How does the interaction work and what happens to your drawing?

How the exhibition uses artificial intelligence (AI)

When you create a drawing in Edvard Munch: Connect the Lines, the system compares your drawing with more than 7,000 digitised drawings from MUNCH's collection.

Artificial intelligence (AI) helps the system identify connections between your drawings and drawing and Edvard Munch's drawings. AI models are computer systems trained to identify patterns and relationships in and between images and texts. Rather than understanding art in a human sense, they recognise similarities, recurring features and connections within large collections of data.

Before the exhibition opened, MUNCH's archive of drawings was analysed using several AI models.

• CLIP (Contrastive Language–Image Pretraining) is an AI model developed by OpenAI that analyses visual characteristics such as lines, shapes, composition and structure.
  
• GPT (Generative Pre-trained Transformer) is a language model developed by OpenAI. In this exhibition, GPT helps generate short descriptions of drawings and identify possible motifs, subjects and themes.
  
• Text-embedding models convert visual and textual information into searchable representations that allow relationships to be identified across thousands of drawings.

To make the archive searchable, both the visual analyses and the text descriptions are converted into digital representations that can be compared across thousands of drawings.

Together, these analyses create a digital map of visual and thematic relationships across MUNCH's drawing archive.
 
The AI does not generate new artworks and does not imitate Edvard Munch's style. Its purpose is to help visitors discover drawings and connections within MUNCH's collection that might otherwise remain hidden.

How the matching process works

When you make a drawing, the visual AI analyses how your drawing looks, identifying characteristics such as lines, shapes, composition and structure. At the same time, the language AI generates a short description of what the drawing appears to depict, such as people, objects, landscapes or other motifs.

These visual and textual analyses are then compared with the analyses already created for more the Edvard Munch drawings in the museums archive.

The system identifies the closest matches by combining two types of similarity: how closely drawings resemble each other visually, and how closely they relate to one another in terms of subject matter or theme.

It does not search for exact copies. Instead, it looks for drawings that may share similar forms, motifs, compositions, moods or other visual and thematic characteristics.

The Munch matches are displayed on your iPad, allowing you to explore unexpected connections with Edvard Munch's drawings. 

YOUR DRAWING

       ↓

VISUAL AI (CLIP)

How does it look?

Lines • Shapes • Composition

       ↓

LANGUAGE AI (GPT)

What might it depict?

Motifs • Subjects • Themes

       ↓

SEARCH SYSTEM

Compares with 7,000+

Munch drawings

       ↓

DISCOVERS CONNECTIONS

       ↓

DISPLAY MUNCH MATCHES

What happens to your drawing?

Your drawing will only be stored on MUNCH's own servers if you choose to share it on the screens in the exhibition.

The drawing itself is not permanently stored by external AI providers. MUNCH retains control over the visitor-generated data collected through the exhibition.

Your drawing may be displayed in the exhibition as part of the collective screen gallery of audiences drawings, alongside the Munch drawings it has been matched with.

Help create a new cultural dataset

When you submit your drawings and its matching Munch drawings to the screens in the exhibition space, you have the option to give your consent to your drawing being stored and shared as part of a data set for research and other non-commercial purposes.

If you choose to give consent, MUNCH may retain your drawing and information about how it was created after the exhibition has ended. This may include the finished drawing, the sequence of strokes used to create it, and information about the drawing process itself.

MUNCH is exploring the creation of a consent-based dataset that may be shared after the exhibition through initiatives such as Mozilla Data Collective and other public-interest data-sharing platforms.

Our intention behind the creation of such a dataset is to establish a unique record of drawings, that could support research into creativity, learning, drawing, human-computer interaction and cultural participation. It could also help artists, educators, researchers and cultural institutions develop alternative approaches to artificial intelligence that are based on transparency, consent and public benefit.

Most AI systems today rely on datasets that are commercially controlled and largely inaccessible to the public. As a museum, MUNCH is interested in exploring whether cultural institutions can help create more open and diverse knowledge resources that support research, education and innovation.

Providing consent is entirely optional and will not affect your experience of the exhibition. Any future sharing of data will be based on consent, transparency and responsible governance. 

Why is MUNCH using AI?

MUNCH cares for more than 27,000 artworks by Edvard Munch, including over 7,000 drawings. Many of these drawings are fragile and can only rarely be displayed.

Through Edvard Munch: Connect the Lines, MUNCH explores how technology can help more people discover, engage with and learn from this largely hidden part of the collection.

The role of AI in the exhibition is not to replace human interpretation or create new artworks. Instead, it helps visitors navigate a vast archive of drawings and discover unexpected connections that might otherwise remain unseen.

For MUNCH, projects such as Connect the Lines are also an opportunity to explore how museums can use artificial intelligence responsibly, transparently and in the public interest.