Paper accepted at AI4CHIEF 2026: Introducing the “Meta-Body” of Artworks

The paper explores how AI can support storytelling around artworks without stripping away cultural context.

We are pleased to share that our paper “The Meta-Body of Artworks: Sequence-Based AI Storytelling for Culturally Situated Heritage Narratives” has been accepted for presentation at AI4CHIEF 2026 – The Symposium on Artificial Intelligence for Cultural Heritage and Indigenous Futures, which will take place in Paris on April 16–17, 2026.

The paper is co-authored by Dr Abol Froushan, Cristine H. Legare, Marianne Magnin, and Marco Cappellini, bringing together contributors from RIZOM, The Center for Applied Cognitive Science (CACS) at the University of Texas at Austin, The Cornelius Arts Foundation, and ArtCentrica.

For readers interested in the practical implications, a short section later in the article outlines how this research can inform the design of AI systems for museums, archives, and other meaning-rich environments.

About the AI4CHIEF Symposium

AI4CHIEF gathers researchers exploring how artificial intelligence can support the preservation, interpretation and transmission of cultural heritage.

The symposium addresses topics such as:

• AI-supported digitisation and analysis of cultural artefacts

• knowledge representation and heritage data infrastructures

• immersive media for cultural engagement

• computational approaches to archives and historical materials

• governance and culturally appropriate access to heritage data

The conference is organised by Le CNAM and Penn State University, with proceedings published by Springer.

Our contribution: structured meaning around each artwork

Our research explores a question that is becoming increasingly important as museums, archives and cultural institutions adopt AI tools:

How can artificial intelligence help people engage with cultural heritage without stripping artworks of their historical and cultural context?

Many current AI systems treat cultural objects as isolated data points:

• A painting becomes an image file.

• A sculpture becomes a database entry.

  
  

AI models generate visual mash-ups, descriptions or stories from these fragments. Yet this is not how culture actually works.

Artworks gain meaning through relationships: relationships with other works, with historical moments, with curatorial interpretation, and with the communities that continue to engage with them.

Our paper introduces the idea of the meta-body of artworks to describe this broader cultural structure.

What is a "meta-body" of artworks?

When museums digitise collections, artworks are typically stored as individual objects in a database.

However in practice, artworks are rarely encountered in isolation.

• Curators place them in exhibitions. 

• Historians connect them across periods and movements. 

• Visitors experience them through stories, sequences and associations.

• Communities interpret them through memory, identity and lived experience. 

The meta-body of artworks refers to this wider living structure.

It is the network of relationships, narratives and contexts that emerges when artworks are encountered together.

It includes curatorial paths, cultural meanings, and the evolving interpretations that develop around works of art over time.

Instead of treating artworks as separate database entries, the meta-body perspective sees them as part of an evolving cultural ecosystem. In this approach, AI is not used to invent new meanings for artworks. It helps navigate and reveal the relationships that already exist between artworks, histories and communities.

Empirical validation

The framework was tested through a pilot study at the European University Institute, examining how artworks influence metaphorical and narrative responses within structured reflective loops.

Across 43 participants and five sessions with A/B conditions, results show that artworks modulate the carrying capacity of symbolic fields depending on abstraction levels and contextual grounding.

Findings indicate that:

• representational anchors stabilise symbolic traversal after abstraction overload

• double abstraction (abstract theme combined with abstract imagery) can produce measurable symbolic collapse

Read More About the EUI Pilot

Practical implications for the cultural sector and beyond

Beyond its conceptual contribution, the paper proposes a practical framework for designing AI systems that respect cultural context and relational meaning.

For the cultural sector, this approach offers concrete pathways to:

• organise digital collections as meaningful sequences of works, rather than isolated objects

• integrate curatorial knowledge and community perspectives into AI-supported narrative tools

• embed permissions, sensitivities and governance rules directly into metadata and digital infrastructures

• develop AI systems that help audiences explore relationships between artworks, rather than generating generic narratives.

For museums, archives and heritage institutions, this perspective can guide the design of digitisation pipelines, knowledge graphs and narrative interfaces that remain culturally grounded and historically informed.

More broadly, the research speaks to a wider challenge facing many domains today:

How can AI systems operate within environments rich in meaning, rather than reducing everything to datasets?

The meta-body concept suggests that AI can function as a navigation layer within complex symbolic systems, supporting interpretation, dialogue and collective sense-making.

While the paper focuses on artworks, the same approach may prove relevant for fields such as education, organisational knowledge systems and collaborative decision-making, where context, relationships and meaning are essential.