Making AI Look Good: The Psychology of Escaping Generic UI

About the Talk

Shani Barzilay challenged the idea that visual design is a matter of taste, arguing instead that it is applied psychology. In an era where 80% of design work has become a commodity, she showed how to build interfaces that stand out by grounding every visual decision in cognitive science.

Key topics covered:

• ●"Taste" is applied psychology: design choices create messages, messages trigger emotions, and emotions drive decisions. It is not subjective.
• ●Color as emotion: using Plutchik's wheel of emotions, color temperature, saturation, gradients, and textures to convey specific feelings (as seen in Arc Browser, Wise, and AI brands).
• ●Shapes and cognition: the Kiki-Bouba effect shows how rounded shapes feel safe and angular shapes feel sharp. Applied to interfaces: when should your UI feel reliable (rounded) vs attention-grabbing (angular)?
• ●Typography carries personality: serif fonts feel polished, sans-serif feels friendly. The choice communicates before anyone reads a word.
• ●A new design workflow for AI: the old linear flow (UX, UI, Dev) is dead. The new workflow starts with developing a design language, building the business case for design, then creating an AI-ready design system using atomic design and design tokens.
• ●Case study with CapitalSnapshot: translating brand values (clarity, actionability, empowerment, trust) into concrete visual strategies for a fintech product.
• ●Design systems for AI: why vibe-coders fail. Atomic design, design tokens, component states, and Storybook handover are essential for production quality.