AI Gives Itself a 10 out of 10

Last year, Nike released a spot called “Sunshine”, a tribute to sport, movement, and energy. Someone fed it to AI, and the result was surprising: the remake scored better than the original. Not emotionally, but statistically. 
Neurons, an AI-driven testing platform, gave the remake higher impact scores for both brand building and conversion. 

It sounds like a turning point, as if AI can not only help create, but actually improve. Until you question the yardstick itself. Because who judged the AI version? Exactly: AI. 

And that raises an interesting question, does AI simply prefer the AI-made version because it speaks the same language? Sharper lines, higher contrast, cleaner edits, those are precisely the signals algorithms recognize and translate into “clarity.”

But people see differently. We notice the noise, the small gestures, the imperfections that make something feel authentic. What AI reads as confusion or inefficiency might actually be the emotional core for us. That faltering moment, that slow glance, those are the things that stay with us.

The danger, then, is that we start optimizing advertising… for AI. That we learn how to feed the models, but forget how to move hearts. A self-reinforcing loop: AI creates content that AI approves of. Sharper, clearer, more efficient, but perhaps also flatter.

That doesn’t make AI worthless. On the contrary, it’s an incredible tool for faster iteration, testing concepts, calculating variations. It gives us prototypes at a speed that was unthinkable a year ago. It lets us play with forms and scenarios that used to be too costly or time-consuming.

But we have to recognize the bias. AI mostly measures what it can recognize. And what’s harder to grasp: nuance, warmth, rawness, gets less weight. Yet those rough edges are often what make a brand human, building connection and affection in the long term.

So the real challenge isn’t: let AI make the ad, we already can. 
It’s: let AI help us get to the essence faster. 
Use it to sharpen, not to replace. Let it help us realize ideas efficiently, but not decide what matters.

Performance is one thing. Making people feel is another. 

At Serviceplan MAKE, we use AI intelligently across the entire marketing and communications value chain, from strategy to generative applications to measurement. Always to gain efficiency, sometimes to boost effectiveness but never to replace human creativity.

Written by Francesco Caccamese, Strategic Director at Serviceplan Belgium