[music] [Jay Ganaden] How's everyone doing? Welcome everyone. Thank you for braving the weather, braving Miami, and joining us here. Thank you so much. I'm Jay Ganaden. I'm Director of Global Agency Partnerships. It's a new team at Adobe. And I am focused on getting customers to use Firefly as much as possible. I was part of the original team on Firefly when it was just myself, an engineer, a designer, and a marketer. And we've met so many people, so many customers. And over the last year or so, we really focused on how to operationalize it. And so today, I'm really excited to have this panel. Let me get this guy. I'm excited to have a panel here today that's going to walk us through some of the use cases and the practical considerations of getting Firefly up and running in your organization. Thank you. This is a very full house. If there are people that are standing, there are some more seats over this way if you like.
And so, there you have it. Let's go ahead and start to bring some folks on that are joining me today. I have a fabulous panel.
And so, I want to bring up Jocelyn Chang. She's the creative director for Amazon Devices. Let's give it up.
So, Jocelyn has a similar kind of career path as me. I started out in management consulting as did Jocelyn. We also had stints at Google, not together, but in different versions of it. And she's also worked with the very talented Jennifer Morla if you are graphic designers in the room here. Also joining us, I've got Mike Pell.
Come on up.
Mike is from Microsoft. He's the envisioneer behind the Microsoft Garage in New York City. And he's also known, and this is a little-known fact, He is the co-inventor with John Warnock, one of the founders of Adobe of PDF.
Yeah, see? Nice.
And he's in a-- And then, lastly, but not least, I've got Kelsey De Weerd coming from Paramount Plus, where she is the director of design, formerly at Disney, among other places. So, with that, this is who's going to talk about Firefly today. So, thank you. And we're here because we're going to talk new creative flows. The Gen AI jumpstart.
So, what this means is, you know, what are the use cases that customers are implementing? So, we've got three very different use cases that we're going to share with you today. But first, anyone here know what PechaKucha is? Alright, one.
Nice. Two. Thank you. So, PechaKucha is something I've participated in before. You get 20 images and 20 seconds. And then the slides will automatically advance. Right? So, that's the hot seat portion of what we were talking about earlier in the description. So, joining us now, we've got Jocelyn. She's going to hop on stage here and talk to us about what she's doing with Amazon.
[Jocelyn Chang] Thank you, Jay.
This is my first time doing a PechaKucha. Wow, they advance fast. Hi, everyone. I'm Jocelyn Chang. I'm a creative director on the Devices Brand Studio team at Amazon. I'm super excited to be at MAX today to tell you all a little bit about our creative studio and how we're approaching using generative AI.
So, first, a little bit about who we are. Many of you have probably maybe used Amazon at some point in time to order something. Well, there's a lot of different teams besides Amazon.com. And I'm on the team where our org creates Kindle, Fire TV, Ring, and Echo products that use Alexa. So, on the Brand Studio team, we focus on creative storytelling. So, inspiring customers to choose and use our products. We craft visual motion assets and written copy to tell stories using a mix of live-action, CGI, as well as motion graphics. Our work spans the digital and physical worlds anywhere from Amazon.com to social, to brick-and-mortar retail stores like BestBuy.com to events like our upcoming fall launch. And this is just a selection of some detail pages that our team created. To better scale our work and make sure our brand is consistent to our customers, our team also focuses on creating those brand guidelines that we all know and love, as well as photography style guides. And these we give to the devices marketing teams so when they create assets with external partners and other internal partners, everything's on brand. So, now that you know about what we do, there are two areas that we've been using Gen AI. One is in creative development such as ideation, creating concept decks. And the second is production. So, resizing as well as retouching. For production in the past, when we've had a photo, like what you see on the left, and have to make it really large and wide, like what you see on the right, we would have to just roll up our sleeves and photoshop away for hours and hours to make those new pixels. I'm sure some of you have felt that pain. But now we have a much faster solution. Generative expand, as many of you have probably already been using. It's a super powerful tool that helps us quickly expand the small scene into a big one without doing all that intense manual work.
