Tackling the conundrum.
Yes& Founder and CEO Bob Sprague sat down (if one can truly sit down on a Teams call) with Yes& VP of Digital Strategy Chrissie Koeppen and President and Chief Creative Officer Josh Golden recently and—to his surprise—found them more in agreement than not on the topic of AI vs. authenticity. What follows is an edited transcript of their conversation.
BOB: Thanks to you both for joining. So, here’s the question: as agencies and brands we’re all being told we need to embrace AI. At the same time, we know that what consumers want more than anything else is authenticity. How do we reconcile those two things?
CHRISSIE: First, we should talk about which AI we’re discussing. There’s a difference between AI-generated and AI-assisted. In the industry we have to find a way to make that really clear, because, as we know, one of the big keys to authenticity is being transparent. If we’re transparent about how we created something, even if it is purely AI-generated, it can be authentic. For example, there are AI influencers on social media, and GenZ loves them. But guess when they love them the most—when they are paired with a human counterpart.
JOSH: When you go to a museum and you see the Monet hanging in D’Orsay, you can see the paint and see that there’s something visceral and emotional that reflects Claude’s strokes (and I use Claude purposely because that’s also an AI platform).
BOB: I saw what you did there.
JOSH: Then you could go to the museum store and buy a print, and it’s not the same thing. I like to believe that AI is at its best an observer of human behavior and an imperfect interpreter of it. So, authenticity always requires some level of collaboration with the tool.
BOB: So, what do we think authenticity really means? When people say they want authenticity from a brand, what are they asking for?
JOSH: I think they mean consistency to a promise that the brand has made, and the intersection of that promise with what they actually want from the brand.
CHRISSIE: Yes, it’s about your core brand values and consistently conveying them. Are you showing up authentically as you are in a way that resonates with your audiences. It’s a nuance, but humans are good at nuance.
JOSH: Yeah, and that’s what AI is not very good at.
BOB: So even if you’re transparent, and even if you have a human interlocutor, you’re still taking something that did not originate from a human being and putting it out as a representation of your brand.
CHRISSIE: We could push back a little on that, Bob, because we have to train AI. The results are only as good as the prompts we put in. The better question may be how AI-generated content is performing. As opposed to surveys, which I tend not to love because they represent what people say they’ll do (versus what they actually do), Patel Digital has studied the way people respond to purely AI-generated content. Whereas AI-generated content takes about a quarter of the time to produce, human-generated content gets 5.44 times more traffic and engagement on a website. We can talk about ads, too. Human-produced ads get 45% more impressions and 60% more clicks. Why? We know that 70% of decisions are based on an emotional trigger, right? That’s what humans are good at: understanding and conveying emotions and nuance.
JOSH: Where AI might be remarkable is in its ability to read what people are looking for quicker than humans can read what people are looking for and put out human-generated content in the appropriate place. So, you neither lose the quick, computer-driven insight nor the emotional intelligence of the human-created communication. You can use an LLM to check you, to put stuff in and ask ‘how might I make this better,’ and what it’s doing is trying to ascribe a certain set of rules to the other things around it. But human emotion is where you break the rules, and that’s where the collaboration comes in.
BOB: OK, so how do things change if AI keeps getting better, especially in the gen AI world, and the AI we see now is not what we see in three months, or six months, or five years from now?
CHRISSIE: Until AI is able to bridge that ‘uncanny valley,’ it doesn’t go much further. That term was coined in 1970 by a roboticist named Masahiro Mori. It’s the phenomenon when something is almost human, but then isn’t quite. We can see it. It’s almost like the closer it gets to human, the more we’re like, whoa, no, creepy.
JOSH: And this may be prove to be a foolish statement, but I don’t know that AI ever fully spans that uncanny valley. Because the human collaborator still has to constantly differentiate themselves, we keep moving further astray of AI out of self-preservation and out of an inherent sense of what makes us different. You know deep fakes are deep fakes because they are purposely trying to trick us. Just like seeing magic done by a magician, we’re always going to be aware of the magic even if we don’t know how it’s done.
BOB: So, six months ago everyone was saying “AI will never replace a human, but a human using AI will replace a human not using AI.”
JOSH: I think the argument becomes moot when you think of AI as a tool. We’d be foolish not to avail ourselves of it. I mean a carpenter with a hammer is going to replace a carpenter without a hammer if they need to put a nail into a board. The carpenter doesn’t see the hammer as a replacement of the carpenter.
CHRISSIE: I couldn’t agree more. Bob’s statement is fear-based thinking, and whenever you’re coming from that place of fear you are less able to see the opportunity sitting right in front of you.
JOSH: And when we go back to the original question of authenticity and Chrissie’s point, it would be inauthentic to say that nail showed up in the board by itself. But to acknowledge that a hammer was used in hammering the nail, that is authentic.
BOB: So, what would you tell an ad agency or a brand that was struggling with this conundrum, of how to use AI but also deliver authenticity for consumers who want it?
CHRISSIE: First, they all need an AI policy like we have at Yes&. And then, start small and experiment. See what works with specific customer and audience segments. As long as you’re being open and honest about what you’re doing along the way.
JOSH: Yeah, and I’d say that you have to understand what need you’re trying to solve. AI as a solution is useless without a problem. So, I think you have to ask, ‘how might I employ AI when something is taking me longer than it feels like it should?’
CHRISSIE: How does it save us time, how does it make us more effective, and what does that look like?
JOSH: And we know that the Holy Grail of AI is if we can get it to do our timesheets.
SUBSCRIBE TO THE AMPERSAND NEWSLETTER FOR MORE INSIGHTS FROM YES&: