In a recent update to our LPs, we talked about the current macro market conditions and how we’re “playing it out” alongside founders at TheGP. We used Midjourney, a text-to-image AI program, to create the following image in our update:

The image not only fit the topic we were writing about, but it also made us think about the founders we were seeing “get up to bat” in spite of the rain. The image sparked our creativity and helped us better articulate what we were seeing in the market—that the most exceptional entrepreneurs were adjusting their “game” in order to win. The Midjourney software closed the skill gap between having good taste and creating interesting, high-quality visual work. It also inspired a new way to express our view of the market in writing.

In other words, software became our creative partner in the process. This is creative compute.

In the past, software tools have certainly helped people express themselves more effectively, captured creativity, and reduced the cost (in money, time, and prerequisite skills) to create art (via digital cameras, photoshop, Instagram, etc.), but the new wave of large, foundation models that foster image and text creation mark the first time that software has become a proactive partner in human creativity. These models are showing us the power of software that has evolved from a reactive tool to an active protagonist.

To spark creativity, you need a surprise. Dots need to be connected in new ways so that previously unseen relationships can come into focus. Working with GPT-3 or Midjourney to create text or images doesn’t feel like working with a new tool that simply makes you more productive. It feels more like working with a new colleague who helps you see the world a little differently, who shapes your thinking and extends your creativity.

When you work with someone new like that, you have to learn to “speak their language.” Their work influences and inspires your own approach as you get a feel for how they understand the world. I first had this experience working with sneaker designers, where I had to learn to communicate visually through mood boards rather than writing a memo describing a “fast-looking shoe.” Learning that skill unlocked my creativity and allowed me to participate in the design process. When I figured out how to evolve my communication to fit the interactive style of the design team, I got exponentially better at turning concepts into sneakers and much more creative because of my collaboration with my colleagues.

I feel the same excitement I felt then when I use Midjourney—I love figuring out the prompts that lead to the images I like most. And I can feel the potential of these foundation models in their ability to change the way we create, research, and work. We see hints of this today in projects like Stable Diffusion becoming more approachable and effective through the curation available from folks like RiversHaveDreams, KREA and PromptBase allowing users to access prompts that have already been designed to drive specific results. It feels like we're on the verge of unlocking creative computation for many more users and a much broader set of industries. Better UI, new workflows, AI-first tools, verticalized applications, and the infrastructure stack required to support them are still needed to bridge the gap between foundation models and commercial value creation.

AI can be horizontal by design. It can help us associate disparate concepts and provide inspiration with new connections. Rather than speed us up, it can create value by slowing us down to show us things we didn’t imagine—this is a new human/computer interaction paradigm. You see this in simple sentence completion tools, and you see it in the way Midjourney, Dall-E and Stable Diffusion construct images. There is imagination required to understand or interpret new images and evolve instructions. The effort to decode the prompt or hone the result requires and inspires creativity.

When AI surprises us, we have to use our imagination to take us from where we are to what the system created. This inspiration is the energy needed to drive critical insight and breakout ideas. The real value of AI lies in its ability to expand human creativity.

I’m excited to see how far we can go.