superman? Was it Spider Man? With great power comes responsibility. Yeah.

Peter Parker.

So yeah, Peter Parker's uncle was right. Would you, as a cmo, let's say you're recruiting into your marketing team and this is like a digital marketing trailblazers like you are in the interview process, would you ask the candidate, any candidate anywhere in marketing, have you used chat GPT? And if they said no, would you still consider him as a candidate?

Wow.

Whoa.

Wow, that was a definite.

He didn't even. Wait a second, his head was moving because obviously he's trained to anticipate questions. Wow, that was quick.

Okay, on one hand I have sympathies, right, Because a lot of my good friends and professional relations have not bought the book. I know that because I see every transaction. Right. On the one hand, I have massive sympathy for people like, we're busy. We're, we're crazy busy.

Right?

The economy's tough, we're distracted. Who's got time for a book? Okay, I get it on one level. On another level, if you believe half of what I say, like I talked to a professor yesterday and he was like, this can save 20 or 30% of every executive's time. If you think we're even vaguely right, even vaguely right, you should do what we did in December. Stop everything and work this out. And if you don't, or if you haven't, or if you haven't shown any curiosity, I have questions.

Well, hey, let's go the other direction. ChatGPT, all text, all conversation, stable diffusion. Talking about how we actually take images right on the creative side that's going to be just as exponentially impactful. What do you think about stable diffusion and the ability to actually take those images and start creating, doing the creativity from inpating to outpading to, I don't know, generating image to image translations like boom.

Video. All of that. I mean all stunning. This is all the last six months we've been seeing these. I mean David probably saw four years ago, but the rest of us, I don't know.

Is that really David? Like we don't even know. He's got this avatar so embedded, right? I mean he's got pixels in the. In the raster level moving around.

Like trust me, Ray is trying to replace me with some generative AI co host and it's going to happen. So my days are numbered. But go ahead.

309 episodes, enough training data.

Oh, totally.

You know what? We should just have the show run on its own now. Like we'll just go and watch and break out some popcorn. I'll come over to your house, you know, volume.

We'll just watch our own show, 309 and almost thousand guests and have it write a book. Like what are the. What are the leadership lessons? Business lessons, Innovation lessons actually.

Yeah, that sounds like great. Intern project.

Thousand words. An average business book.

So we are looking for a summer intern to use chat GPT to write our Disrupt TV book. Anyone interested, you can.

I would give the entire book revenue to whoever helped create that. You know, I'm just saying. Just saying.

We're here with David Boyle, director of audience strategies.

You gotta come back.

You gotta come back. You can follow him on B, E G L E N. Is that I? No, it's an L. And thank you so much for being here.

So thank you. David, congrats on the book, really congrats. That's amazing.