‘Is it even worth it?’: Confessions of an agency exec on client challenges with generative AI peak

This conversation has been lightly edited and condensed for clarity.

Marketers are already being asked to do more with less. Generative AI’s promise was efficiency, but there’s client apprehension. What has that looked like for your agency? 

Meta’s new generative AI tagging, “Made with AI,” I think that that is wonderful. All the platforms are actually moving toward this tag that will tell an end user, a consumer, a person that what they are seeing has been touched by generative AI. The problem that we’re having right now is that if you even remove a sun flare using [Adobe Firefly] within Photoshop, something that you would use the healing brush for, it will tag that photo as being made with generative AI. 

Can you spell out why some marketers may have an issue with that?

Let’s say [we have] an automotive client. There’s a picture of a beautiful vehicle and now there’s a tag that says “Made with generative AI.” You can imagine that they are not pleased with that,  because they do not want to be communicating to a consumer that what they are seeing, this product, is not real, or it’s been enhanced. And so we’re in this very strange space where photo retouching has been happening since there was Photoshop. And now, that photo retouching, because it’s being done with a generative AI tool, is now calling into question the credibility of the image or asset itself. So that is the situation that we’re currently dealing with .

What does that client apprehension mean for how your agency uses AI-powered tools? 

Until we figure that out, is it even worth it using Firefly within Photoshop because clients don’t want that tag. From a creative team’s perspective, they’re like, ‘We are saving hours using this stuff for some of these more image enhancements.’ It’s such a small thing, but it just shows you there are big ripples that this is going to create. You’ll question a lot of stuff. It’s embedded. If you look at any of these tool sets, for example, Unity and the gaming engines [Unity, a game engine and development platform, recently published a report stating 62% of game developers are currently using AI tools for game production], those tool sets have generative AI in them. That’s how they’re producing environments. We’re not going to be able to get away from it. 

Does that change your opinion on generative AI and how it should be used? 

I think though, that this speaks to the amazing use cases for it, especially around very specific use cases where it’s brilliant. These tiny, yet deep, applications are where we see so much time savings, scalability, you see the increase of productivity. You’re able to get to that thing, that brilliant thing, much faster. That’s where it could be remarkable. The broader application is where you can see it might be going a little off the rails. 

How do you talk clients through how the agency uses AI tools? 

In that case, it’s being really clear and very tight [with clients] around the definition of what is the use case, these are the tools that we’re using, it’s proving out, showing end product. It’s really getting the client comfortable, and understanding that the tools we’re using are covered. They have the stamp of approval from a safety perspective that these tools are not using anything that’s being input as training data. Eventually, we get clients there. They become more comfortable.