# **Artificial Intelligence in Museums: Applications, Challenges, and Ethical Considerations**
## **1. Introduction**
The presentation explores how artificial intelligence (AI) is transforming museums through three key applications:
1. **AI for audiences** – Enhancing visitor experiences.
2. **AI for museum professionals** – Improving efficiency in curation, conservation, and research.
3. **Museums as forums for AI** – Using exhibitions to foster public debate on AI’s societal impact.
---
## **2. AI for Visitor Experiences**
### **2.1 Personalized and Interactive Exhibitions**
AI enables museums to create dynamic, immersive experiences by leveraging vast collections that would otherwise remain inaccessible.
#### **Examples:**
- **Phototechia (Musée de la Photographie, Charleroi, Belgium)**
- The museum holds **100,000+ photographs and 1.5 million negatives**, most of which are never displayed.
- AI analyzes images based on **objects, colors, and emotions**, allowing visitors to explore the collection interactively.
- Visitors can **curate their own "galaxy" of images**, engaging with photographers and shaping their own experience.
- **Ask Dali & Dali Lives (The Dalí Museum, Florida, USA)**
- **Ask Dali**: An AI-powered audio experience where visitors interact with a **deepfake Salvador Dalí** via a lobster phone.
- Example questions: *"Why are the clocks melting?"* → *"Think of them as a vast dream."*
- **Dali Lives**: A **deepfake screen installation** where Dalí engages with visitors, takes selfies, and sends them to their phones.
- **Marketing vs. Education**:
- **Pros**: Highly engaging, innovative, and attracts visitors.
- **Cons**: Raises **ethical concerns**—visitors may mistake AI-generated responses for authentic insights.
- **Question**: Is this **education or entertainment**? Does it provide meaningful learning, or is it just a gimmick?
- **Fondation Rubiton’s Chatbot (Paris, France)**
- Visitors scan a **QR code** to ask a chatbot questions about artworks.
- **Risk**: **Reduced human interaction**, potentially undermining museums’ role as spaces for **debate and dialogue**.
---
## **3. AI for Museum Professionals**
### **3.1 Streamlining Curation and Research**
AI automates time-consuming tasks, improving efficiency in **cataloging, conservation, and provenance research**.
#### **Key Applications:**
- **Automated Metadata Generation**
- AI generates **descriptive tags, titles, and catalog entries**, reducing manual labor.
- **Document Transcription**
- Museums hold **handwritten archives** (letters, records). AI transcribes them, making them searchable.
- **Cross-Referencing Collections**
- AI links related records, improving **cataloging accuracy**.
- **Fighting Illicit Trafficking**
- **ANCHIS (EU-funded project)**: AI scans objects to detect **stolen or looted artifacts**.
- **Provenance Research**: AI reconstructs **ownership histories**, identifying looted works.
- **Predictive Conservation**
- AI models analyze **humidity, temperature, and other variables** to predict deterioration in artworks.
---
## **4. Museums as Forums for AI**
### **4.1 Exhibitions as Critical Discourse on AI**
Museums are increasingly using AI not just as a tool but as a **subject of exhibitions** to foster public debate.
#### **Example:**
- **Le Monde selon Lya (Musée Marie-Jo de Pomme, Paris, 2023)**
- Explored **analytic vs. generative AI**, encouraging visitors to **critically engage** with AI’s societal impact.
---
## **5. Ethical and Practical Considerations**
### **5.1 Key Principles for Responsible AI in Museums**
The speaker proposes three guiding principles:
1. **Human-Centered AI**
- AI should **enhance** (not replace) human interaction.
- Example: Chatbots should **complement** (not eliminate) museum guides.
2. **Ethical AI**
- Must respect:
- **Copyright** (e.g., using deceased artists’ voices/likenesses).
- **Cultural heritage** (avoiding misrepresentation).
- **Transparency** (disclosing AI’s role).
- **Privacy** (protecting visitor data).
3. **Responsible AI**
- Should align with museums’ **educational and public missions**.
- Must be **accountable** for its impact (e.g., avoiding misinformation).
### **5.2 Legal and Ethical Challenges**
- **Case Study: Ask Dali**
- **Legal**: Requires **copyright clearance** (e.g., Dalí’s estate permissions).
- **Ethical**: **No consent** from Dalí (deceased), raising questions about **authenticity and exploitation**.
- **Question**: Should museums use AI to **resurrect historical figures** without clear ethical guidelines?
- **Educational Value vs. Marketing**
- **Criticism**: AI experiences (e.g., Dali Lives) may prioritize **engagement over education**.
- **Defense**: Even if superficial, they **attract new audiences** who may later engage more deeply.
---
## **6. Q&A Highlights**
### **6.1 Legal Aspects of AI-Generated Historical Figures**
- **Question**: How do museums handle **copyright and consent** when using AI to recreate deceased figures (e.g., Dalí’s voice)?
- **Answer**:
- Museums work with **legal teams** to secure permissions (e.g., paying estates).
- **Ethical dilemma**: No direct consent from the deceased, raising concerns about **exploitation**.
### **6.2 Educational Value vs. Marketing Gimmicks**
- **Question**: Are AI experiences (e.g., Ask Dali) **truly educational**, or just **marketing stunts**?
- **Answer**:
- **Pros**: Attracts visitors who might not otherwise engage with museums.
- **Cons**: May lack **depth**, requiring **traditional mediation** for meaningful learning.
- **Conclusion**: A **balance** is needed—AI can **hook** audiences, but museums must ensure **substance**.
---
## **7. Conclusion**
AI offers **transformative potential** for museums, from **personalized visitor experiences** to **streamlined curation**. However, its adoption must be:
✅ **Human-centered** (enhancing, not replacing, human interaction).
✅ **Ethical** (respecting copyright, transparency, and cultural heritage).
✅ **Responsible** (aligning with museums’ educational missions).
Museums must **critically assess** AI’s role—balancing **innovation with integrity** to ensure technology serves **both audiences and art** effectively.
---
**Final Note**: The presentation was part of a **multilingual event** (English/French), with a last-minute cancellation of a French talk by **Frank**, a prominent librarian.