SeeM
SeeMuseum
AI-assisted online museum for guided art exploration
Designing a non-intrusive AI curator that helps users move from discovery to deeper understanding of artwork
UIUX Designer Intern · AI Exhibition Experience · SeeM · Live Product
05/2024 - 09/2024
The Core Challenge:
Guidance Without Interruption
Merging the gap between art enthusiasts and elite museums through constraint-aware AI experiences.
Insights from early research synthesis were used to inform navigation structure and feature prioritization.
Online museum experiences often fall into two extremes. They either show too much information at once, causing cognitive overload, or rely on AI chat interactions that interrupt the visual experience.
For SeeM, the core challenge was:
How can AI guide users through unfamiliar artworks without taking over the act of looking?
Research synthesis helped identify three design constraints:
-
Users enter with different levels of art knowledge
-
AI-generated content can easily overwhelm the viewing flow
-
Exploration needs structure without feeling controlled
Core Design Principle: Progressive Exploration
Reveal information gradually based on user intent and viewing depth.
Instead of showing artwork details, AI explanations, and recommendations all at once, SeeM uses a progressive exploration model.
The experience starts with orientation, then gradually introduces context, detail, and AI guidance as users move closer to the artwork.
AI supports the viewing flow — it does not lead it.
Personalized Discovery
Reduce Early Choice Overload
- Limiting initial options
Reduce early choice overload with a small, relevant starting set.
AI-assisted Curation
Position AI as a Curator, not a Chatbot
Use AI to sequence and contextualize artworks, not to interrupt users with constant prompts.
Structured Browsing
 |  |
|---|---|
 |  |
Structured Browsing
Support multiple entry points while keeping information density controlled.
To balance immersion and understanding, I designed a progressive zoom-based exploration model
Contextual AI Guidance
Core Interaction Model: From Overview to Detail
A zoom-based exploration model that balances immersion and understanding.
To avoid overwhelming users, I designed a four-stage viewing model.
Each stage reveals a different level of information based on how closely users engage with the artwork.
Stage 1: Distant Shot
Overview
Builds spatial orientation before users focus on individual artworks.
Stage 2: Mid Shot
Context
Introduces essential artwork information without crowding the canvas.
Stage 3: Close Shot
Inspection
Supports closer visual exploration while preserving exhibition context.
Stage 4: Super Close Shot
AI Guidance
AI appears only when users show deeper intent through pause, zoom, or focused viewing.
AI guidance is triggered by user intent, not system eagerness.
AI as Supporting Layer
-
AI Guidance appears only when users pause or zoom
-
Questions are suggested based on the viewed region
-
AI does not interrupt the primary exploration flow
This approach keeps visual immersion intact while still supporting learning and curiosity.
It prevents AI from interrupting the core experience, keeping immersion as the primary focus.
Designing for Different Exploration Intent
Users browse, collect, and revisit for different learning goals.
Exploration does not end when users leave an exhibition.
SeeM separates saved exhibitions from saved artworks because they support different behaviors:
-
Exhibitions help users revisit a curated learning journey
-
Artworks help users collect individual pieces of interest
-
Revisits support long-term learning instead of one-time browsing
The retention loop is based on learning intent, not generic saving.
Separate flows were designed for saving exhibitions and saving individual artworks, allowing users to revisit content based on learning goals rather than treating all saved items the same.
SeeMuseums
See Your Own AI Generated Online Museum
SeeM is not a concept about AI features, but an exploration of how interaction design can guide attention, pacing, and understanding in complex content environments.
This project reflects my approach to designing AI-assisted experiences with clarity, restraint, and user intent in mind.