However, it's pretty important to note that generative expand isn't a one-button solution. It's not one and done. For those of you who have tried it, you know it's better to work in really small increments. And at the end, you may need to do a little bit of light retouching, just so the final result is picture perfect, as we all aim to do. The great news is, even with that light retouching, a job that used to take eight hours, a full eight-hour day to make these wide assets, as well as some really large tall ones, now can take less than an hour. So, we've gotten similar time savings across a lot of these production use cases. Now, I want to switch gears a little bit and talk about some of the things we did at the start of our AI journey, way before we even had Firefly. One thing was to have our entire team, the design team up to leadership, really align around the principle that the creative quality bar for visual expression does not change with AI. So, even though we want to move quickly, be efficient, our same quality standards apply for AI, and we still look for the same things, such as product accuracy, photo realism. Does the context or scene, is that appropriate for our product? We do all the legal checks.
Now, looking at product accuracy a little deeper, the AI image absolutely needs to depict our product accurately. So, we check for things like the shape, the color, the texture, the size, and it's absolutely critical in marketing our products that we are faithfully representing them. Another thing we look at is just photo realism. This is a unique issue when we're using AI, when images can come up with these hallucinations or distortions, where on first glance it looks fine. And then when you look closer, you'll see the things that were circled in red were little artifacts. The next part of the journey was to figure out the best way to help our team get excited about AI and how it applies to our work. We provided regular updates that we called lunch, learn, and discuss. And the focus was really discussion. These were highly interactive, where we did a lot of listening. And through these, we were able to identify early adopters and AI enthusiasts, some of whom became core members of our AI tiger team. And then through these sessions, it became clear that it was really important to have a sense of play drive the AI integration, to give our creatives a fun, low-risk way to try AI. So we kicked off an AI challenge to create an image that is a mass ad for an animal-related product. Whether it's a product service, organization, or cause, the animal could be real or imagined. And the contest was for our entire studio, not just visual creatives, but also strategists and project managers. Our team not only had fun making kittens with lightsabers, but they got prompt writing practice, and this generated a lot of organic discussions, organic informal discussions about the potential and also the limits of AI.
So ultimately, it was not kittens or dogs, or even a real animal that won. It was an ad for Sasquatch Salon, where Bigfoot meets big hair. That got the most votes and the bragging rights, and of course, a gift certificate to Amazon.com.
And I'll just leave you with some of these wonderful other entries we got. On the top left, Zoo Optical, because even Ziggy deserves 2020. A Furberry collection right below that, and a Black Labbath tour poster for any of you old enough to know that band. Thanks so much, and now hand it to Mike. Or Jay.
I just want to give it up for Jocelyn. She says it's her first time. Isn't she a natural? Come on, guys. Thank you for being the first to do the PechaKucha. And I'm going to hand it off to Mike, who's a seasoned pro. You can see he's really into country and things.
So, come on up, Mike.
Cheers. [Mike Pell] All right. Hey, everyone. My name is Mike Pell. I lead the Microsoft Garage in New York City, which is the company's worldwide innovation program. And it is true that I did help to invent Acrobat and PDF with some friends of mine way back when. I was employee number 491, so thank you for having me today. What I want to show you is what I've been doing. I do a lot of work with data viz. And so, for me, the most interesting thing to do with generative AI is to help decorate some of the things that we do within the data viz space. All of you know that we have been dealing with this problem forever. Right? Does anybody disagree with me that charts are just terrible? I mean, come on. Really? Really? I'm pretty sure we can do better than this.
