Welcome to Creatives on AI, our series of conversations with people across the industry exploring the nuance behind AI’s role in our work. We’re shifting our focus past the polarizing one-liners and digging into how AI may help, hurt, and fundamentally reshape our creative worlds. While AI isn't systematically replacing creative roles, it sure is changing things (and fast!). If you want to understand how people are navigating this seismic shift, stay tuned in for new perspectives each week this month.

‍

This week we talked with Karin Fyhrie, founder of Sovereign Objects, a seasonal creative studio that brings together a singular mix of artists, designers, and thinkers to transport companies into the future they’re building.  

‍

Thanks for joining us, Karin! To start, give us a better sense of your studio and what a typical day looks like for you?

‍

I’m the founder and ECD of Sovereign Objects, a seasonal creative studio. Examining what a typical day looks like touches on a deeper question I’ve been trying to figure out over the last five years. Is it possible to design a more sustainable studio model for creatives, where personal work isn’t always at odds with what a person does to pay the rent? In any given week, I, Karin, am striving for a 50/50 split between commercial design projects and an art practice free from constraints — but that split is different for each team member. 

So nothing is typical, when it comes to how I spend or keep time — it’s constantly evolving. Right now, it looks like three days a week building brand systems for companies, from Netflix to Google X to Sana AI. Then one day a week working on personal ventures, currently that’s a new video series called Dream Logic, where I’m exploring some of the creative contradictions of AI with my friend Kirby Ferguson. That leaves maybe one day a week left open for whatever creative urges bubble up to the surface. At least this is the balance I aim for. It’s never that tidy.

‍

‍

If you had to sum up your creative process, how would you describe it in one sentence?

‍

I show up, put the time in, and trust that, somewhere along the way, something will unexpectedly light up and get interesting — even slightly dangerous. Then I follow that spark to a place where I can find meaning and beauty (and try not to worry about what the world will think, which isn’t easy). 

‍

Was there a specific moment or project that prompted you to experiment with AI?  

‍

Yes. Back in 2022, we were creating a healthcare brand for women age 65 and up. When we searched for FPO photos—on Instagram, Google, Getty, wherever—we found nothing that reflected the women we’d interviewed. The visual landscape offered a flat, uninspired take on aging: mostly passive, overwhelmingly gray, and uninspired. Yet the women we spoke with had a ton of energy and attitude, regardless of circumstance. They woke up determined to make the most of each day—all without apology. That mindset was nowhere to be found in the imagery.

So my colleague Emilia Obrzut proposed an alternative path to proof of concept: using early AI face apps (this was before DALL·E), we “aged up” younger women in existing photos which had the vibrant art direction we craved. The results were perfect—they offered a more accurate vision of the vitality we saw in our target audience. That ultimately informed a photoshoot featuring real models aged 65 and up, captured with the same visual energy we had only seen applied to younger subjects. Notably: that business is now booming.

What’s interesting to me about this project—and this AI use case—is that it does the opposite of how GenAI is typically used, and what it typically produces. Instead of perpetuating tired visual tropes, it helped us create something that should exist, but doesn’t. AI couldn’t get there on its own. We had to imagine it.

‍

When did you start exploring AI more seriously?

‍

In 2023, not long after the Silicon Valley Bank crash, business got strangely quiet. Not just for me, but for many. Sovereign Objects was only three years old, so it was still hard to tell if this business dip was “normal.” In many ways, yes—and it’s since rebounded. But from my time at COLLINS and IDEO, I couldn’t shake the feeling that something more deeply unsettling was unfolding. Creative agencies of all sizes — especially smaller ones — were buckling under mounting economic pressures. Like a tectonic death-grip. There, I said it.

The SVB crash amplified all my existing fears. Then GenAI arrived—and multiplied them by 100.

Amidst this growing concern, I decided to take a sabbatical, reset, and consider new possibilities. (This was partly thanks to the seasonal approach of Sovereign Objects, but also a recognition that its structure needed to evolve. I was finding that being at the center of a network meant to support creative health—including my own—was putting me at risk of burnout.)

