Mask as Logic

My exploration began during a museum visit, when I stood before an exhibition of traditional masks from cultures around the world. What struck me was not only their visual richness, but the depth of intention embedded in each one. A mask is not assembled quickly — it is built over time through layers of material, symbolic meaning, and shared cultural agreement about what a particular form communicates to a particular audience. Recognizable, singular identity, the kind that carries weight and travels across years, does not emerge from a quick decision. It is built slowly through deliberate choices about what to show, what to withhold, and why.

This understanding set the terms for everything that followed, because the contrast with how identity is built today could not be sharper. AI image generation tools and template platforms have made visual production accessible to anyone — a person with no design training can now generate a logo, a color system, and a suite of brand assets in minutes. But access to visual production is not the same as access to identity. The outputs these tools generate are trained on statistically common design patterns. They are very good at producing things that look like brands, but they are not equipped to help someone discover what their brand actually is. The visual surface arrives before the underlying identity has been established, and the result is a landscape of identities that feel polished but indistinguishable.

The scale of this problem is significant. In the United States, roughly 33.2 million small businesses account for more than ninety-nine percent of all employer firms, and around 35 to 40 percent of these operate in sectors where brand identity is not optional but existential — including restaurants, retail, e-commerce, hospitality, beauty, tech startups, and direct-to-consumer businesses. In these spaces, the brand is often the first product. Another 30 to 35 percent fall into professional services, where a cohesive identity consistently separates those who grow from those who plateau. Across these segments, hundreds of thousands of businesses require clear and differentiated identity systems, yet most cannot access them because professional brand strategy is priced for organizations larger than they are.

The research that shaped this work falls into two connected areas. The first examines how masks function as communication systems — not decorative objects but precise instruments for projecting identity to an audience that shares the same cultural knowledge. The second addresses the gap between how humans describe identity and how machines interpret it. Humans understand identity through story, value, and emotion, while machines process it through pattern, frequency, and relationships between variables. When a person tries to communicate their brand vision to an AI tool, something is almost always lost — not because the tool is limited, but because no translation layer exists between human intention and machine instruction.

This led to a key realization. The problem is not that AI produces poor design. The problem is that the discovery process that traditionally shaped design decisions is missing. Many AI tools generate outputs quickly, but they do not guide users through the process of understanding what their identity actually is. From this, the central proposition emerged: the challenge in the generative AI era is not producing visuals, but translating human identity into a form that machines can interpret without losing meaning. Identity discovery must happen before generation, not after. In this context the designer's role as a translator between meaning and form does not disappear — it becomes more important and at the same time more invisible.

In response, I developed Maskit, a system designed to capture signals of identity and translate them into a structured framework that machines can use to generate coherent visual systems. The tool operates through two phases of structured questioning — visual and psychological — establishing parameters that can carry across multiple creative applications including logo systems, website content, visual assets, and communication materials. What makes it more than a questionnaire is the way each part is structurally connected: every question adds a layer, every answer refines the previous one. The final output is not a fixed visual style, but a system of decisions about what to reveal, what to hold back, and how to maintain consistency — encoded clearly enough that a machine can extend it without losing the meaning defined by the user.