And so, I think all of us as designers, marketers and people who care about communicating could do a much better job. The way that I like to do this is by sort of connecting the dots between these different apps. Now, it may look like a really long workflow when you look at it this way, but honestly, it's not, it's super fast. I found a great technique I want to show you very, very quickly. So, I'm going to connect the dots in like however many seconds I have left. So, I'm kind of digging being in Miami. I've been here a couple times. You know, I use Microsoft Copilot quite a bit to help accelerate what we do with innovation at Microsoft. There are certain parts of everybody's process where you can sort of speed things up. I want to find out. If I really wanted to move here, right, I'm sort of in the market for a luxury condo. I said, Hey, Copilot, can you show me the price of luxury condos the last five years and then project forward for me for the next five years? Actually did a pretty good job. It spit this right out. I was like, okay, impressive. Yeah, I think you got this. And then, of course, what do you do next, always? You bring your data into Excel because you got to clean it, you got to arrange it. You have to make the most beautiful chart you've ever seen. Right? Okay. This is what I'm talking about. Really? No. Okay. This is what we do next. So, we have our data. We got it into Excel. Now, we can copy it. And I bring this into Project Neo. So, remember what Michael did this morning? He was amazing, wasn't he? Project Neo. I use Neo for something very, very simple. I compose the scene in a way that I want it, because it drives me crazy when I go into a Generative AI application of any kind, I'm generating images and they're nothing like what I want. So, I found a very simple technique to create something and use it as the layout to be able to compose what I want. So, it's a very, very simple chart. Take this into Firefly and lock it and you see in the upper left corner, I was able to use this as reference. So, my image is now going to be the composition reference. And it does a great job, right? I just said, hey, you know, generate something that sort of looks like it belongs in Miami. I get a great chart. I kind of like the one on the lower right. You know, for me, had some elements of photo realism. There's like people walking around. Looks much more interesting. From here, it's simple to just be able to export this. Take it into, of course, everybody's favorite PowerPoint program. I'm just going to say, for all of you who are, you know, non-believers, PowerPoint is the most powerful prototyping tool on the planet. I challenge any of you, right? In a PowerPoint, throw down. I can do anything in PowerPoint. But what I really wanted to do is just put text on top of my image. So, it's really good at this. And it even like did this stupid angle thing, right? So, it looks like it's in perspective. You know, you've got to make it look a little good. So, now I've got the basic information. I have my background. I have my text. What do I do next? Just what we saw this morning, right? Take it into Adobe Express because now I can add some motion. And so now I can easily export this into just a little bit of a video because you know that the kids like videos these days, right? And if I want them to pay attention, I've got to do it this way. This was so simple, right? So, in just a few steps, I went from having my data to being able to compose it to getting it into a form that everybody can understand. So, not bad, right? For just a few steps. It does not take very long at all. And so this is how you can use Generative AI to be able to not just decorate, but really to bring this stuff to life. So, see? We did it. And it was pretty quick. So, thank you.
All right, give it up for Mike. Thank you, Mike, for taking us through data viz and suggesting that PowerPoint is actually a prototyping tool. It is.
No, I've been put in that position before. So, I get it. And so, joining us next, we've got Kelsey De Weerd. Joining us from Paramount+. Come on up. Gonna take you through a really fun project.
[Kelsey De Weerd] Hi, I'm Kelsey. I'm from the design team at Paramount+. I'm going to walk you through a really fun social activation we did for the movie If. If you haven't seen the movie, it stars John Krasinski, Ryan Reynolds, and Cailey Fleming as Bea. And Bea discovers she can see everyone's imaginary friends. So, Bea embarks on a journey to reunite these long-lost Ifs with their kids before they disappear. As part of the journey, she draws in her sketchbook. For the launch on Paramount+, Our social team came to my team and said, Hey, can we have Bea draw our followers' Ifs in real time? We thought, that sounds cool. Let's talk logistics. But, spoiler alert, we made it happen. And on the day we posted this prompt asking followers to describe their imaginary friends. And in return, we'd reply back with a drawing as if we were Bea. We created over 70 Ifs, like this furry little red one here, and the snowman for our fans, for our brands, like the chocolate bar from Forrest Gump. And more here, one of my favorites, the fan girl from Awesomeness TV and the walking mustache from Anchorman.
