Creative Exploration — Vibing with AI!

Vibing with the Multiverse Explorers! — Image generated via Runway ML

This article is part of the ‘Cultivating Creative Intelligence: An Article Series for the AI-Driven Future.’ In this series, we explore how AI is redefining creativity, design, and leadership, drawing insights from the Multiverse Machine project, the CTRL+ALT+EVOLVE movie, and responsible Agentic AI research.

Are you ready to dive into a world where creativity isn’t just about the finished product, but the fascinating journey of discovery? That’s the core of what I experienced building the Multiverse Machine, a project that became a deep dive into Creative Exploration — Vibing with AI!

What is the Multiverse Machine, and Why Did I Build It?

Imagine an app that lets you pull inspiration from infinite alternate universes — worlds filled with unique concepts, visual styles, and even their own vocabulary. That’s the Multiverse Machine. It’s a tool designed to spark creativity and challenge our pre-conceived notions of design and art.

Why build it? Like many in the age of AI, I was fascinated by the potential of artificial intelligence to augment human creativity. But I also had pressing research questions that went beyond just generating images: Who or what truly has the power to create in this new landscape? And critically, what defines “good” when the tools and possibilities are changing at lightning speed? While the Multiverse Machine’s initial audience was graphic designers creating posters, these fundamental questions — who creates and what defines good — resonate deeply with anyone building in today’s landscape, from artists to software developers.

The Journey Began with Uncommon Seeing

Inspired by concepts like DaVinci’s challenge to find pictures in coffee stains — a form of “uncommon seeing” — my initial exploration aimed to push AI’s creative boundaries. I first attempted to create a “Creative Image Detective” app using technologies like Tensorflow.JS and COCO-SSD via Bolt.new. The goal was to get AI to generate code that could “see” things both creatively and objectively. When operating in “creative” mode, the app could examine existing images and finding a snake in the texture of bear fur or perceiving “happiness” in the way a bush was lit. It was an attempt to produce directed hallucinations, a kind of amped-up AI creativity. In “objective” mode, the app stuck to reality, tasked with identifying known and visible objects, for instance locating a bear in a picture that contained a bear.

Turns out, getting AI to imagine things that aren’t explicitly there, based on subtle visual cues, is tough! While the results were sometimes interesting, consistently achieving that kind of interpretive creativity from AI proved challenging.

A Different Approach: Shaping the Experience to Spark Creativity

This led to a significant shift in my approach and a new app called the “Multiverse Machine”. Instead of focusing on getting AI to be creatively interpretive, I focused on using AI creatively to shape the user experience. The concept of “Uncommon Seeing” was still central, but the method evolved. The goal became setting the stage for people to react, to find their own sparks of inspiration from the diverse outputs. The challenge was also broader; the app needed to support the creation of posters, requiring proficiency in multiple modalities and literacies — images and words working together.

This meant building an app where designers would connect with ideas and images from vastly different “universes.” The AI’s role was to collaborate in tapping into that limitless data stream, to be the wellspring of unexpected concepts and visuals that could provoke, inspire, and challenge human creativity.

The AI Toolkit: Vibing with a Creative Collective at Speed

Bringing this multi-modal, universe-spanning vision to life required more than just one tool; it meant orchestrating a creative collective of AI. This wasn’t just about “vibing” with a single platform; it was about pulling together a set of tools that embody the spirit of vibe coding — a rapid, iterative approach that emerged in early 2025.

Bolt.new became essential for its capabilities in building and deploying the application structure with incredible speed — timelines simply not possible with traditional hand-coding. Gemini was my key collaborator for generating the core content of the universes, helping to craft unique vocabularies, descriptions, and initial concepts. Tools like ImageFX, Adobe Firefly, and Runway ML were critical for visualizing these abstract worlds and their inhabitants, providing the visual inspiration for explorers. Gemini helped with the visualization as well, supporting me in creating the prompts that would generate the visuals. And last, but not least, Gemini also help with analysis of the resulting images providing critiques of the images supporting the development of a visual literacy for aspiring designers. This combination of tools, used in concert, allowed me to rapidly translate my ideas into a working reality, demonstrating the power of this emerging practice.

