Google’s Global Head of Creative Innovation, Matthieu Lorrain on AI & Creative Functions

Welcome to BrandBots, an AI marketing podcast designed to teach you practical strategies for implementing AI into your marketing business. I'm your host, Nick Taylor. Today we welcome Matt Lorraine, the global head of creative innovation at Google to the show. With a career spanning over 16 years at Google, Matt has cemented his status as a pivotal figure in the intersection of emerging technology and storytelling. Matt works across product design and partnerships to invent the next generation of content experiences. It has been leading global teams to develop cutting edge creative platforms with iconic brands such as Gorillas, Star Wars, the NBA, Louis Vuitton, Space Invaders, Lego, and Nike. Matt also works closely with Google's R&D and product teams, including XR and DeepMind, to incubate early use cases of creative experiences using Google's creative technologies. This episode, Matt shares insights from his recent talks at South by Southwest, discussing the future of technology, particularly the integration of AI in creative and business processes. He delves into the very interesting concept of liquid content, where he envisions a future where content is highly personalized and dynamic, adapting to individual users' needs and context. Matt also addresses the challenges and strategies for businesses integrating AI, stressing the importance of starting small with internal processes and pilots to manage risk effectively. His perspective on AI integration in business is really practical as well as forward thinking. So this conversation should offer some very valuable guidance for any organizations looking to navigate the AI landscape and integrate it across teams and functions. But further ado, please enjoy this conversation with Matt Lorraine. This episode starts now. Alright, Matt, good to see you. Thanks for having me. Of course, you just got back to South by Southwest. Absolutely. And you have a really good tip, which I recommend to everybody going to South by Southwest. Don't drink. Yeah. Don't drink, try to sleep, stay hydrated. In my case, also, I met it Schultz. Yeah. Yes, sweet. I was there for 36 hours. Oh, okay, yeah. Normally it's three, four days and it's party all day, all night. And you fill your mind with stuff all day. It's more than humans can take, I think. I feel so. Those big conferences, you're right the first day, you're excited, the second day, you're a little more quiet and usually the third and fourth day, you're wondering why you stayed that long. So I tried to avoid that these time. And 36 hours was for personal reasons. I didn't have much time, but yeah, that felt right. And exciting to be in South by. A lot of discussions on AI and the future of any technology you could think of. So that was great. Great, great. Did you give us just a little background who you are, what you do? Absolutely. Hi, I'm Matt Sula-Han. Matt, almost of my friends. I work at Google. I am the global head of creative innovation in a partnership team called EBX. And I've been at Google for a very long time, 16 years. Wow. So I've seen a lot of evolution, a lot of product launches, a lot of hype cycles and technology cycles. My job right now is to be a bridge in many ways between emerging technology and big partners and big IP and storytelling and creativity. So I pretty much get to work with product teams at Google when they launch new creative technologies such as augmented reality, generative AI, in the past video, YouTube, interactive video, etc. And my job is to work with those product teams to understand what are the use cases for those technologies and how it can be used to solve business problems, tell their stories and gauge with fans. It comes with some great perks. It's very exciting to be the first one playing with the toys. It's also a real privilege to work with big partners, great IP, for example I worked with Star Wars, with Marvel, with GrowwRitters. It also comes with the territory, it comes with the challenges which is when you work on antestate technology, you are a bit of the guinea pig for it, number one. And number two, because I'm a bridge, as I said, it also means a lot of contact adaptation because product teams, engineering teams speak, think very differently that brands, agencies and every day is also that challenge of how do you translate and how do you speak both languages and how do you understand both contexts and how do you align those. Yeah, that makes sense. So you spoke, I had self-by-self-west. What, three times? You had three different? That's right. I had two official talks, panels and one side. The first one was a great talk on designing the future, with great co-panelists such as Chris Downe, who is a chief digital designer, sorry, at Mattel. So we talked about how do you design the future innovation, how do you organize your team, your organization, to maximize your innovation output, how do you design meaningful innovation? Yeah. I had the second talk more on virtual stages. Cool. Yeah, I've done a lot of work on augmented realities. So what does it mean to combine physical and virtual environments? Wow. And I had a third talk at the House of Canada, with Chris actually also, and a few other people, augmented creativity, which I feel is going to be. Yeah, I feel like I is a choose your own adventure, which of the three talks should we dive into first? Yeah. I feel like the first one, which is implementing AI meaningfully into your creative team, right? Yeah, designing the future definitely involves those days thinking about where