The fans really caught on. Here's an example from Criminal Minds, where the fans went wild for their If, really just expanding our reach to their audience as well. So, how did we do it? Teamwork, imagination, and while, why we're all here, Adobe Firefly. We had been using Firefly for background extensions, for some retouching, and testing it out on another project where we wanted a variety of backgrounds in a similar style. And gave us the idea, could we try this for If? So, we jumped in. The designers were excited. They found inspiration from the movie from Bea's drawings.
And played around with the settings, we created our own hand-drawn style reference for the character.
And once we were all aligned, we briefed the design team so everyone had a similar style set. On the day of, we made sure to humanize the prompts to really get the essence of what the user was asking for. And we would pop the drawings into Photoshop as needed to make some tweaks. So overall, we had six designers, two social media managers, and in two hours we created 70 drawings, which was under two minutes per reply.
Here you can see most of the collection of the Ifs we drew.
We had a lot of fun. The fan response was great. The team said it was the best project they had ever worked on and are eager to do it again. A lot of imagination all around here. So to that, we say congrats team, congrats Firefly. Let's see what we can do next. Thanks.
Thank you, Kelsey. And thank you panelists for joining us in this 20x20 PechaKucha. What'd you guys think? We're 19 minutes in. And that's all the slideware. Well, I've got a few to show. But that's all the slideware that we're really going to walk you through. Now we're going to kind of get into the practical of how we made all these campaigns happen. So let's get into those details. So I'm going to have a seat and we're going to have a chat. Okay, so the work, right? So let's start with that. What are the use cases? Clearly, we've seen production. We've seen data viz. We've seen elements of social and how we've integrated Firefly with that. So let's maybe dive into this project that you had with production, Jocelyn. And tell us a little bit more about this. I've got this kind of video rolling for you. But tell us about how you discovered Firefly for this and how you've rolled it out across your team and those kind of things. Yeah. So production design is probably one of the most obvious use cases or the low-hanging fruit. And it was where we wanted to start with Firefly.
And as soon as we onboarded Firefly, we didn't want to just say, Hey, everyone, go at it. We wanted to be a little more prepared and thoughtful about it. So as I mentioned in my presentation, we had established some principles and guidelines about image quality as kind of a first step. And then secondly, we found one of our most experienced Photoshop retouchers. And had her go back and look at some business use cases that we already did where generative expand and generative fill would have been useful. So then she went back and redid some of those using these new features. And from that, we had documentation for how much time savings we would get. And also, more importantly, she was able to really drop down tips and tricks. And she was very thorough about this and did even screen captures and video recordings. This is one of them. And that actually became a Firefly guide that we gave to our production design team. That's great. So you were able to document the process and through the lens of someone that does it every day. That does the work, every day. And the exact type of work that other production designers do. So tell us, was she happy to use AI in this case instead of... She was absolutely thrilled. Like, she was slapping me as she was using some of these features and saying, Oh my gosh, this took me eight hours before and now it's one hour. But oh, look, I do have to fix some of those water stains because it made it look weird. And this is how I'm doing it. And so there is this mix of excitement and also understanding some of the limitations that we still do need to do some manual retouching. You still need to have that eye for detail. And one of the things that also came out of that is that we are adjusting our QA process. So we do have a QA process and team for production design in general. But because not everyone is a professional retoucher, per se. We want to make sure in this QA process we're also able to capture maybe some of some of those little weird AI artifacts that come in. Or maybe in this example with the pool if that grid doesn't align perfectly. And so part of that is saying now we want to see these images in full res or the QA team needs to see them in full res as opposed to small and in a PDF. Thank you for the PDF, but sometimes it is not the best format to view. I always say like half the people that I meet who know about this they say, Wow, thank you. The other half say I really want to stab you. I can't stand working with PDF. We love the PDF and typically it works just fine. But in this kind of transition period we want to be able to catch any little things that the team might not be catching. The photo realism principle that you had. Yeah, this is the photo realism. Okay, that's great. And so you're able to document this and distribute it across your team. Now did it stop with Amazon devices or have you been able to kind of... You know, I think a lot of large enterprises are like