Beyond the Code: Addressing Core Questions

For me, the technical build was only part of the story. The true heart of this project was exploring those fundamental research questions that continue to shape our creative landscape: Who or what truly has the power to create? And, in a world of accelerating AI capabilities, what defines “good”? The Multiverse Machine project became a hands-on investigation into these questions.

What the experience revealed was the power of the human working with AI entities as a creative team. This wasn’t about AI as a mere tool or co-pilot; it was about a partnership. And critically, I believe this team — comprised of myself collaborating with tools like Gemini and Bolt.new — was more creative together than we would have been working separately. The AI brought speed, unexpected outputs, tireless iteration, and diverse perspectives, while I brought intuition, curation, vision, and the ability to synthesize feedback into a coherent direction.

This contrasts sharply with the first app attempt, which perhaps put too much reliance on the resulting code or the app itself to perform the creative heavy lifting — getting AI to interpret images creatively proved a difficult task to automate. The Multiverse Machine, however, takes a different path. It brings much more forward — both in terms of the rich, provocative content from diverse universes and the interactive experience of exploring that content. Its success lies in how it empowers future users to be creative, providing them with sparks and tools for their own “uncommon seeing.”

This brings us to the question of “good.” Was the code built with Bolt.new “good”? Were the images and vocabularies generated by Gemini and other tools “good”? What about the image critiques that were generated? In the context of this exploratory project, the initial answer is that they were more than good enough. My primary goal was exploring an idea, rapidly prototyping a functional application that could help investigate my research questions and enable a specific kind of creative exploration for users.

The code built was robust enough to serve as a proof-of-concept and prototype, allowing for rapid iteration based on feedback. And yes, I believe much of the code, while built with speed using vibe coding, would end up serving well beyond just this initial phase. There’s a clear path forward and a “lift to go bigger” — I’d love to enable users to create new worlds on the fly, to truly expand the multiverse themselves.

But what has been built now offers significant value. It serves an audience well, especially those seeking to learn about images, engage in uncommon seeing, and approach creative work differently. They can explore vast possibilities, get critiques to deepen their understanding of visual language, and find inspiration beyond their usual sources. These are not minor achievements; enabling this kind of learning, exploration, and empowerment through a functional application built with a lean, collaborative AI approach? Those all sound like amazing wins in addressing what creation and “good” can mean in this changing landscape.

Going Beyond the Prompt

Within this project and the CTRL+ALT+EVOLVE movie, I also found cases where the ideas that emerged were different than what I initially thought I asked for. They weren’t necessarily hallucinations, as much as creative interpretations of my prompts. There’s a fun post-credits scene in the movie that emerged from images I collected while working on the main narrative. The images didn’t fit the main story, but they were interesting! And when I put them together in a sequence — they spoke of a whole new potential story for my characters! It shows how sometimes, especially when vibing with AI, we just stumble into the next creative idea by being open to the unexpected.

Creative Exploration — Vibing with AI! is about embracing this fluid, iterative approach. It’s about recognizing that with the right tools and mindset, coupled with the speed and unexpectedness of AI collaboration, we can amplify our creative potential and explore possibilities that were previously out of reach.

This journey building the Multiverse Machine has reinforced that Creative Exploration — Vibing with AI! is a powerful new paradigm. It’s about embracing rapid iteration, leveraging a creative collective of AI tools, and trusting the process of human-AI partnership to discover possibilities previously out of reach. This was just the first step in understanding how AI reshapes our creative world. In the next article of this series, ‘Cultivating Creative Intelligence,’ we’ll delve deeper into The Creative Multiplier: Scaling Impact Through True Human-AI Collaboration, exploring how this goes far beyond simply using a “co-pilot” and necessitates a rethink of how we structure teams, define “good,” and even challenge the rise of “shadow AI-T.” Follow me to continue exploring the future of creation and collaboration.

P.S. Want to experience the Multiverse Machine firsthand? I’ll add a link and a bit more background in the comments below. Please note, this is hosted on a free tier, so usage limits apply, and performance may vary depending on traffic. The Multiverse Machine is detailed in a more extensive visual essay exploring AI creativity, which I plan to make publicly available on my GitHub repo soon. Stay tuned — follow me for updates on the essay’s release and the rest of this ‘Cultivating Creative Intelligence’ series! I’m always happy to share insights or ensure you get a chance to explore the app smoothly — feel free to connect or send a message!