AI is going, but also how can you use AI in your process? We can start that. Yeah, absolutely. So how, I feel like we're in a time when AI is still, it's not new necessarily conceptually, but I think people are now getting around to realizing I need to integrate this into my business. If I have a marketing business, if I have a business that's remotely creative, and I feel like some people do feel like they are behind. So what do you think are some of the keys to integrating it at the speed where people need to? I think it's impossible to be a year ahead. I think the best you can hope for is to be like weeks if not months ahead of the curve, but for business owners, for people running marketing teams, what do you think are the things that they really need to keep in mind at this venture or at this stage? It's a very good question. AI is, I think in many people mine right now and they are trying to figure out, okay, what does it mean? Always are going to change my business. How do I use it to drive better impacts, better creative, better engagement with my fans or my consumers? What I'm seeing, and I've seen a lot of stats and insights on it, is that almost everybody at that stage has tried AI. By AI here, I mentioned specifically, I prefer to large language models, which is tools like Gemini and Chad G.P.T., but also diffusion models, which is tools such as mid-Journey for images. Almost everybody has played with it because it's fairly easy to write a prompt, but what I also see is that very few people actually have fully changed their processes or really one deep enough to understand how to use it for the long term. The recommendation I usually give people is, number one, you're going to start using it in your internal processes. It's safe, probably much, to play into experiments. Maybe you may want to start pilots or carve a space where you can experiment without having any risk of scoring up pretty much on your content or on your relationship. That's the number one. You can't cheat your way to other expertise. You need to upscale yourself. The way you can do it is by practicing, by experimenting with real use cases and also educating yourself. I think people need to start the same way that we've digital. It took a while for people to really incorporate digital. It's been 25 years now that it's been around and you still have some time to divide between regular commercials, for example, digital. I think it will take time, but again, practicing, educating yourself, understanding the limits, I think it's important. There are a lot of hopes around JNI, but there are also real challenges right now. In terms of copyrights, ethics, what are the risks, what are the limitations, but also understanding the creative limitations. It's not a silver bullet for everything. Last but not least, I think as you said, things are moving so fast. It's pretty incredible. Every week there's a new tool, every week there's a new method. The main recommendation I also tell people is pretty much you need to change your mindsets into a constant learning approach. New tools are going to come again all the time. Some tools will rise, will become very popular maybe for a year and then might be replaced. You need to organize your team and aware where those teams can be very nimble, very flexible, and very quickly learn and integrate, but also pivot to our new practices. I think that makes a lot of sense. I'm finding people are stuck between trying to learn advanced methodologies with the current tools like Chatchee BT and like Mid-Journey versus Shining Neutoy Syndrome and constantly chasing. Okay, my God, there's all these new tools coming out every single day. I have to learn all of this stuff. But I do find it's meaningful to have a dual approach where learn the tools that you already know, but learn them really well, but also stay on top of what are the things that are coming down the point. It's very true. And I think you outlined that very well. It's worth investing into the main tools that are right now being the most successful again in one of the large and large models. You have Gemini from Google, you have Chatchee BT, you have Claude, those are probably the main tools for images, same thing, you have Mid-Journey, you have Daly, Google also worked on images. They are like the open source tools. And so it's probably good at it to pick the right, the few tools you want to invest in, go deep, understand them. in the meantime. maybe have like almost like a different team or a few different people whose job is to do quick evaluation, quick testing. So it becomes like that kind of dual kind of work stream or take a few tools, invest in it, learn how you can use them in your pipeline, in your workflow, in your process, while constantly looking at the new one, the new tools, the new platforms testing them, and the one you think are really good, then you can bring them back into that deep dive process. Yeah, how do you advise business owners and business practice leaders who have them, who have marketing teams, who have design teams? There was a time, you're not that long ago, where it was all about agencies and marketing teams going digital. Now they have to go AI. And it's for some companies, it represents a real challenge to get people to not only use the tools, but understand them and the implications of them. For business owners and marketing team leaders who are stuck with that challenge, how do you advise a approach integrating it within their teams? You draw the parallel with digital, I think that is a good inspiration for me, good reference, because if you look at digital, for many