Why Generative AI Interests Cultural Institutions
Generative AI (text, image, sound) is no longer reserved for technical teams. It's becoming an accessible tool for concrete tasks: writing multilingual labels, creating communication visuals, generating accessibility descriptions, designing personalised mediation experiences.
The 5 First Steps
1. Identify a repetitive, low-risk task — already done manually, easy to verify
2. Test with non-sensitive data — public content, not internal databases
3. Train 1-2 reference people — avoid uncontrolled mass access
4. Define a usage charter — what can be used, for what, who validates before publication
5. Evaluate after 3 months — what was accelerated? what problems?
What Generative AI Doesn't Do (Yet)
- It doesn't "understand" your collection in the museographic sense
- It doesn't replace curator expertise on attribution, restoration, historical contextualisation
- It sometimes produces plausible but false content (hallucinations) — human review remains essential
Talk about your situation (15 min, free) →
A project in mind?
Book 15 min →