Art & Ideas Archive

KADIST and e-flux have joined forces to create a unique digital platform that connects their archives and collections through the power of artificial intelligence. At the core of this collaboration is a tool that cross-references texts from the e-flux Journal with works from the KADIST collection, offering users a dynamic and intuitive way to navigate contemporary art and critical discourse. Whether you're a curator, student, or curious visitor, the platform reveals unexpected connections between ideas and images, theory and practice.

How does the platform work?
The platform pairs approximately 1,200 e-flux Journal essays with corresponding 2,500 artworks from the KADIST collection, enabling thematic and conceptual connections between texts and visual art. On this platform, artificial intelligence is used as a tool for semantic linking: the system analyzes existing articles and artworks descriptions but does not generate new artworks or new texts. Instead, it reveals relationships that already exist but that may not be immediately visible when consulting e-flux Journal and KADIST separately.

The interface is structured around two main entry points:
Topics: AI-generated thematic prompts suggesting conceptual entryways into the platform. Search: Users can enter any keyword into the search field, and the platform surfaces related artworks from the KADIST collection alongside relevant e-flux Journal essays.

From there, users can follow conceptual threads, move between texts and images, and continue exploring through linked ideas. Key ideas from the essays are highlighted and directly associated with specific artworks, making connections explicit and accessible.

Is this platform curated or automated?
Both. The connections are generated using artificial intelligence, drawing from existing metadata, texts, and artworks, while remaining grounded in the editorial and curatorial frameworks of KADIST and e-flux. 

Who is this platform for?
The platform is open to everyone. It may be especially useful for:
- artists and curators
- researchers, students, and educators
- writers and critics
- anyone curious about contemporary art and ideas
No prior expertise is required – just curiosity.

Is user interaction tracked or used to train AI models?
No user-generated searches or interactions are used to train external or third-party AI models. Interaction data may be observed in aggregate (for example, to understand which types of queries are most common or where the interface needs improvement), but it is not used to retrain the model, monetize attention, or extract behavioral data. This platform is not designed as a feedback loop for AI optimization, but as a public-facing research and mediation tool.

How does this project respond to concerns around AI and authorship?
These concerns are central to the project. All texts and artworks remain fully attributed and linked to their original contexts. This experiment positions AI not as an author, but as a mediator, one that opens possibilities rather than providing answers.

What is the ecological impact of this AI tool?
This platform is designed to keep computational use to a minimum. It relies on pre-processed data and a largely static, optimized database, rather than continuous or real-time AI computation. The system does not use generative models or perform heavy calculations each time a user searches or navigates the platform. Most of the AI processing takes place upstream, during the structuring and linking of existing texts and artworks, not during day-to-day use. As a result, the platform functions primarily as a lightweight research and mediation tool rather than an energy-intensive AI system.