By Mallory Gafas
As Generative AI disrupts the way businesses operate, create and engage, unprecedented opportunities emerge for brands to innovate content, enhance productivity and grow value.
At the intersection of creativity and marketing, how can enterprises embrace GenAI to maximize benefits while minimizing risk? Where can organizations leverage new technologies to reimagine the customer experience? What are the implications of GenAI for creative workflows and content development?
To address these and other pressing questions, marketing leaders and top creators gathered in Cannes, France for a panel series hosted by Adobe and Forbes. In the first of three panels, entitled GenAI In The Wild: A Look At How To Employ In The Enterprise, Forbes senior-vice president of research and insights Janett Haas moderated a discussion with the following speakers:
- Eric Hall, Senior Vice-President And Chief Marketing Officer, Digital Experience, Adobe
- Phil Regnault, Partner And Principal Adobe Alliance Leader, PwC
Read on for highlights of their impactful conversation, which has been lightly edited for length and clarity.
Q: Eric, at Adobe’s summit earlier this year, you said: “The lines have been blurred between your marketing, your sales, your reps, your products and support. In today’s world, your digital experience is your brand.” So, what do seamless customer experiences look like in this era of AI?
Eric Hall: As we interact with our consumers, or as consumers interacting with brands, we’re not seeing the department. Our customers don’t see the department that we’re part of, and they experience different aspects [of] our brand through an app, they experience our brand through a website. Whatever the experience is, that is the product. A lot of times, that is very digitally driven, and as AI gives us the ability to have a lot more automation, that’s an opportunity to make things better, and do things faster. It’s also a way to trip up and make a lot of mistakes, so we want to avoid that. But, I do think we’re in a space now, where there’s a lot that will be possible coming online over the next couple of years, and we might as well embrace it—because if we don’t, someone else will.
Q: What does it mean to be a marketer in the era of Generative AI?
Phil Regnault: It means we’ve entered the golden age of marketing. I think this is the best time to be a marketer. If, maybe, the origin of marketing was kind of the Mad Men era of the 50s, and advertising and all that—I think now this is, in some ways, a resurgence of the value of marketing actually being observed as being contributors to positive outcomes.
What’s changed over the last 18 months is that the clients that we work with, the executives that we survey, are saying that marketing is going to play a much more prominent role in this notion of business model reinvention. Two years ago, 48% of our CEO clients responded saying that they believe that their own business model would go the way of the dodo by the end of the decade. They were saying that their companies were just not going to survive. We just released a survey this week with an update on that stat: 64%. That’s across the C-suite, but if you survey CMOs, it’s actually in the 70th percentile. So, there’s this consensus developing in the C-suite that these business models are no longer going to be relevant unless they reinvent. Coming back to Generative AI, [we are] breaching this new frontier leveraging GenAI to augur the golden age of marketing, in my opinion.
Q: In terms of maturity [level of Generative AI], are we moving beyond that proof-of-concept project stage to harnessing the power of AI? Where would you say we are?
Eric Hall: Well, I think we’re at zero-dot-one. Seriously, I think that very few companies have had the chance yet to train AI on their data, and then have that be shown in process to their customers. So, what folks are doing now is very much in the playground phase; there’s a lot of work happening, certainly with creativity and just with basic core content. And that’s fun, because we’re seeing the death of the blank page, because ideation now is very easy—and that’s really valuable. I no longer have to write a brief, hand it to someone else, and say, “Please, based on these words, interpret the image that’s in my mind.” I can actually go and ideate myself with tools like Adobe Firefly, and get back 10 different ideas in the space of minutes that it would’ve taken someone else hours and hours and hours to do. …
However, for the enterprise, for any business of any size, to really take advantage of AI, you’ve got to have all of your privacy and IP protections in place—most companies are still figuring that out. You have to be able to train the AI on your company data, and then allow that to be part of workflows; no one is doing that at any kind of scale yet. … And then, ultimately, the AI has to end up in workflow. It can’t simply be a tool over here, where I mess around and I bring it back to do my real work. The ability to take the AI, put it into applications, have it trained on your company data, and provide a really unique level of automation—in terms of both what the marketer is doing, but also what we’re showing customers—[is] at the beginning.
Phil Regnault: I would say we’re two out of ten [in terms of GenAI maturity level]. But, it’s less about the placement on that spectrum, as much as the rate of acceleration because that is actually just staggering right now. So, whereas 18 months ago, maybe the focus of the conversation was a lot on the tech, the models, the algorithms—now, it’s shifting into the actual application of those models and innovation. Like getting the AI assist embedded in the Adobe Experience Platform [and] mainstreaming the workflow.
But what’s more interesting than the developments on the technology front, is the attitudinal shift that’s happened, and again, the rate of change there. I think what’s really important, particularly for these large language models, is that they’re no longer closed black boxes. Companies like Adobe open them up to allow you to actually customize those models, add your brand book in there, and achieve more brand compliance—not to mention regulatory compliance as well. [That] opens up a whole new frontier. So, that attitudinal shift is happening overnight, seemingly.
Q: I’m interested in hearing about the impact of Generative AI within your own marketing organizations and how you’re bringing that out in your work with customers.