Around that time, DALL·E was making the rounds, and the images I saw were just awful. All cartoon boobs and neon blue. I thought to myself: surely, there must be a way for creative talent to get their hands on this and make pictures worth looking at.

Then a friend introduced me to Greg Hochmuth, a creative technologist, who taught me about the process of fine-tuning. It allows for the training of custom datasets that can be used to more strongly influence the LLM’s underlying aesthetic. That opened up a new level of control—and with it, a different kind of practice I had not at all expected. It was distinctly uncommercial; something more personal and more overtly psychological.…

‍

‍

‍

‍

How would you describe the way you use AI now?  

‍

It has contributed to a revitalized practice of making work with no boundaries, no preconceived purpose or audience; where I chase down my own curiosity and obsessions. I’m not convinced that what I produce in AI is ‘art’ exactly. But the process, the ideas, the glimpses that emerge through using AI feel revelatory of something deep in my psyche — at least to me!

Let me explain how I use it. It’s late at night. I’m alone in my studio logging onto Stable Diffusion. There’s no brief. No goal. No need state or use case. I begin with an “idea field”—a cluster of loose subjects or moods I may want to explore. I’m not wedded to any of them, but I have to start somewhere. (Over time, I’ve learned that the beginning is always the least interesting).

From there, I follow a thread for six hours or more. The process becomes trance-like—images evolve and overwrite one another in a kind of hypercollage. What starts as women crawling in unison might morph into backlit not-quite-gorillas, then dissolve into some technofolkloric battlefield. The subject shifts. The tone shifts. It’s not linear—and that’s what’s freeing.

I work iteratively, layering image on image into what becomes a stream of conscious stop-motion sequence. Most of what the AI generates—90% or more—I don’t even download. But in that process, something accumulates beyond the 1000s of images. A texture. A pattern. A kind of subconscious map. It feels like a very unadulterated space of pre-thought—sidestepping reason to access the unconscious. Again I hesitate to describe what I’m doing even as ‘art’, but it does resonate for me with what I find most interesting about the surrealist movement: a search not for  a style, but for a way of shortcutting the (self-policing) conscious mind and connecting with some kind of deeper self-knowledge.

‍

‍

‍

‍

What has been the most surprising thing about integrating or using AI tools?

‍

To be clear: I’m not just letting the model take the lead, but I’m not fully in control either. It’s a negotiation. A tug-of-war. I come to understand my own interests through refusal, by continually rejecting what the AI surfaces. That’s the real work. A kind of revelation through negation.

Each discarded image sharpens the edge of what I’m not after. Slowly, something emerges — something unexpected, something I wouldn’t have known to look for. When it lands, I recognize it with my eyes, my gut—not with logic. After so many years in leadership, it’s been refreshing to get reacquainted with that as a compass.  

‍

If you were to prescribe a metaphor to AI, what would it be?  

‍

I’d say an overarching metaphor might be sculpting. I think of LLMs like a big stone I start with. Have you ever seen a geode? From the outside, it looks like just another ugly rock. There’s nothing special about it—except the possibility that something extraordinary may be hidden within. I think of the training data as that outer layer. Mundane. Rigid. Resistant. And from that bulk, you chip away—again and again—in search of a potential gem from another planet. If what I find looks familiar, I don’t bother pocketing it. I’m rock hounding for some dazzling, improbable outlier. 

And that’s just half of it. I’m also trying to surface something in me—some half-formed feeling or idea. It’s not just image-making. It’s almost like emotional excavation.

Tapping into the fear around AI for a minute, what are your concerns around AI and its widespread adoption? Where do you think you fall on the spectrum – optimistic, skeptical or alarmed? 

‍

I’m intrigued, but skeptical… bordering on alarmed. Let me name some fears specific to our industry:

• Artists train these models—without pay or permission.
• Bias is built in 
• We may no longer believe what we see.
• The technology devours energy.
• Many creative jobs will disappear at a rate not easily replaced.
• Everyone who can prompt is now mistermed a “designer.”
• The value of craft is being steadily diminished.
• The world is about to be flooded with even more ugly.

This is not the full story, of course. 

‍

Have you adopted any philosophies around AI and its adoption?

