Adaptive Museum - AI-Mediated Wayfinding

A museum is not only a collection of objects. It is a navigation problem.

To visit a museum is to move through overlapping systems: architecture, exhibition narrative, crowd behavior, labels, lighting, maps, personal curiosity, fatigue, memory, and time. A visitor rarely experiences the museum in the linear order imagined by the curator. They drift, hesitate, skip rooms, return to objects, follow other visitors, lose orientation, or become absorbed in unexpected details.

Adaptive Museum proposes a research project around this condition. It asks whether AI-mediated guidance can support wayfinding and interpretation without removing the visitor's autonomy. The goal is not to make the museum behave like a shopping app. The goal is to understand how adaptive guidance changes movement, attention, cognitive load, and memory inside a spatial narrative.

Can AI guidance help without over-directing the visitor?

The main research question is: compared with a fixed museum guide, can an adaptive guide reduce disorientation and cognitive load while preserving the visitor's sense of exploration?

This question contains a tension. If guidance is too weak, visitors may miss important content or waste energy finding their way. If guidance is too strong, the museum becomes a route to be completed rather than a place to be discovered. A good adaptive system must therefore balance clarity and openness.

The project would compare two conditions. The first is a traditional guide: fixed map, fixed route, fixed labels, and the same information sequence for everyone. The second is an adaptive guide that changes route suggestions and information density according to location, dwell time, selected interest, task progress, and possible signs of disorientation.

The system should be spatial, not just mobile.

The first prototype can be built as a web or Figma-based experience connected to a virtual museum plan. It does not need a real museum partnership at the beginning. A non-NDA virtual gallery allows the study to control layout, exhibits, route complexity, and information density.

The prototype should include three layers. The phone layer provides route suggestions, object information, and task prompts. The spatial layer includes signs, landmarks, room names, and visual cues in the plan or 3D scene. The ambient layer can be speculative in the first version: lighting, sound, or directional cues that might later become physical feedback.

Importantly, the adaptive behavior should be transparent. The system should not pretend to know the visitor's mind. It can respond to observable signals: where the visitor is, how long they stay, which theme they selected, whether they backtrack, and whether they fail to reach a target after a reasonable time.

The comparison should measure movement and memory.

A pilot study could recruit 12 to 20 participants and ask them to complete museum tasks under both conditions: fixed guidance and adaptive guidance. The tasks should include finding an object, following a thematic route, answering a memory question, and choosing a self-directed next stop.

The study can record task completion time, wrong turns, backtracking, route completion, skipped rooms, time spent at exhibits, and information recall. After each condition, participants can complete short usability and cognitive load measures, then answer interview questions about confidence, control, trust, and whether the guidance felt helpful or intrusive.

The strongest result would not simply be "adaptive is faster." In a museum, faster is not always better. A richer finding might show that adaptive guidance reduces anxiety at decision points while preserving optional exploration. Another possible finding is that adaptive systems help some visitors but over-structure the experience for others. Both outcomes would be valuable.

This project connects architecture, HCI, and design research.

Adaptive Museum is a strong master's application direction because it is specific enough to test, but broad enough to grow. It connects museum and exhibition experience with wayfinding, information design, user research, AI adaptation, and spatial interaction. It also allows a clear continuity from Spatial Critic.

Spatial Critic studies how spatial conditions affect navigation behavior in a controlled prototype. Adaptive Museum applies that logic to a more culturally meaningful environment. The question shifts from "how do people move through spatial ambiguity?" to "how can intelligent guidance support movement, attention, and memory inside a public cultural space?"

This makes the project suitable for HCI, Human-Building Interaction, interaction design, design computing, and research-based master's programs. It shows that the applicant is not only producing screens or renderings, but building a research question, prototype, study method, and design implication.

The project must avoid false intelligence.

The danger of an adaptive museum system is that it may present guesses as knowledge. A visitor who pauses might be interested, tired, confused, or simply waiting for someone. A visitor who skips a room might be bored, overloaded, or following a personal route. The system should therefore use cautious language and reversible suggestions.

Privacy is another concern. The first version should avoid facial recognition or personal identity. It should work with anonymous movement, selected preferences, and session-level behavior. This keeps the project aligned with ethical research practice and makes it easier to conduct a pilot study.

1. Define the virtual museum layout and route tasks.
2. Create a fixed-guide prototype and an adaptive-guide prototype.
3. Add a study protocol with measures, participant flow, and interview questions.
4. Draw a system diagram showing visitor signal, adaptation rule, interface output, and research measure.