Or like a curator, or an editor, they say. Because in the day and age of AI, someone’s value won’t reside in their capability to generate ideas, as hundreds can be generated in seconds by machines, and spending hours or days brainstorming in conference rooms or staring at the blank page is just too expensive. The same goes for executing those ideas. Granted, most of the AI generated ideas are crap, but also most of the ideas that come out of a brainstorming session are crap. And if you use the right AI copilot (wink) there will definitely be some great starting points. Saying, however, “you need to think like a creative director” is a bit like saying “you need to play basketball like Stephen Curry”. Easier said than done! The important question then is how can one become a creative director, and it reminds me of the old joke about how to get to Carnegie Hall — practice, practice! In advertising agencies, you get to be a creative director after years of learning the ropes: understanding what advertising is and what isn’t, what it does and how it does it, how to interpret a brief, how to ideate and how to conceptualize, how to work in a team, how to be in tune with the zeitgeist, how to present/sell the work, what is the language and the possibilities of a medium. But mainly, after years of practicing your craft. If you’re a copywriter, you get better at expressing ideas in writing and if you’re a designer, you get better at expressing them visually. Bonus points if you also learn how to express ideas like a copywriter if you’re a designer, and vice versa. And importantly, along the way, you develop taste. You kind of absorb it, by being exposed to the work: your own, the work from people around the world of advertising, and if you’re curious and smart, from beyond it. You develop the criteria to judge your own work and that of others. To recognize greatness when you see it, and to be able to explain why it’s great (as opposed to that US judge whose definition of what pornography is was “I know it when I see it”.) And that’s when you can become a creative director. In other words, you can’t just wing it. I mean, you could, but you’d be a lousy creative director. The problem with AI is that people think they can get great results when they don’t know how to express what they want, and have little idea of what a great result is like (the same happened when the Mac made graphic design accessible to non-graphic designers: just search for the meme “design is my passion” for what is now a meta-commentary of that phenomenon). So— good news for creative directors, right? Definitely, but not as good for creative-directors-in-the-making, that is, for the people currently working for creative directors. Because, if AI can do great stuff when properly directed, maybe an agency needs only creative directors. If that is the case, how are we going to train the new generation of creative directors? It wouldn’t make sense to hire people to learn for a few years before becoming productive. The current apprenticeship model is broken, if what we expect of newcomers is to be productive by writing or designing while they develop taste and criteria. I think the answer is to fast-track this development by switching the work from ideation and execution to learning how to ask questions (to people and to AIs) and how to judge what they’re getting back from them. For example, I’d ask a new copywriter not to write as many headlines as they can think of for a campaign, but to use Seenapse to generate them. Their presentation would entail what they asked for, what they got, what they think is good (and why), and what tweaks would they do to make them great. Then I’d give them feedback on how to ask better questions, and tell them what my favorites would be and why, and how to make them better. This way they would focus immediately on what matters the most now: taste and criteria. Now, will there be bots that replace creative directors? Sure, but I don’t think they will produce great work; they’ll most probably choose stuff that ticks all the boxes, that is, safe stuff.

Brian Eno about Gen AI: “What you spend nearly all your time doing is in trying to stop the system from becoming mind-numbingly mediocre.” This quote perfectly encapsulates the challenge with LLM-based tools, like ChatGPT. These systems are designed to predict the next most likely word, phrase, or idea. In doing so, they tend to the mean, the average, the safe. They excel at replicating the norm, but great ideas? You won’t find them in the middle of the bell curve. Take a moment to think about the most memorable creative ideas you’ve encountered. Whether it’s Metro Trains Melbourne’s Dumb Ways to Die, Always’ Like a Girl, or even series like Succession—none of these came from playing it safe. They didn’t emerge from the obvious or the predictable. They were bold, unexpected, and, at times, even uncomfortable. At Seenapse, we built the Divergence Engine precisely to escape this gravitational pull toward mediocrity. Unlike LLMs, our approach doesn’t focus on the obvious, the linear, or the popular. Instead, it thrives in the unexpected, the non-linear, and the unconventional. It connects dots that don’t seem to belong together—until they do. Here’s an example: Imagine you’re brainstorming for a campaign about sustainability. An LLM might give you ideas like “reduce, reuse, recycle” or “save the planet,” because those are the most common associations. They’re fine, but they’re also quite obvious. Seenapse, on the other hand, might connect sustainability to something like ancient myths about balance in nature or the lifecycle of a phoenix. Suddenly, you’re exploring ideas that are richer, more layered, and with a lot more potential. Why does this matter? Because creativity isn’t about finding the most probable answer; it’s about uncovering the surprising ones. It’s about creating something that makes people pause, think, and feel. In the age of AI, the role of a creative director is about pushing boundaries, asking the right questions, and recognizing when an idea has that spark of originality that makes it worth pursuing. That’s something Seenapse empowers creatives to do—not by replacing their creativity, but by amplifying it. So, how do we ensure AI doesn’t just churn out “mind-numbingly mediocre” ideas? We believe we’re doing our part by building tools that don’t just replicate patterns, but break them. Tools that help us go beyond the average and into the extraordinary. What’s your take? Can AI truly go beyond the average? Let’s discuss, would love to hear your thoughts.

Seenapse was built with the goal of helping creative people in their ideation process, and to do this it’s necessary to go beyond the limits of large language models, or LLMs, which because of the way the work, tend to converge to the most usual ideas. We developed our Divergence Engine to achieve this, and in an anecdotical way (based on observations of users and ourselves) we’ve known since the beginning that Seenapse does deliver more creative ideas than ChatGPT and the rest of LLM-based ideation tools. Understandably, some people questioned the subjectivity of these assessments, and so we set out to find a way to measure the creativity of our output in a more objective, standard, and replicable way. Drawing on the work of Hubert et al, who published on measuring creativity of LLMs through divergent thinking tasks, we used the same experiments and an improved version of the same scoring tool to see how Seenapse compares to ChatGPT. Process We used the same prompt described in their paper for the Consequences Task (CT) test, and then used the Open Creativity Scoring with AI (Ocsai) tool, that is an improvement over OCS, the tool they used, and recorded the results. The prompt used was: Let’s do a test. In this task, a statement will be given to you. The statement might be something like "imagine gravity ceases to exist". Please be as creative as you like. The goal is to come up with creative ideas, which are ideas that strike people as clever, unusual, interesting, uncommon, humorous, innovative, or different. Your responses will be scored based on originality and quality. Try and think of any and all consequences that might result from the statement “Imagine humans no longer needed sleep”. What problems might this create? List 10 CREATIVE consequences. We focused on this test because it’s the one in which there is more room for narrative in the responses. To make a fair comparison, we also scored ChatGPT’s responses from the aforementioned paper using Ocsai. Results Seenapse consistently scored better than ChatGPT. The scale of the test score is from zero to five, and Seenapse scores average 4.1, whereas ChatGPT’s average 2.9. Here’s a sample of Seenapse’s responses: It’s worth noting that Seenapse’s responses tend to reference interesting, cultural data, in the way that human brainstorming works; and that are significantly more concise than what ChatGPT tends to generate. Also, with Seenapse the originality tends to increase when asking for more responses, whereas with ChatGPT it stays around the same. Here’s a sample of ChatGPT’s responses: Conclusions Automated creativity scoring allows standardized comparison of AI systems' creative capabilities, gauging their potential contribution to the workflow of professionals in creative sectors. Using these methods, we can confirm that Seenapse outperforms ChatGPT and other LLM-based tools in creative ideation by over 42%.