‍

I’m not using AI because I fully endorse it. I’m using it to understand it—to get closer to its limits and possibilities. As a creative, and as a woman, I feel a responsibility to help shape this space before its foundations calcify at breakneck speed.

Technologist and writer Jaron Lanier calls it “lock-in”—the way early design decisions in technical systems tend to solidify over time, becoming difficult-to-undo defaults. These systems are engineered without artists, encoded with the male gaze, and designed to mainline consumer engagement (at whatever cost necessary). AI is likely to shape how we, and future generations, think, see, and live. If artists and people in the humanities don’t step in to show what else is possible, the values embedded in these systems risk becoming permanent.

I understand why many creatives don’t want to touch it—that engaging with AI can feel like complicity, or like participating in our own erasure. But for me, personally, I think it’s more important to engage and critique from a place of fluency, even if imperfectly. If I’m lucky, and open to it, maybe something creatively exciting will emerge along the way. That seems to be the most fertile way in which art and innovation converge.

‍

Do you think your personal philosophy will change over time?

‍

My philosophy is necessarily changing all the time, because the pace of AI warrants it. I’m open-minded – even ambivalent – about AI at the moment, but that could change in a heartbeat.

As for rules of use, I don’t have hard lines drawn—not yet. More like loose boundaries:

• Use it creatively first, not commercially. 
• Limit to one long session, two or three times a month. 
• Aim to unearth something I haven’t seen before.

‍

What’s your stance on transparency around AI use and integration?

‍

While the information environments we spend so much time in aren’t doing us much good psychologically, I’m increasingly concerned about something deeper. It seems likely we’re approaching a threshold where people will no longer trust what they see. And when that trust erodes, the consequences could be serious, because seeing is foundational to how we make sense of – and feel safe within – our environment.

I agree with the trend forecaster and writer Sean Monahan who recently said in his 8Ball Substack, that we are moving from a world of “real until proven fake” to one of “fake until proven real.”

A responsible solution to this would be to design ways to distinguish what’s real from what’s AI-generated. But to do that properly, at scale, would require extensive regulation. Given the regulatory landscape of big tech, I’m not holding my breath. 

What interests me now is how people will respond. I don’t think we realize how massive this shift could be. What happens next? More mistrust? More polarization? A further unraveling of shared reality? I suspect we’ll see some kind of movement where people will start turning back to their senses, what they can trust in their bodies and their immediate surroundings to ground themselves in tangible, tactile reality.  

‍

Is there any part of your creative process you refuse to automate?

‍

I think it’s a question of how we, as creatives, designers, filmmakers, artists, and writers, can collectively define what we want to automate, and how emergent technologies might extend rather than replace human creativity. That responsibility doesn’t fall on artists alone; it also depends on how tech companies choose to build these tools—and who they’re built for.

How do we keep originality, aesthetics, and experimentation at the center of our efforts to build a more interesting world? And how do we ensure these tools are designed to support that vision, rather than default to sameness or prioritize efficiency above all else?

‍

What advice would you give to young people wanting to get into the industry today? 

‍

#1
‍
It will be increasingly difficult to do anything extraordinary.

Though extraordinary will always be possible.

Fight like hell.

#2
‍
People are desperate to be inspired.
‍
In Latin, inspire means ‘to breathe life into.’
‍
That spirit could be you.

#3

Are you inspired?

If not — find that light.

Then keep that flame from dying.

#4

Prepare to be many things.

Because design is always changing.

Use that flame to guide you.

#5

Be good to your body.

Be good to your mind.

Give a shit.

Find beauty wherever it hides.

‍

For fun, if the robot apocalypse does in fact come to fruition, what's one skill you have that will help you survive or thrive in a robot-led world?

‍

I can make a computer cry. 

‍

Karin Fyhrie (née Soukup) is the founder of Sovereign Objects, a seasonal creative studio that brings together a singular mix of artists, designers, and thinkers to transport companies into the future they’re building.  

She held leadership roles at IDEO and COLLINS—where, as Managing Partner of the San Francisco studio, she helped the company to become “Design Agency of the Year” three years running. Her work has shaped brands and initiatives for Robinhood, Dropbox, Netflix, Google X, the MacArthur Foundation, the Exploratorium, and more.