Eric Hall: [There’s a] couple different areas that we’re having fun with right now. Certainly, we’re all in on the notion of [doing] interesting things from the initial content creation front. Our teams are building Firefly, and they’re the power users of Firefly. So in all manner of aspects of Adobe, people are pretty excited to be like, “Here’s my deck where I have now added all of these amazing images from Firefly, which I could not have even done before.” Even people who aren’t in creative [roles] are now taking creative and putting it into their documents.
Our social teams are [also] using Adobe Express, which has Firefly embedded, to do very quick-turns on anything they’re trying to produce. Our performance marketing teams, who are responsible for all of the Adobe paid media, [have] completely changed how they do A/B testing. Now, what they can do is put a hundred variations in the market, and [in about] 24 hours, have pretty good telemetry on which of these is working. Then, they can take the content attributes of what’s working better in market, and build out variations on those attributes instantly. And then, they can put a new set of a hundred things in market. So, even the way that they’re thinking about data-driven testing has been totally turned on its head because you’re not limited by variations anymore. So, there’s a lot of different ways that our folks are experimenting and breaking new ground.
Phil Regnault: It’s not just that creatives and core marketers get to do more and experiment more, and deliver better and faster outcomes with solutions like Firefly and other GenAI solutions—but everyone else in the enterprise can become a marketer now and contribute to the marketing cause. I think that’s interesting; everyone becomes a creator, basically.
A year and three weeks [ago, we] announced that we were going to make a billion dollar investment in GenAI. Fast-forward now a year, and there’s been a lot of impact. … On the business end of things, we subscribed to Firefly. We [also] announced a few weeks ago that we are now the largest enterprise user worldwide of OpenAI. We’re their first reseller of OpenAI. So, we’re very serious about this. And we’re applying it internally. First, about six months ago or so, we released an internal tool called ChatPwC. It’s available to all 370,000 people in the firm, and they use it for a myriad of daily tasks. But, we’ve gone a step further now [and] created a destination for them on our internal sites where they can, depending on their job function, actually use these tools that we’ve built with GenAI to speed up their work. For some tasks, [there’s an average] 20 to 30% time savings—and for some tasks, as much as 70%. So, imagine what the marketer can do with that newfound time.
Q: We know that data quality, data security, data privacy and ethical concerns are top of mind. Talk about what responsible AI means to your organizations, and how you’re leading the conversation about responsible AI practices.
Eric Hall: When we set out to build our AI models, this is a very thoughtful consideration for us. We obviously have a huge investment in the creative professional community. It’s a group of people globally that we care a lot about, and as we think about the role of any technology and its impact on creativity and intellectual property, these are issues that are very much core, and near and dear to our heart. When we set out to do this, we said we have to be able to do three things: We have to be accountable, responsible and transparent.
Accountable, meaning we have to take full ownership over the outcomes of the models we build. As we think about what we are setting out to build, what does that mean? What’s the data that’s going to be used in that? How are we training the models? How are we testing the models? How are we putting guardrails on models? All of the things that you need to do to sort of manage any aspect of potential harm in a lot of different directions. The responsibility part has a lot to do with the data that you choose to use. So, we set out very early to say that we are not using any data that we do not have clear intellectual property rights to use. And so, Firefly is built on licensed content and/or things that are clearly in the public domain. We set out very clearly to say, “[Being] responsible is about the data that you use.” Because, at the end of the day, all AI models are an augmentation of a lot of data. Then the third piece is transparency. Can you open up the hood? Can you see into the engine and say, “Where is this coming from? What’s the data provenance?” [These are] all the issues that you should care about, and if you don’t have those three things—accountability, responsibility and transparency—there’s a problem. That’s how we set out to build the AI.
Years ago, Adobe was a founding member [of the] “Content Authenticity Initiative,” that now hundreds and hundreds of other technology organizations [are part of], and PwC has joined as one of many partners with us. And that’s about being able to trace content from its inception: Who built it, when did they build that content, what changes have occurred to the content. And, ultimately, that’s going to be a paradigm shift.
Phil Regnault: As far as the definition of responsible AI, we view it through two lenses. One is through just the technology risk, and being responsible in mitigating that risk, and the second one is more about the empathy related to how you apply it.
On the technology risk, there’s the underlying data. Are you using proof data? Is the algorithm itself [a] responsible algorithm? Does it take into account the right levels of diversity and inclusion, for example? Is there no hidden bias in the algorithm? Make sure that the output is reflective of what you would hope it to do, not to mention from a regulatory compliance point of view. … On the empathy side of things, I think this is also controllable, but it really goes back to the attitudinal shift I was talking about before. Is the marketing that you intend to do with your GenAI of the creepy-cool variety of personalization at scale, or is it one that is truly customer-centric? And I think that that requires the CMO to actually inspire their teams, as well as their peers throughout the enterprise, to adopt that customer-first mentality, and actually just ask the question: “May I use your data?” As opposed to trying to be underhanded about it, be transparent and cede ownership of the data, whether it’s actual personally identifiable data, along the privacy realm, or your creations. Get explicit permission to use it and I think everything will turn out fine.
Q: Last question. It’s a fill-in-the-blank. If 2023 was “the year of the playground” for AI, and 2024 is “the year of production,” then 2025 will be the year of—blank—, for marketers?
Phil Regnault: Trust.
Eric Hall: Reinvention.
Janett Haas (Moderator): Trust and reinvention are a great way to end this panel. I want to thank both of you for sharing your insights and your leadership with us. Thank you to everyone at Adobe for this great partnership, and making this event series possible.