years you had digital marketing teams on one side, regular marketing teams on the other side. And progressively people started to realize that, that doesn't make sense, it's marketing. And so marketing has an other home component, has maybe a TV component, you may have a local ads component, and you have a digital component. Across everything, I think AI, we're going to get into the same situation, where maybe at first you may want to have a few people in your team specializing AI, but over time everybody's going to use AI the same way today as everybody use PowerPoints or Photoshop. So I think that's my first point. Be ready for your whole organization to adopt AI in different capacity. Two is, as I said, because it's early, it's probably a good time for pilots. You need to call a safe space where you can manage the risk, you manage the challenges. The different ways you can do your pilots, for example, you can decide just to do an internal pilot. So okay, let's do a project, we're going to look at how AI can help on the development process. Something that is not publicly very visible, but something where we can bring already a lot of value. Or you can try to take a small pilot, which is okay, we have maybe a lower risk campaign or a smaller market. Why don't we try that market on that smaller campaign, how it goes before scanning up to my big splashy moments. I would avoid going straight, the big super boom moment, maybe you may want to test it before a little bit. That makes sense. What I see also is a lot of companies right now are still hesitant to use AI in production. So something that is going to go out there and engage with consumers. But a lot of them are already using it internally again in the pre-production and in the creative development process. That for me a good example of edging your bets and managing the risk is it's very low risk to start thinking how you can have better ideas, better strategy, better insights, and better even prototypes internally. And once you better understand then all the implication, that's when you're going to move into production in big launches and maybe back to your commercials or maybe changing your consumer relationship. That makes a lot of sense. And I think it's all about those low risk pilot programs internally. Are there new roles and new titles you're seeing emerge like chief AI officer or anything like that. Somebody whose job it is to come into these organizations and make sure that they're integrating. We've all been through multiple technology cycles and tech hives, especially I think the last few years it felt like it was accelerating. We had a very quickly, we had the voice, we had the web free, we had the metaverse and now we have AI. And each time we came with titles. I feel like the poster child was chief metaverse officer. I'm even now here like prompt engineer. Yep. So I'm a little, I'm always like a little skeptical of like names quickly put together, especially when they start by chief. I think like the C suite has like relatively stable jobs, which is we have a chief marketing officer. You're probably going to keep that for many years. Yeah. But it's true that the skills, the expertise and at some point in the job title my evolve. When it comes to where obviously anyone who knows how to build models, finding models, people who are good at machine learning pretty much, machine learning engineers, or even AI research scientists, those are like in very high demand right now. Yeah. The good news is that you don't need an AI researcher to work with AI, right? There are a lot of tools and APIs that make it easier for you to incorporate. And so very often you just will need even like very good developers or creative technologists, people whose jobs to connect the doubt between technology and creativity and marketing. And then more than trying to create almost like new job titles, I think it's more existing people should start to change their process and incorporate it into their work. So if you are a strategist, you should probably already start thinking how can I improve my ability to identify insights to articulate a strategy to create persona with AI. Yeah. If you're a creative director, obviously you can start thinking on how do I use AI as a creative partner to develop ideas, to develop mock ups. Again, if you're a designer and not director, it's a great prototyping tool, it's a great quick visualization tool. Yeah. If you're a producer, I believe you can start thinking how do I improve my actually my project management, with other ways to automate all the aspects of a project that actually converse some and premature value. So it's more how I'm thinking about it is like the existing jobs are evolving and it's more the skills and the tools are changing more than the titles you believe. And we briefly touched on the idea of liquid creative. Yeah. Do you want to tell us what that is, what it means? Yeah. So the theme of liquid and water is very big in my thinking and there are two angles. As we were talking about different specialties and like how to think about the future as a creative professional, the first angle is when people ask me, okay, so what does AI means for creative professionals? I like to say, I like to quote Bruce Lee actually and say, you're going to be water. Yeah. Incredible quote from Bruce Lee that is more relevant than ever. What does he mean is that you need in a world where the tools, the practices and the experts are going to change faster than ever. You need to adapt a mindset of constant learning, constant adaptation, constant flexibility. But you also need to learn not