Collections Policy
The collection is organised as a documented digital archive of the history, workings, social consequences and possible futures of artificial intelligence.
1. Purpose and scope
The collection includes historical objects and documents, scientific papers, source code, software, interfaces, datasets, model cards, institutional and legal records, and other born-digital objects relevant to understanding AI. The history of AI in Greece is a distinct collecting priority.
2. Acquisition and selection
Evidence is included when it has a clear curatorial purpose, adequate provenance, a reliable source and a lawful presentation route. Primary sources, institutional archives and material with clear reuse rights are preferred.
3. What counts as an exhibit
An exhibit may be a photographed physical object, historical document, paper, source code, software interface, dataset, model card, screenshot, government or legal document. Physical materiality is not required; provenance and documentation are.
4. Original, reproduction and curatorial representation
Every exhibit separates the evidence itself from the way the museum represents it. Photograph, scan, facsimile, reconstruction, diagram and artistic rendering are not treated as equivalent.
5. AI-generated imagery
An AI-generated image is not historical evidence by itself. It may be used only as clearly labelled curatorial illustration when it does not replace available authentic or better-documented material.
6. Born-digital evidence
Websites, repositories, model cards, interfaces and datasets are treated as evidence of their time. URL, institution, date and, where possible, stable or archived references are recorded. Content is not copied when reuse rights are unclear.
7. Rights and licences
Online availability is not permission to reuse. The museum uses its own work, sufficiently clear public-domain material, Creative Commons material under its terms, or material used with permission. Otherwise it links to the primary source or uses a documentary metadata facsimile.
8. Museum records and metadata
Each exhibit records a permanent accession ID, a separate exhibition position, title, date, creator or institution, evidence type, original/reproduction status, provenance, primary source, image rights, certainty where applicable, relationships to other exhibits and the date of the latest curatorial review.
9. Relationships and context
Documented relationships such as “led to”, “influenced”, “responded to” or “uses” are part of the collection itself, not merely navigation.
10. Correction, reclassification and deaccession
The collection is living. An exhibit or representation may be corrected, reclassified or withdrawn when stronger evidence, a rights issue or a substantial curatorial problem emerges. Established URLs are preserved where possible and changes are recorded.
11. Access, reuse and licence for original content
The collection is free, bilingual and requires no account. Documentation remains public. Original curatorial texts created by the Museum of Artificial Intelligence are licensed under CC BY-SA 4.0, unless stated otherwise. This licence does not extend to images, historical records, code or other third-party material; reuse rights for those items follow the licence of their original creator or institution.
12. Care and sustainability
The museum hosts locally only material it can lawfully host, uses versioned assets, checks broken links and records curatorial review dates. Stable references and documented provenance are preferred for born-digital objects.
13. Public corrections and proposals
Suggestions for new evidence or corrections are evaluated by the same standards of provenance, relevance and rights. Inclusion is not automatic simply because an object is popular or visually striking.