this where each team probably has their own way of going about rolling things out. But we are definitely consolidating a lot of the learnings and sharing them with the rest of Amazon. And so we did have a presentation. Well we had a Firefly kickoff where I presented and that was Amazon-wide. And we do have a few internal resources that are for all the designers and creatives at Amazon that we're sharing some of this material. So I'm hoping it will influence the rest of the design community at Amazon. So they don't have to retread through some of the things your team has learned. Exactly. That's great. And then the other thing that I loved about your presentation was, I mean, kittens, come on. And lightsabers, I mean, why not? And so, you know, I gave this presentation about a year ago On navigating the excitement and anxiety of generative AI. And so, and you've got these sort of polar opposites of like, oh man, I'm so excited. I can do all these things now. And then the other part's like, well actually, that's kind of what I've taken my whole career to learn. And so one of the things that I liked about your presentation was how you were able to sort of navigate that through the lens of this challenge. So tell us a little bit about the challenge that you built out and got that as a means of getting your team more comfortable perhaps with generative AI. Yeah, so I think navigating excitement and anxiety of Gen AI so encapsulates how people are feeling, people are excited. But especially our team, because we didn't have access to Firefly for a little while, there was also a lot of anxiety. What does that actually mean when we roll it out? And just being able to see what's happening in the news every day, you hear AI this, AI that, it can make videos, it can make this. It did create a little bit of anxiety. And so, the idea with the challenge was to make it as easy and low lift and fun as possible. So we actually rolled it out on a Slack channel that was pre-existing. It's an informal channel. That's a lot of fun. It's called One Cool Thing. It's just for our devices brand studio. And in the channel every day you could submit the coolest thing you saw online or maybe something you made. And so that kind of informal tone was already part of this channel. And so we took the month of May and made it the AI challenge for that whole month. And I think that really helped set that tone to do that in the context of this channel. Also, we wanted to make sure everyone could participate. So it wasn't just a designer challenge. We wanted our copywriters, our project managers, our strategists.
And pretty much anyone who wanted to try AI and get hands-on experience to go ahead and do that. And so we were really explicit that we actually don't want copy on top of the image, even though we can all do that in PowerPoint or some other way. We said, don't bother. No copy on top of image. Just generate the image. Focus on the core concept and add your copy in the Slack. And we got a ton of participation, close to about half of our studio submitted entries and almost the entire studio voted. Okay, great. And so what kind of volume were you talking, image-wise? Because you have a lot to sift through in that content. Well, it was all kind of automated, the way that we did the tally. So it was not that manual. But that reminds me, one other thing was, I think for me, the best or the most useful part was seeing all the organic conversations about AI. Because sometimes people don't want to talk about their excitement or their fears in a more formal setting, and to see people bring that up in a Slack channel was really informative for me as someone who is helping roll out AI for our studio. And Slack is just one of those things where it's meant to be a bit informal, right? Versus doing it all in Jira tickets.
But okay, that's great. I love the challenge. And then the challenge was also something that was a component of what you worked on, Kelsey, except this was a challenge that was done on social media and out to viewers or other artists that were interested in film. So maybe tell us a little bit about your challenge here. Did social come to you? Did you go to social, like how did this all work? Yeah, the social team came to us and the challenge really was how do we get this done in real time? Do we hire an illustrator? Do we have to hire three illustrators? Because they wanted this like immediate, like people post, "my If is..." And then we give them back this drawing right away. And then we're creating characters. Is there an IP issue there? So when we thought of Firefly, we got really excited because it meant we could use our design team and use all of them. It wasn't just one person. We could use a group of people and keep a really consistent style. So it made it really fun for everyone. I think another thing that we were able to do was like control the prompts a little bit. I think you see the owl right here holding the heart. And the user said, this is my deeply departed dad in fluffy owl form. And we were all in a Slack channel together discussing the prompts. We're like, how do we do this justice? Do we skip this one? Let's try it. So a couple of us were trying it and we had to actually go in and say, can the owl hold a heart and put a thumbs up? And so we kind of made it into what it is. And if you can see the response back there, they loved it. And it brought everybody smiles honestly. And tears maybe.