to be too stressed out about it, right? Because constant change can be very stressful. That's for me the idea of be water. How do you become more nimble and more adaptive and more flexible in how you use those tools, because change is going to accelerate and that's going to be a constant. When it comes to the other side, which is more the content and the creativity, I've been very interested in the idea of what I call liquid content. And so what I mean by liquid content is that I believe we're moving to water world where content, stories, experiences are going to become liquid, meaning they're going to adapt and be customized more and more to every single fan, to every single user. And you will adapt based on the context, on the interaction, on their needs, on their desire. So think about that a little bit like an evolution that multi-decates trend to one more and more customization. Yep, right. We started with broadcast TV and content was not very customized. It was a same for everybody. Cable brought a little bit more niche customization. Online video brought much more customized content like more like fringe content would be possible if you're a fan of very specific type of very specific hobby. You'll have content for that. And I think with liquid content is the evolution of what we're now. We're probably going to have billions of individual content experiences. Yeah. And what I mean by customization is also that idea that the content will for a single original content, you could imagine how it could change across time. So the length, a same video may be shorter and longer across space and screens. Maybe you watch a video on your phone, it's a vertical video. And the day after you have a headset and you step into the video now and you're surrounded by the scene interaction, right? Maybe you watch a video and then you suddenly want to talk to the characters. You're going to be able to do that fairly soon. I mean, want to play the film. This scene is incredible. What if I could play? It's like an RPG. Yeah. Our direction. I saw again recently Disney Plus is doing a lot of interesting experimentation around turning live games. They did that with an FL and with the NHL. They're turning live games into animated games where they replace the real live action by, for example, the Toy Story live action. Oh my god. So those are all examples of what I mean by liquid content. And it's of course the story is also. So it's my ambroring to imagine what it means for creative in a world where you you become more of a world builder, defining the rules of engagement, the characters, the place. But then the every single individual story potentially become like a unique interaction between a fan and the AP and the contents or and the brand. It's fascinating. I just had a conversation with Philip Mags who's the creative officer at Superside, which is a creative as a service company. Yeah. He was talking about the idea that brands are going to basically relinquish the idea of having a like static series of brand guidelines and a brand book and a style guidelines. And instead brands are going to be programmed to be able to self articulate. So you'll be able to program AIs that can articulate all of these different elements of a brand instead of having everything locked up into PDF. Yeah. You're going to program AIs to be able to articulate brand values, visual. around brands, stock images, whatever sort of branding assets. You can train the AI's to basically create self-articulate and create. - Yeah. - Which is, I feel like it's along the lines of the liquid content. - Absolutely. I think that's what Philip articulated is almost like the brand expression version of liquid content. I absolutely agree. And I like that example because I worked, for example, with Star Wars a few years ago, and working on the Mandalorian, we created an augmented reality experience for them. And so it was incredible to see how important it was to understand all the rules, the startling rules of the Starworld universe. They call that the canon. And a huge part of the creative process was to learn the canon. I watched pretty much every single movie many times, taking notes. - That's the good kind of research. - It was good research. But after watching the same film full-time, so start to feel like work. They have also playbooks. We had a lot of reviews with them. And for me, I was like, "Okay, so that canon that world building is going to remain, but the individual expression of that canon with the AI is going to be automated." And I think you could do a similar parallel with brand guidelines. Brand styles. Brand has an identity. Has guidelines. Has a voice. Has a nod direction. Has also a history, a legacy. And this is not going away at all. I think it's more going to be where a lot of the energy we're going to. It's like framework thinking a little bit. And almost like systemic thinking. But then the AI, indeed, that's a logic liquid content. It's probably going to be there to create the individual expression in interaction with the fan. It's actually interesting. It's almost like the fan becomes potentially a co-creator. The creative act, part of the creative act might happen at the end points. Whereas a fan, as a consumer, I engage with that brand. I engage with that AP. And maybe I actually ask for something specific. Or maybe I send an intention to my based on my interaction. Interesting. So we're not quite there yet where brands can self-articulate. But in terms of where we are now, what do you think are some of the essential either AI protocols or things that design teams and creative teams need to