Why generative AI is an optimization tool, not a creative savior
Many teams now accept AI for brainstorming as long as it stays out of the final work. In this clip from You’re Not the Audience, Parliament Creative Director Calvin Ross Carl argues that’s backwards: generative AI is an optimization technology built to predict the statistically likely, so it can’t make the fully lateral leap where creative magic happens. He also explains why optimization should protect innovation rather than replace it, and why the biggest threat to most brands isn’t a big swing that misses but blending in. It’s from the You’re Not the Audience episode on best practices and innovation, How Best Practices Kill Innovation.
Watch the full episode
Outline
Summary
Calvin Ross Carl takes on one of the most common uses of AI today: generating ideas and brainstorming. AI tools are fine if they help, he says, but it’s worth remembering what the technology actually is. Its job is to predict the statistically likely next thing, which means it can only extrapolate from prior data and prior best practices.
That’s why the growing acceptance of AI for concepting, as long as it isn’t used for final work, makes him nervous. Concepting is the moment where the magic happens, when a person reaches into the ether and pulls out something completely unexpected. Viewers in the chat agreed that AI may help with mood boards and mockups, but never produces a truly lateral idea.
Calvin closes with a broader point about optimization. Every company needs it, but it should protect innovation, not replace it. Brand strategy is about uncovering a deep, often immeasurable truth, and the biggest threat to most brands is becoming indistinguishable from competitors who all follow the same best practices.
Highlights
Generative AI is an optimization technology that predicts the statistically likely.
Using AI only for concepting gives away the moment where creative magic happens.
AI output tends to be adjacent to your thinking, never a fully lateral move.
Every best practice exists because someone once ignored a prior best practice.
The biggest threat to most brands is becoming indistinguishable, not taking a big swing.
Key insights
AI extrapolates from the past
Because AI is fed only by what has already been done, its ideas are extensions of existing data rather than genuinely new directions.
Brainstorming is where the value lives
Treating concepting as a safe place for AI misreads the creative process. The early leap is the hardest, most human part of brand and design work.
Optimization should protect innovation
Sales, data, and testing are essential, but when every new idea has to prove itself with numbers first, optimization ends up replacing the chance at innovation.
Brand strategy isn’t measurable research
Organizations often expect strategy to deliver provable data points, but its real job is uncovering a deep truth about a company, which is far more ethereal.
Detailed analysis
The clip reframes the AI debate around what the technology is designed to do. If a tool’s purpose is to predict the most likely outcome, then asking it for unexpected ideas is a category error. Calvin doesn’t dismiss AI outright; he acknowledges it may be useful for pulling references or mockups, and allows that machines may get better at lateral thinking someday.
The second half connects that argument to how companies operate. Agencies like Parliament sometimes butt heads with clients because those clients are built for measurable optimization, while brand work depends on knowing when efficiency is the wrong tool. Calvin acknowledges that a big swing can hurt when a company bets everything on one project, but for most brands the larger risk is sameness.
Conclusion
AI can support creative work, but it can’t supply the leap that makes a brand distinct. Calvin’s advice is to keep the magic human, let optimization protect innovation, and remember that nothing about best practices is set in stone.
Transcript
But the one thing that I think gets talked about a lot right now is how people are using AI for ideas and using that for brainstorming and things like that. That’s cool. AI tools, sweet. Use them if they help you. But we just have to remind ourselves: a lot of folks are using AI for generating ideas, for brainstorming, but it is worth remembering that AI as a whole is an optimization technology. That is what it is. Its whole job is to predict the likely next thing, statistically. That is its actual job. And it’s incredible how it works, but if you’re always predicting what is statistically likely, you’re only basing it off of prior data, prior best practices, all of that stuff. So you’re not ever really getting a truly new, innovative idea. It’s just an extrapolation of what you’ve fed it. And that’s always the big red flag for me when I hear a lot of people talking about using AI in brainstorming.
I have a lot of really complicated thoughts about AI and using it in creativity. But one thing that always makes me nervous is you’ve started seeing almost an acceptance where, when people are talking about AI, they’re like, “No, no, no. As long as it’s not for the final stuff in the end, I’m okay with it.” If you’re using it for concepting, you’re using it for brainstorming, that always puts me back on my heels a little bit, because that’s the moment where the magic happens, especially in brand and design work, where you reach into the ether as a human and pull out something completely unexpected. I’ll say that’s a lot harder. It’s much more difficult to do that, nearly impossible, but somehow we find ways to do it.
That’s why I always get really nervous when I hear about people using AI for generating concepts and brainstorming: you’re taking away that chance of finding true magic. You might find something insightful. You might find something that you can sell to your boss. You might find something that proves a point or a perspective, but it’s really hard to get a truly unimagined idea.
So, Orion says here, “It’s good for mood boards, example imagery, but not for the act of inspiration itself. I’ve never felt inspired by a creative solution that GPT has spit out.” Yeah, I think that’s totally how people are often thinking about this, and I totally get it. Mood boarding stuff, it’s probably good for pulling things. I’ve seen people do it for design stuff, like it’s good for throwing designs into mockups and things like that. But that’s where I come back to the same thing that you noted there: I never end up being inspired by anything it spits back at me, because anything it would spit back to me is just adjacent to what I’m thinking about. It’s never a fully lateral move.
And yeah, Jared’s here: it’s only fed by what has been done. Yep. That’s the challenge. We humans are our own little filtration machines, and the only way we ever get to new ideas is by understanding something about the old. But for some reason, machines—yet, I should say yet—aren’t great at that super lateral thinking where you jump completely outside of the knowledge base and run with a different idea. So maybe we’ll be there someday. I don’t know. But in the meantime, I’ll keep it old school for now.
Okay, that’s a lot of thoughts about optimization and innovation, but if I have any final thoughts about it, I do like to also remind people—because I don’t want to seem like I’m totally dragging on that side of the business that’s focused on optimization, like the sales department, the people who need their data—those are good. Every company needs optimization. But it should be there to protect innovation, not replace it. That is definitely one of the biggest challenges: optimization often replaces a chance at innovation, because you have this crazy idea and someone asks, “All right, let’s see some numbers behind it,” or, “Let’s test it,” and you’re like, “We won’t know until it’s done.” That’s always the challenge.
So it is definitely worth remembering that every best practice that has ever existed and exists today is there because someone ignored a prior best practice. That’s a really important thing for us to remember: nothing is ever set in stone. Even when somebody says you need to follow up, do this, crunch these numbers, sometimes it’s okay to say no. And the big goal is to just know when efficiency and optimization is the wrong tool.
We run into that a lot, especially in our side of the biz, brand strategy. Brand strategy is often perceived as a research process that results in some kind of measurable or provable points or measurable or provable data. That’s actually really the opposite of what brand strategy is usually focused on. It’s usually about uncovering some deep truth within you, so it’s much more ethereal. It’s much more immeasurable. And that’s what happens when these orgs hire outside help, when they bring in an external party like Parliament: sometimes we butt heads, always in a nice way, because they’re a company that’s focused on that kind of creation of an idea as opposed to our style.
There is generally a pretty big fear of making that choice. But it’s also important to remember that the biggest threat to most companies isn’t taking a big swing. Usually, if you take a big swing and it misses, it’s not actually that huge of a—I mean, sometimes, if you’ve invested all of your millions into one single project—we’re seeing this some in the video game world now—maybe that big swing can be bad. But the biggest threat to most brands isn’t taking that big swing. The real big threat is becoming indistinguishable from all those other brands, because you’re all just operating off of best practices. So yeah.