So very interesting then. Maybe tell us a little bit about how you got to this art style with Firefly. Yeah, we really took inspiration from the movie and the girl's sketchbook, and she's drawing with markers. We looked at a couple other style references. And then we drew our own so that they all had this marker hand-done style. And then at the end of every prompt we had a consistent set of words that said like hand-drawn style, kids. So the beginning of the prompt was all the differentiator from the users and the end of the prompt was all the same throughout our six designers. Great. And so now that you've kind of got this sort of challenge contest through, what are some of the other use cases that you're seeing at Paramount? Oh, you'll have to stay tuned to find out.
Okay, fair enough, fair enough. Alright, so shifting gears here a little bit. So Mike is an accomplished author, in addition to inventing Acrobat. Right, and so he's written a book here. This is not a paid placement. It's me doing my research. So Visualizing Business, now it makes sense. Data viz for your presentation. But in this, you talk about this new approach called 4D. Can you maybe walk us through what that means and how that might be relevant to the audience? Sure, so for many, many years I've done a lot of work in data viz. I've always been kind of a 3D-person from way back. And I've always loved to help people communicate clearly as much as I can. And so I've written, actually this is the third book I wrote, about this technique of taking AI and data in a more cinematic or spatial approach to how to present information. And so the notion of 4D, four dimensions, is really just adding time to the things that we try to communicate. Now, many of you are video editors. Time is part of what you do every single day. But the vast majority of business people never, ever deal with time as a concept. It's just not something that they do ever. And so to me it's ridiculous that we don't have time as an element of what we do in data visualization. Because there's always some data that's come before. And many of us would like to play "what if" and sort of project forward. So in this book, I just talk about, I invented a time machine to help people to be able to rewind and look at existing data and then sort of come forward into the present and then fast forward to project forward. So that's kind of the notion of 4D time being added to data, which of course many of you are already very familiar with in your work. But most people are not. So kind of taking it maybe to this next level. You mentioned Project Neo. We didn't realize it was actually going to be on stage today. So delightfully brought by Michael on stage. I love that he was able to just-- Michael, are you here? But I love his style and just kind of laughing his way through it, right? To those of you that saw it. But this is project Neo. It is part of Adobe Labs. And so tell us about how you discovered this, how you're relating it to this concept of 4D and time.
So yeah, take us through that. Well, so I was trying to figure out how to shorten my process. So I know 3D modeling and animation. I've used all those programs that many of you know. That 3D software is some of the most complicated next next to music software, or frankly, video editing, video editing is super hard as far as interface goes. Project Neo is really simple. I just needed a really lightweight mauler and I saw online. There was an Adobe evangelist who published his little video of taking Project Neo out exactly what I did out putting a frame to use as basically locking the composition of what they wanted and taking it into Firefly. And I thought, wow, I can use that. And this is going to be really easy because I would spend forever. Sorry, like when you ask Firefly to do something or anything, you know, for that matter. You don't get the same thing twice in a row and you can't for the most part make it compose things exactly the way you want until I discovered that technique. And now it's easy because I can lock everything. Yes, I can decorate it in lots of different ways. But the composition is exactly what I wanted. So that technique was super simple. If you haven't used Project Neo, I would encourage all of you to just try it out because I found it, just like Michael, to be very approachable even for an expert. Yeah, so what you're talking about is, for those of you that are kind of newish to Firefly, we use something called structure reference. It allows you to lock in the layout of a particular image, and you upload that image. It's kind of a way that we initially had stepped into custom models which do give you that consistency that you're looking for. But yeah, very interesting. What are some of the use cases, the additional use cases here? I mean, you're a data viz guy? Is that kind of the scope of envisioneering? Yeah, so in envision side, I'm an envisioneer which means I can both envision the future and know enough about technology to be able to engineer rapid prototypes. So I actually use this technique now to show people how they can put very quick things together in Neo and then they can go off and do the kind of communication they need to do to do a presentation to management. Or to go back and run a campaign or whatever they have to do. So for me, it's been super handy to have something like Neo That's a much more lightweight mauler that's web-based. So I don't have to go on to a super heavyweight program just to get something simple. And for data visualization, you generally would not have an image like this castle. You know, you use primitives. You use very simple shapes. And so it's absolutely perfect because Firefly gives you the materiality and the cinematics and the lighting and the little things that make something really be able to make an emotional connection with people. Like in my example of the Miami luxury condos, it didn't matter what that was. It was visually interesting. And these days, that's a really important aspect of what we do.