have in place for start developing? Yeah. It's a big question. There are a few pointers, I would say. One is start looking at your workflow. Yeah, that's usually what I recommend. Start what is your existing process? Whether it's a creative process, marketing process. And you can start looking at where are the gaps, all the inefficiencies, all the repetitive work, and how can I start bringing value there? I'm a big fan of trying to start from your existing process and find out how to improve it versus starting from the AI tool. I always like to say technology is not an idea to tool. So you have an existing process you probably know very well. You can start thinking about, OK, so how do you plug AI into this? Yeah, how do I, you know, again, where are the efficiencies? I'd recommend to start on the insights, strategy, production, because this is where it's also a much lower risk. The second thing is start, again, creating process for you to test, to pilot and to measure the results of how AI tools and platforms can help you process. This needs to be intentional. You need to organize it. You can't rely just on like your one curious team member who want to play with AI and expect him just to come back to say, hey, here's what I learned. You probably need to be a little bit more structured and disciplined. Third is you need to understand the challenge and the limitation. And I know there are a lot of questions. What does it mean in terms of liability? What does it mean in terms of copyright? Yeah. What are the risks? How can you plan for that risk? How can you mitigate that risk? How can you also be responsible in the way you use AI? Right? Whenever we use a technology, we have a responsibility to proactively mitigate potential negative consequences. And finally, stay and remain water, as I said earlier. Change your mindset. You may be worked on certain processes for many years. So how do you drive change at a much faster space? And how do you change your organization culture in general to be able to adopt that change much faster? I feel like that's a huge insight, the fact that don't start with the tool, start with your process and adapt your process to use the tool. Instead of, oh my god, we have to start using AI. Where do we start? It's okay. What are my day-to-day tasks? Yeah. How can AI alleviate some of the burdens and elements of those tasks? It's a good starting point. It's also in my career. I've been working in re-emerging technologies for 20 years now. And in my career, what I've seen is that change is difficult. And sometimes it's tempting to just swipe every few minutes. Start from scratch. Right. And sometimes it makes sense actually to forget the start new foundations. In the meantime, very often that faces where too much opposition and friction and they make it unproductive. So I always push for crawl, walk, run, approach in general. Right. It's a classic one, but it's very powerful. So start by crawling, start by playing, start by carving a space where you can try things. That's also what I mean by start with your process. Start with what you know. And at first maybe innovate with incremental thinking. Like what can I do? And then if you see the right results, then you can start to accelerate. And ultimately maybe you'll completely change your process. But if you want to start to go straight for the multiverse pretty much, it's very likely that you'll stop in the middle because this is too much for you. So start by climbing a few hills, get quick victories, surely positive impact, drive change and transformation in your organization. And then you can start to get a bit more ambitious. Once you're used to climb small mountains, it's much easier to go for the big one. And I think that's what team leaders need to keep in mind. Because don't be so ambitious, don't try to scale Everest. Bring your team to base camp. So they can see whatever it looks like. Just have small wins because what that does is it enables them to understand how the AI technology works. And then they'll understand how it works. And they'll apply that thinking to, oh my god, I can use this for this other workflow that we haven't even been talking about. And I think it's more facilitating people in your team to understand the framework of AI. More so than how do I use this tool. It's more like a sort of paradigm shift in terms of how they think about work if you think about it. Absolutely. And you don't go straight for a marathon. First you start my running 5Ks. Yeah. In the meantime though, it's not enough just to run one 5K and say that's it. I'm ready. Yeah. You need to start running every day pretty much. And so I think that's the same with AI. Before completely dramatically changing how your organization works, you want to start with small wins. In the meantime, it needs to become a muscle. It needs to become a habit. It needs to be part of your practice pretty much. If you really want to understand what it will take to get the peak, the big multi. So what do you think? Obviously there are the kind of headliner tools that everybody knows about like mid-journey and chat GPT and Gen and I. What do you think, what do you consider to be the essential tools right now? Not very surprisingly, larger grid models and such as Gemini and GPT. Yeah. I think the number one tool for credit is now and there's just so much you can do in terms of improving your brief, finding insights with its articulating strategy, helping expand on ideas. So that for me is the number one tool and it's going to evolve, I believe, to what creative agents, creative partners that become your sounding box. So I