OVERVIEW

Turning Passive Gallery-Goers into Active Explorers

ArtSense addresses the challenge of enhancing the museum visitor experience by moving beyond passive observation toward active engagement. I noticed a gap in existing interpretive tools, where traditional wall labels and audio guides often fail to foster deep cognitive retention or proactive curiosity.

ArtSense mobile app shown in a museum gallery

Objectives

Enhance engagement through AI-powered audio guides, chat, and gamified learning

Optimize interpretive tools using real visitor behavior data

Reduce cognitive fatigue with dynamic, AI-assisted content delivery

CHALLENGE

Where Museums Fall Short for Today's Visitors

As I researched and prototyped ArtSense, three core challenges emerged from observing how visitors actually move through galleries, how curators feel about AI in museums, and where existing interpretive tools fall short.

Information depth gaps
Challenge:
Visitors often have lingering questions that standard guides and labels fail to answer, forcing them to rely on external search engines.
Solution: Through the museum's archives, I designed a system where visitors can ask art-related questions through the AI curator within the mobile application — directly pulling information from the museum instead of relying on Google.

Pacing behaviors in museums
Challenge:
There's a mismatch between the slow pace of audio guides and the "glance-based" behavior of visitors who want to move quickly between artworks.
Solution: Visitors can use their camera to capture artworks and save them to a private personal collection in the mobile app — letting them learn more about works they want to explore in greater depth, both during and after their visit.

Barriers to meaning-making
Challenge:
Without scaffolding, visitors often struggle to synthesize complex information, leaving them unable to form the deep, personal connections required for long-term cognitive retention.
Solution: A gamified kiosk placed in resting areas provides cognitive scaffolding for deeper engagement. The kiosk challenges visitors to test their knowledge of art movements through an interactive multiple-choice game and a personalized AI photo booth that generates portraits in various art styles.

How might we move museum visitors beyond passive observation toward active engagement — bridging the gap between in-the-moment discovery and long-term cognitive retention?
RESEARCH

Mapping How People Actually Experience Museums

This research focused on optimizing museum interpretive tools by integrating AI-powered dialogue and gamified interactions to support active meaning-making. I structured the research around three key goals:

Investigate and analyze the visitor journey throughout the museum

Conduct a competitive analysis of digital apps used in New York museums

Interview a creative technologist at The Met on AI's role in museum experiences

Mapping the full visitor journey helped me identify the moments where existing interpretive tools either supported or interrupted natural museum behavior.

ArtSense visitor journey map

A competitive analysis of digital tools across major New York museums (The Met, MoMA, the Guggenheim) surfaced common patterns — and gaps where AI-driven, snackable interpretation could add real value.

Competitive analysis of New York museum digital apps — part 1 Competitive analysis of New York museum digital apps — part 2 Competitive analysis of New York museum digital apps — part 3

Across primary research, three behaviors stood out as the strongest signal for how ArtSense should be shaped:

01

Glance-based behavior: Visitors primarily scan for visual "hooks" before committing to deep dives.

02

The "reflection" zone: Learning doesn't only happen in front of the art — it happens in cafés and resting areas after the visit.

03

AI usage: Curators are wary about AI, but understand that visitors are already using ChatGPT and Google to learn more about artworks.

INSIGHTS

What Visitors Need to Feel Truly Engaged

Insight 1
High cognitive load in galleries prevents deep synthesis and art appreciation. Visitors usually process their experience after large exhibitions and in cafés.

Hours of dense wall labels and rapid context-switching between artworks leave visitors with fragmented impressions rather than synthesized understanding. The most receptive moments for reflection happen after the gallery — in transitional zones like resting areas, cafés, and exits.

Recommendation 1
ArtSense Kiosk located in resting areas and exits serves as a cognitive bridge when visitors are most reflective. This gamified kiosk serves as an interactive tool to test their knowledge and their ability to notice details across artworks.
ArtSense kiosk placed in a museum resting area, showing the puzzle game

The opportunity to learn and reflect on past artworks through a puzzle allows visitors to evaluate and analyze details of the art movements presented. Through the visitor's portrait, they can reflect on their museum experience, using the rendered image as a scaffold for learning through the puzzle.

