Safe AI Learning Station

Why museums need this

Published survey figures, shown as reported. They describe the sector; they are not results of this project.

• UNESCO–ICOM Global Survey on AI in Museums, 2026 · 400+ museums in 90 countries
• Ithaka S+R Art Museum Director Survey 2025 · 368 directors, mostly US art museums (about 2% Canadian)

Core solution

A compact, disability-led holographic exhibit and workshop system for museums, libraries, schools, tourism sites, and community centres. It combines:

• A holographic display or projection station
• A small set of clearly fictional, locally co-designed AI characters
• Safe AI learning activities
• Accessibility-first interaction
• Human-led facilitation
• Local cultural and historical content
• A simple evaluation and consent process

Character system

Fictional guides, not impersonations of real people, unless explicit permission, consent and cultural authority are in place.

Key rule: every character says, in plain language, that it is an AI character, not a real person, and that visitors should verify important information.

Safe AI learning activities

1. 01

   Ask the Character

   Visitors ask a safe, pre-approved question about a local exhibit, artifact, or historical theme. The character gives a short answer, shows a source or uncertainty label, and invites the visitor to check the museum label or ask staff.

   “Sparky, what might this object have been used for?”

   Learning goal: AI can help explore, but humans and sources remain important.

2. 02

   Spot the Mistake

   The character gives a short local-history explanation containing one deliberate, clearly marked error. Visitors find it using exhibit labels, QR codes, or staff support.

   Learning goal: AI can be wrong; verification is a skill.

3. 03

   Create With AI

   Visitors create a short story, poem, exhibit title, or visual concept inspired by an artifact. The result is labelled AI-assisted, consent is asked before saving or displaying it, and visitors can delete their input.

   Learning goal: AI is a creative tool, not an autonomous authority.

4. 04

   Privacy Mirror

   Visitors learn what they should and should not share with an AI system: personal information, location, passwords, images of other people, and sensitive family or health details.

   “What should I never tell an AI character?”

   Learning goal: Privacy is part of AI literacy.

5. 05

   Accessibility Challenge

   Visitors test whether an exhibit explanation works for different needs: large text, captions, audio description, plain language, and screen-reader-friendly QR content.

   Learning goal: Accessible design benefits everyone.

Museum problems addressed

Safe AI rules

Every station displays and enforces these rules.

1. 1AI characters are fictional. They are not real people and do not have feelings, consciousness, or authority.
2. 2Visitors stay in control. They can stop, skip, repeat, or exit at any time.
3. 3No personal data required. Visitors can participate without creating an account.
4. 4Consent before saving. No visitor creation, image, voice, or story is stored or displayed without clear permission.
5. 5No impersonation. Real historical figures require documented permission, cultural authority, and clear labelling.
6. 6Sources over certainty. Important historical or scientific claims link to museum labels, archives, or staff-reviewed sources.
7. 7Human escalation. A staff member or trained volunteer is available for questions, concerns, or accessibility support.
8. 8Accessibility by default. Captions, plain language, large text, audio options, quiet-mode settings, and physical access are built in.
9. 9No medical, legal, or financial advice. Characters redirect sensitive questions to appropriate human services.
10. 10Transparent AI label. Every AI-generated response, image, or story is visibly labelled.

Fredericton 90-day pilot

Pilot name: Fredericton Safe AI Learning Station. No host has been confirmed yet.

Pilot roles

Revenue model HYPOTHESIS

Pricing hypotheses to test during the pilot. No revenue or attendance is promised.

Evaluation metrics

Only what the pilot can realistically measure. Targets are examples, not results.

Sector scenario 2026–2036 SYNTHETIC

An illustrative comparison of a sector-wide safe-AI capacity effort against no plan. Only the 2026 values are reported data; every later value is an assumption, not a forecast.

A single Fredericton pilot cannot move these sector-wide figures on its own. The “with plan” line shows a target direction that many organisations would have to share. It should be read as a goal to test, not an expected result.

Copy-paste text for partners

FrederictonHolograms Safe AI Learning Station

FrederictonHolograms proposes a small, accessible holographic AI learning station for museums, libraries, schools, and community spaces.

The station uses friendly fictional AI characters to help visitors explore local history, create with AI, learn about privacy, and practise checking AI-generated information. It is designed to support staff, not replace them.

The system includes safe-learning activities, accessibility features, consent controls, source labels, and a simple evaluation process. Characters are clearly fictional and do not impersonate real people without permission and appropriate cultural authority.

We are seeking one local pilot partner to test the concept over 90 days. The pilot would measure visitor engagement, accessibility, staff usefulness, content accuracy, privacy compliance, and operational feasibility.

This is not a claim that holograms or AI will solve museum funding or staffing challenges. It is a practical, community-centred pilot to test whether accessible AI learning can support museums, tourism, education, and creative participation in Fredericton.

## Safe AI Learning Characters

FrederictonHolograms is developing accessible holographic AI learning experiences for museums, libraries, schools, and community spaces.

Our approach uses clearly fictional characters to help people:

- Explore local history and culture
- Learn how AI works and where it can fail
- Create stories, images, and ideas with AI
- Protect privacy and personal information
- Practise checking information before trusting it
- Participate in accessible creative learning

### Our safety approach

AI characters are clearly labelled as fictional. They do not claim to be human, conscious, or an authority. Visitors can stop, skip, repeat, or exit at any time.

We do not require personal accounts. We do not save visitor creations, images, voice recordings, or stories without clear consent.

Important historical and scientific information is linked to sources, museum labels, archives, or staff-reviewed materials. Human staff remain available for questions, concerns, and accessibility support.

### Accessibility

The experience is designed for diverse cognitive and physical needs. It includes plain language, captions, large-text options, audio alternatives, quiet-mode settings, and clear ways to ask for help.

### Pilot invitation

We are seeking one Fredericton museum, library, school, tourism site, or community organization to host a 90-day Safe AI Learning Station pilot.

The pilot will test visitor engagement, accessibility, staff usefulness, content accuracy, privacy compliance, and operational feasibility.