We're prompt-engineering our way to mediocrity

So this is probably the first ever podcast episode you've ever listened to about the use of AI in marketing, right? Okay then, maybe not. So before I start, there's a lot to admire about how AI is being used in marketing now, and the potential of AI within marketing, which we're already seeing the beginnings of, is a lot to be excited about. Okay, all of that boring, non-creative, repetitive stuff should definitely be thrown at AI, and it is. Most of the performance marketing stuff, the pragmatic style ads, the banner ads, the A/B testing, the rationality-based copywriting, you know, stuff like features, tech specs, colors available, all of that stuff, all the basic sort of grunt work, you know, test and learn stuff, AI can do it better and it can do it faster. Um, but that's not really what excites me. It's where it gets interesting is where it can do things that humans either can't do quickly or can't do at all. You know, things like gathering large amounts of data and working with what-if models around that data. You know, AI can link all the moving parts together and help us with strategic planning. For example, the ability to gather category entry points, create more detailed marketing segmentation, identify the optimum position, build funnels and work out the best leverage points, right? Recommend new product development ideas, all of that stuff. That's where it gets really interesting, and how those disparate areas can be linked together so that you can go back and anticipate trends, market drivers, changes in buying habits, all of that stuff. Superb. All all great and wonderful. So I don't want you to come— I don't want to come across as being some kind of Luddite, you know, Neanderthal type person, because I'm definitely not. We use AI pretty much every day within our agency, as I'm sure any forward-thinking agency is doing as well. What I'm talking about is that rather than use AI to make us and our brands better whatever better means in this particular context, and the one that sort of floats your boat. We're focusing on its use to make us faster, even if our standards of quality take a dive as a result. Pretty much every AI-generated image that I see accompanying an article or a blog post or an ad, it just makes me want to projectile vomit. Right? And I'm sure it's not just me. As an industry, we've spent decades teaching people to stop it with this low-quality stock photography stuff that seemed endemic wherever you were subjected to any kind of marketing or sales collateral. You know the sort of thing, right? The man with a headset looking straight at you on the Contact Us page, right? Or the woman sitting on her own laughing while eating a salad. For some reason, right? And yet here we are churning out great swathes of the stuff using AI without a care in the world. The thing is, our brains evolved. We spent hundreds of thousands of years learning to spot things that don't look quite right, and it was in our evolutionary interests because getting it wrong meant death, right? Oh, look over there! Is that an innocent-looking piece of driftwood flowing down the river? Oh no, it's a crocodile, right? So the ancestors who kept missing that distinction didn't stick around long enough to become anyone's ancestors. Our biological evolution was also dependent on our ability to read human faces, to understand underlying emotions, during communication— anger, fear, happiness, whatever. And our brains have become fantastic at pattern recognition, okay? Show someone who's got no interest in AI, doesn't think in terms of images, is not looking at it in a technical way— but if you show, you know, a Luddite, if you show a civilian, as I'll call them, enough AI-generated faces and tell them what to look for, and they start noticing the usual stuff that we see, right? The waxy skin, the shadows that don't quite fall right, and how the eyes look through you instead of at you, right? All of a sudden they can see it, and once you see it, you can't unsee it, right? So when we're doing all of this really basic, lazy, sloppy AI image generation We're inadvertently educating our audience to recognize our shortcuts. The indiscriminate and unsympathetic use of AI-generated imagery within marketing is now seen by our audience in the same way as they looked at those low-quality royalty-free images. We've forgotten, or maybe we don't seem to care, that our audiences use these tools too. So they know exactly what we're doing, and what they're thinking is not how cool and with it we are. They're thinking how lazy we are because we're using AI images that clearly look like AI images. Now, Bill Bernbach, the advertising guru, the person who probably single-handedly invented good advertising, is quoted as saying, nothing kills a bad product faster than good advertising. Now, what he meant by that was that while poor products might linger in obscurity with little to no promotion, excellent advertising will quickly expose a product's flaws. Nothing exposes lazy thinking faster than a sophisticated audience. The craziness doesn't stop here. Since we're not actually being more efficient by doing this AI stuff for image generation. For, like I said, I'm, I'm making a particular use case here. What we're doing is just moving the effort from one place to another. Instead of finding the right photographer, designer, creative director, whatever, we're prompt engineering our way to mediocrity. And irrelevance. It's as though we've decided that it's a good idea to replace a Michelin-starred chef with a very expensive microwave. Psychology tells us that people judge quality by the worst element they notice, not the best, okay? In which case, our brilliant strategy inevitably gets undermined when it's paired with imagery that signals to people we couldn't be bothered. So perhaps the real question we need to ask ourselves isn't whether the AI images look fake, but whether looking fake has become our brand. Thanks for listening. If you like what you heard, please pass it on. I'm not saying I'm right, I'm saying many out there are wrong. Please subscribe, share, comment, whatever works for you. Find out more at kexino.com. That's K-E-X-I-N-O. Hope to see you again here soon.