ArtSense kiosk tablet close-up showing the puzzle quiz interface Twelve art-style portraits generated from a visitor's photo at the kiosk

Art movements and visitor portraits, generated in 12 styles at the kiosk's photo booth.

Insight 2
Visitors abandon digital tools immediately if IDs don't match or content lasts longer than 180 seconds.
ArtSense app showing artwork not found error when an invalid ID is entered
Visitors already use their phones to archive artworks they've seen — they just don't want to type a four-digit code to do it.
ArtSense Search Art screen — enter a numeric ID to look up an artwork

Standard museum apps rely on numeric ID codes pulled from wall labels, which break the glance-based pacing visitors prefer. And audio guides that run longer than three minutes are almost universally skipped.

Recommendation 2
Deliver "snackable" content (90–180s) and let visitors capture artworks with their camera to avoid ID-code friction.
ArtSense capture and collect feature — search only vs capture and search experiences ArtSense Collection page — capture and save artworks via the mobile camera

Capture-and-collect mirrors how visitors already use photos to remember works they want to revisit. Tapping a saved piece opens a short, structured artwork story — short enough to fit a visitor's natural pace, long enough to deepen their understanding.

Insight 3
Curators value interactive engagement but fear the "AI" label due to concerns over factual accuracy and authoritative voice.
AI can be a powerful tool, but curators don't trust the information provided through ChatGPT or Google search.

An expert interview with a creative technologist at The Met made it clear: the institutional authority of the museum is its most valuable asset. Generic LLM output undermines that authority, even when factually correct.

Recommendation 3
Frame the AI as unlocking existing museum archives — not generating independent narratives from unknown data.
ArtSense in-app AI chat, sourced from museum archives and collections

AI-powered descriptions and chat responses are gathered directly from the museum's archives and collections. This preserves the museum's authoritative voice while still giving visitors the conversational, on-demand interpretation they were already seeking elsewhere.

OUTCOME

Built for Curiosity, Validated by Users

Usability testing for ArtSense was conducted with four individuals of varying museum experience, using Notion for transcription and NotebookLM for qualitative analysis. The outcomes revealed strong validation for the core features alongside clear areas for technical improvement.

What worked well

  • Gamified learning: The interactive puzzles and quizzes received positive feedback for making art history engaging and approachable for visual learners.
  • Personalized "capture and collect": The ability to bookmark artworks into a timestamped, visually cohesive Collection was valued as an improved alternative to losing photos in a general camera roll.
  • Minimalist design: Users responded positively to the clean UI and the delivery of short, digestible art descriptions.
  • Social potential: A strong desire emerged to share the personalized stylized portraits and captured collections — highlighting the social utility of the ecosystem.
I love that I can revisit a painting I only glanced at — it's like the museum followed me home.

Areas for improvement

  • AI latency: High rendering times for image generation in both the kiosk and mobile app caused user uncertainty, with many participants mistaking the wait for a bug.
  • Cross-platform friction: Instructions on the kiosk weren't clear enough for users to immediately realize they could use the four-digit artwork IDs in the mobile app for deeper exploration.
  • Kiosk hierarchy: Some participants struggled with the visual layout of the kiosk, finding it difficult to determine how to interact with specific artist and artwork information.
LEARNINGS

Designing With Purpose, Learning as I Went

ArtSense was the most ambitious end-to-end project I've taken on — researching, designing, and building two connected products from scratch over the course of my NYU thesis. Two themes shaped the experience.

Reducing cognitive load to improve art appreciation

Through ArtSense's kiosk and mobile application, the primary goal is to reduce cognitive load so visitors can continue learning without feeling overwhelmed during their museum experience. The capture-and-collect feature received high praise from visitors who want to learn about artworks that interest them without having to stand in front of them for long periods.

Utilizing AI tools to build ArtSense

I used Claude to generate descriptions and identify artworks within each art movement to populate museum data across the kiosk and mobile app. I worked with image generation APIs like Nano Banana to clean up images in the capture-and-collect feature, and used Runway ML to generate portraits of visitors in specific art styles. This experience taught me how to pick the right LLM and image-generation models for specific visitor moments — while building up my own implementation skills along the way.