would really encourage everybody to invest time into becoming better at using those tools. If you also work in innovation or in design, obviously tools such as Midjohn is also incredible. Not always necessarily for production but for prototyping and visualization. Yeah. In my work, part of my job is to try to invent the future and so to come with ideas and try to convince people to support ideas for things that don't exist. Yeah. And imagination has the need. Sometimes you really should and this really will look like. Gen and I is incredible for that. Be able to very quickly have very high quality, high resolution representation in marks of your idea helps a lot. Right. So that's not surprisingly tools most people use. Yeah. But it's more like how you use them. I think that is critical. Right. What about automation? I feel like a lot of business leaders, obviously they want integrated into the day-to-day of their marketing teams, creative teams. Every company's got a bunch of work they hate doing. Yeah. And probably a first pass that a lot of people want to do in their company is what can we automate? So what are your preferred tools and overall workflows for automation that you recommend? For automation, I've been very interested in companies trying to chain models. There's a, for example, a company called Lang Chain. Okay. That provide a pretty much a framework that lets you attach multiple models together to start creating very custom workflows. All right. So it could be, for example, and you can combine that with non AI. So you can, let's say you want to automate, for example, social media work. Those frameworks lets you, let's say, for example, scroll Twitter known as X now, identify on a specific hashtags, X amount of information and then send that to a large language models that my positive information and do potentially extract maybe or summarize the key information. Yeah. And then you can chain that with another model that maybe might do sentiment analysis. And then you can maybe chain that to another model that might use that to then create quick social media ads that react to that sentiment. Yeah. So So, chaining models, I think, is a very interesting way to start automating and create your own very custom tools. In the space of image, I've been very excited by the open source scene and especially right now there's a tool, there's an interface called KofiUI. Oh, yeah, KofiUI. And of course. And of course, it's really what I think. Right now, I think the most advanced AI designers, that's what they're using, they're using KofiUI to create absolutely unique and automated workflows for image. So, let's say you, for example, want to create an asset with a lot of different variation in your background or a lot of variation of point of view, KofiUI can help you for that. So, it's also very interesting. So, Langchain, when you want to create custom workflows for information in general, KofiUI and open source tool, when you want to create more unique automation for visuals. They provide pretty much libraries and different solutions. Oh, excellent. Okay, great. What about life hacks? Not necessarily business. I always like to ask people, what are some of your favorite AI life hacks? So, what are yours? I have a new one, I really, it's actually using large language models as a deck critic. Interesting. I've started that a few weeks ago and it works really well. A lot of us, I'm sure, are doing pitch decks, presentation decks. It's the format to convince anyone of anything those days. Yeah. So, what I started to do is to download my decks as PDF and then upload them into a large language models, for example, GPT and asked the model, okay, he's a deck, he's the context what I'm trying to achieve, my audience, my goal. And I ask him to give me a grade from 0 to 10 and to give me a recommendation on what I can improve. And that works extremely well. So, for example, you have a sales pitch, you will be able to tell you maybe a 7 out of 10, which obviously can be improved as B plus. And then it will give you very actionable information on the narrative flow, the content in the slide. And other names are imperfect. Sometimes you might get a little confused on what slide is what slide. But that's fine, you can refine. So, that's by far my new favorite hack is using chat GPT, Gemini and other tools as a partner that will provide pretty, pretty strong feedback on your deck and how you can improve them. That's great. I feel like people don't use these LLMs as brainstorming partners and creative partners the way that they could. Because it's basically like having three other very agreeable people in a room with you helping you brainstorm your ideas. It's very true, when it's funny, you mention like three other people. You can actually do that with an LLM, you can obviously it's a classic method is to give you the role. But you can actually give it a skeptic or be there. You can give it a lot more. There's an ideation method called the Six Hats, that is very famous. You pretty much can do that with an LLM. Say, "Hey, here's an idea I want to brainstorm with you and I want you to play three different characters. Character number one, I want you to be the agreeable." Really, character number two, I want you to be the devil's advocate. Character number three, I want you to be the underdog. Yeah, and those are like two point very powerful ways to ideate better. It's not going to replace you. Right. There's going to bring perspectives that helps you ideate better. Very smart. I feel like this is a perfect place to end. Matt really appreciate it. Thank you. Thanks for having me.