Creative Problem Solving with AI: Dan Saelinger’s Collaboration With the Harvard Business Review

When Harvard Business Review (HBR) needed imagery that represented the often solitude of a post-pandemic work-life, they reached out to Dan Saelinger because of his ability to create unique and stand-out visuals using generative AI technologies. Though known for his colorful and whimsical imagery, this project required a more solemn look.  And, with a creative that is seemingly impossible to create in-camera, Dan turned to AI as a way to be able to craft the perfect set of images to accompany the article. HBR’s creative director leaned into this approach and trusted Dan’s instinct. 

Leveraging AI as an artistic tool is something Dan has been exploring in recent years. Seeing it not as a threat, but as an artistic tool in which to use to explore concepts and in this case, create the exact imagery the client wanted within budget and time constraints. Dan approached this assignment with a refined process that balances client input with AI’s unpredictability. The result is a layered, evocative series that goes beyond straightforward visuals, using metaphor and symbolism to make viewers question the nature of workplace connection in our increasingly virtual world.

What was the creative direction for this project?

The concept was based on loneliness in office spaces due to remote work. People are coming into offices only to join virtual meetings or sit in mostly empty rooms, creating this strange disconnect. The creative director had seen some AI work on my site where I’d depicted “stormtroopers” in isolated settings, and he wanted a similar vibe, but with office workers or symbolic figures representing isolation. It was pretty open from there so I had a fun time exploring how we could represent a feeling of isolation through imagery.

What is your process like when using AI to generate images?

I’ve learned to structure my AI process to avoid getting lost in endless iterations. I start with a “sketch round,” where I generate a batch of exploratory images based on the story’s themes. For this project, I generated around 20 initial images, exploring ideas like isolated office scenes and surreal cubicles. My goal here is to offer a range of options to help clients see the story visually. I don’t get too hung up on technical issues in this phase, like missing legs or unusual perspectives. I explain upfront that this round is about sketching ideas, not the final polish.

How do you refine the images in later rounds?

After the initial sketches, the client gives feedback, and I narrow the images down in each round. In round two, I focus on specific concepts the client liked, with more detailed variations. For instance, they wanted an astronaut figure at a cubicle to convey isolation, so I created several variations of that scene. By the final round, I aim to produce close-to-final images, only making minor tweaks based on client preferences.

What was a challenge that came up during this project?

We initially wanted to use astronauts as a representation of isolation. However this was right around the time the astronauts got stuck in space, so we felt out of respect, to pivot that direction. I had already done a lot of generating at that point so I needed to pivot quickly. We went back and forth on options, trying to think of characters that still had that same sense of isolation and exploration. We considered everything from a businessman to a beekeeper! Eventually, we landed on the idea of a bell diver, which gave us the same kind of exploratory vibe but in a completely different context. 

With AI, you can make adjustments that might be impossible in a traditional setting. I started by adapting my original astronaut prompts, replacing "astronaut" with "bell diver." However, AI often brings its own associations into the mix—like "astronaut" tends to skew futuristic and sterile, while "bell diver" can bring in a darker, more rustic tone, which wasn’t quite what we wanted. So it takes some finessing to get the look you want.

AI is sometimes thought of as a quick tool. What’s your perspective on this?

There’s a misconception that AI is fast or easy, but pushing it to get specific results takes time and effort. For this project, I generated around 10,000 images to get the right ones. I often need to adjust prompts, try different descriptors, and test compositions. It’s like photography—I’m visualizing, troubleshooting, and experimenting, even though it’s all virtual. It took about 30 hours to get these final images, which is a serious investment in generating images that truly resonate.

We know you like to embrace AI, how do you see it as a tool rather than a threat?

I’ve always seen AI as a complementary tool. The skill isn’t just in using it—it’s in knowing how to guide it. I think of it like sketching for a client, visualizing and creating something from scratch. It’s exciting to see how the technology evolves, but I approach it with the same artistic mindset as my photography. It’s about bringing a vision to life, regardless of the medium. I still created these images, so AI didn’t take a job away from me, rather I used HBR came to me for my expertise using AI in order to get the exact images they desired.