So let's talk about overcoming implementation challenges.
There's the excitement element, right? The anxiety that we're trying to push through. And some of that anxiety sometimes manifests in the organization. Right? Like, how are you sort of overcoming, you know, there's issues of legal and ethical concerns. What are some of the things that you've seen in each of your organizations? And maybe we'll start with you, Kelsey. Sure. Yeah, I think for Firefly specifically, we know that it's all vetted, that it's legally approved. So we don't have to send all of those images through legal, just like every stock image wouldn't have to go through legal. So it just lets our work occur more efficiently.
Yeah, I mean, I think a lot of the known implementation challenges are around legal and security. And I mean, for us, that does feel like it's slowing things down, but that's not necessarily a bad thing. I think we need to be really thoughtful about how we use generative AI. And it was good to have gone through that process because now it's a little faster and easier once those things are approved through legal and security. I think the next challenge is really going to be around how to integrate multiple AI tools and models and make them into a large cohesive, efficient workflow. So just like today, we don't just use Adobe tools. We also use Figma. We may use Microsoft tools. We use them in a way where it's fast. It fits together and works cohesively. So that's going to be our next challenge. How do we redesign our workflows with all these new AI-enhanced tools and maybe these newcomers that are coming along with AI tools? As well as for a larger corporation, I think probably a lot of large enterprises have internal tools as well. So it's an integration challenge that we, of course, haven't solved yet. Sure, sure. And then how about you, Mike? What are some of the things you're seeing at Microsoft? So in my day job, we get some of the world's most successful companies coming into the garage to talk about innovation. And the first question they ask is, What are we supposed to be doing with this AI stuff? The second question is, can we lay off half of our staff? The answer is always no. You can't, right? Because we need people to do this. So part of the implementation challenge is that there's a lot of anxiety still about people not really understanding what you can and can't do with this. So part of my job is to try to demystify this and just take them through very simple examples and show them. Sometimes this works great. Sometimes you don't know what's going to come out the other end. This is the very first time ever in the history of technology that we've made something that you cannot predict exactly what's going to happen when you do something. And this is still probably difficult for all of us. You really don't know what's going to come out the other end. It's a process, right? It's a lot of experimentation. People are not used to that. So as great as it is, there's an adaptation that you have to do internally in your culture to be willing to experiment. Absolutely. And this is part of the workflow thing that you've mentioned, Jocelyn, right? It sounds like we still have human curation as an element that is part of this workflow. It seems consistent across all of your teams. So let's shift a bit and talk about the future of work and workforce. And by this, I mean, how are we adapting our creative teams? What are the practical considerations there with creative teams? And again, I was talking through this idea of the excitement and anxiety of generative AI. And on the one hand, there's this sort of narrative that creativity will be democratized. Another where there's a feeling of, you know, hey, it took years to learn my craft and find my voice, right? And so in that sort of milieu, where are we with skills, like new skills for your teams, to bring them up? And are those skills actually vastly different than they were two to three years ago? And what skills are undeniably human? So maybe we can start with you, Mike. So I think for all of us in the room, you cannot be the person you were five years ago. You can't be the person you were five months ago, right? This is going too fast, right? We can't go backwards from here, right? Things are going to continue to get faster and faster. So it's kind of up to all of us to want to learn, right? And to be willing to try and experiment. If you're unwilling to just try and see what this is all about, it's not going to turn out well, right? We're seeing how quickly organizations are adapting to this kind of stuff. And it's a lot of fun. I mean, I think that most people who try this out think it's a great experience. But there's some people who are just resistant, right? It's human nature. You know, we don't want to have to learn anything unless we have to. Well, you have to this time. You really have to pay attention. And so the skill set is different than the hard skills that many of us had to learn coming up. The new skill set that you need is to be flexible and to be a great problem solver and critical thinker and be willing to experiment with your creativity. So, you know, all of this is a superpower to us. It's not an impediment. It's not something that is going to take our jobs away. It literally is going to make you better at what you do, but you have to try. Yeah, when I worked at Razorfish, the UX VP that hired me said, you know, Jay, you've got to be able to design around uncertainty, right? And I think that's paramount, even paramount today.
