Curiosity

What if clients started cutting agencies out of the creative process? What if the brand team types up a brief and feeds it straight into an AI à la Adam West’s Batman in 60’s TV series. One-way operation. Not much room for interpretation. Not a whole of discussion.

To explore this question we conducted a study with professionals in the branding industry at large. 200+ respondents participated in evaluating the output of just such a scenario.

Five test conditions we set up along a spectrum of AI involvement from “All Hail the Skynet Overlords” to “John Connor 4eveh.” A single generic brief was created to simulate what an agency might receive from a client to create a new brand and it was shared with the five “agencies.” Each one was instructed to create a brand poster that included: a main visual, brand name, brand icon and a tagline.

1. Skynet, Inc: This approach tasked one AI with executing all elements of the brief as a single endeavor.
2. Band of Bots: This approach fed the brief to four different AI but only asked them to do one of the four elements based their strengths* The four elements were then ‘assembled’ into one execution by a human.
3. collaBOT: In this scenario the objective was to “collaborate”, or at least iterate with the AI. The initial brief was tweaked and nudged a series of times as each AI output spurred new ideas in the human. The process was repeated for five cycles, the fifth iteration serving as the final.
4. Dance, Droid!: Maybe where we are now, this scenario is the human directing the AI to produce a specific idea on demand.
5. 100% Human: Just like it says. No bots employed.

*Prior to this study a Pilot Study was conducted to understand capabilities across multiple AI and optimize parameters and prompts.

Not surprisingly, the industry is still wary.

The perceived degree of AI participation in the creative output resulted in significantly lower evaluation ratings regardless of actual AI participation.

In other words, it didn’t matter how much AI was actually used or not, the more a study participant believed the output used AI, the less they liked it. I guess this says a whole about perception in general: if you believe it, it’s true for you. Also, keep in mind the emotional aspect of this study being that it relates directly to our livelihoods, passions, etc. I.e., there was a TON of strong emotion shared during the course of this study. Emotion that I would summarize as viscerally charged.

We know it when we see it, even if we don’t.

There was near Zero correlation with how many perceived AI prompts were used and the actual number

Meaning, participants spotted AI involvement, but it wasn’t clear to them how much AI was used. We know from Finding #1 that the more you think AI is involved, the less you like it. But if we are not great at telling precisely how much, what then?

It makes me think of, “more is not necessarily better,” when it comes to creative concepting. And there’s a difference between ‘having an idea’, and ‘bringing it to life.’ The craft of using design to successfully solve business challenges? Hours and hard work. The deep commitment to understanding the real human needs to be addressed? Hours and hard work. Both benefit (to a point) from the thinking you invest, the more thinking, the better the product.

And creativity? That can happen in a flash, as we all know from eventful showers and slow elevators. But AI is fast too. AI can serve up ‘options.’ Seemingly forever. It can even make wacky combinations ad infinitum. But is that ‘creativity’?

Set two AI’s to ‘work together’ creatively, and you get two algorithms that may not be wildly different from each other, which may have been taught with similar material. So far, so meh.

But throw two humans on the task and you get two individual algorithms that are by definition quite unique. And that’s before they have even had a chance to learn much or have potentially wildly different ‘programming’ experiences. The interactions are exponential and unpredictable. ie Creativity.