Kelsey, why don't you tell us about some of the skills on your team? Yeah, I think curiosity is vital. You have to want to learn and keep growing. But I also think you have to be a decision maker because there's so many opportunities. Like you could sit there and generate new images all day, but we need your vision and we need you to make decisions so that we can choose the image and move on and, you know, get to the end goal. That's awesome.
I think, you know, it's easy to identify some of those hard skills, just like if you're a visual designer, it's good to become more comfortable with writing and specifically prompt writing, understanding the syntax, the order of writing, how to give art direction to a machine versus to a human. In terms of soft skills, I would echo what kind of has already been said.
I think embracing a mindset of learn and be curious and really being comfortable iterating and experimenting and exploring and understanding that things are not going to be one and done. You're not going to get it right on the first shot and that's okay. That's good. And I mean, that's kind of the rough and sort of unscientific process of design, right? I mean, I think there is a lot of that experimentation and while we have new tools to do prompt writing, for example, these are things that you need to hone and include as part of your workflows.
That's great. So now you are all design leaders, right? So let's talk about creative leadership. How has this changed your sort of direction of people on your team? And yeah, let's talk through that. Kelsey, do you want to maybe start us off? Sure. I think it's a lot of what we just talked about. It's encouraging the team to be curious and giving them the room to explore. You know, that means time sometimes, so arranging our workflows that we can give that freedom. And then in the end, we have to make some decisions based on what we're allowed to do. But I think before that brainstorming is kind of like, nothing is a bad, and a bad solve until we have to get to the timeline.
Sure. Mike? I would think that we all have an opportunity now to encourage people to figure out new kinds of workflows. Like something that you never thought that maybe you could do before. So, for example, in my world, we always had a model, something in 3D, and then render it and then figure out how to do it. And now we can just compose an image. Well, maybe we could figure out how to go from a composed image back to a model at some point. There's like these crazy new things that are available, but if you stay locked in to the way you've always done things or locked into the way that you always approach the problem, you'll never be able to leap forward in some new way. So I try to encourage people to just come at it from a completely different angle and just see what's possible.
Yeah, I'd say really embracing a continuous learning attitude and helping those on your team who that may not be natural to them for them to be comfortable with that. I think another piece is that in the past, we may not as creative leaders been thinking as much about new skills and development and how important that is. But now with AI and not that many people having five years of AI experience because it did not exist, thinking through what we need to do for our team to help support them develop those skills to help them work better with AI and be comfortable with AI and trying new things in new workflows. It is something that's fairly different from attitude and cultural standpoint.
Cool, well, we are at time. We're over time. But thank you, everyone. Let's give it up for our panelists. Jocelyn Chang.
We'll give it up for Mike Pell.
And Kelsey De Weerd. Thank you so much. Thank you for spending the afternoon with us. And go hit some happy hour.
Cheers, everyone.
