What AI can realistically do for cultural institutions — and where the limits are.
Module 2 made the problem visible: administrative work is consuming time that belongs to the mission. Now the question becomes practical. Which AI tools address which problems? And just as importantly — which problems should AI not touch?
This module cuts through the noise. No vendor promises, no science fiction. Just a clear map of what AI does well, what it does poorly, and how to tell the difference.
AI tools are not general-purpose. They are very good at a specific set of tasks — and it is not a coincidence that those tasks overlap heavily with the administrative workflows cultural institutions struggle with most.
Reading, summarising, translating, drafting, and reformatting text. This covers a surprising range of daily tasks — from grant reports to exhibition labels to donor emails.
Finding structure in large, messy datasets — classifying images, flagging anomalies, matching records, or identifying objects in a collection. AI is fast and consistent where humans are slow and variable.
Routing, reminding, formatting, sorting — tasks that follow predictable rules. When a process is the same every time, AI can handle it without human attention.
The gap between vendor claims and institutional reality is wide. Not because the technology is fraudulent, but because context matters enormously. A tool that works well for a large national museum with a digitised collection and a dedicated IT team may be useless — or harmful — for a small regional archive with two staff members and a spreadsheet.
AI tools work best when they are narrow in scope, matched to existing workflows, and supervised by the people who know the collection. The institutions that see real results are the ones that start small, measure honestly, and expand from there.
Five types of AI tools are commonly used in cultural institutions. Five workflow problems from Module 2. Click a tool on the left, then click the workflow it best addresses on the right.
Reading about AI tools is one thing. Choosing the right approach in a real situation is another. Here is a scenario that cultural institutions face regularly.
The question is not "should we use AI?" — it is "which workflow, which tool, and how do we measure success?" Boards respond to evidence. A small pilot that shows measurable results is more persuasive than a large promise.
The best AI projects start with the people who know the work. If a tool cannot be evaluated by the archivist or registrar who will use it, it is probably not the right tool yet.
Not every AI tool is right for every institution — but the right tool, applied to the right workflow, changes what is possible.
[1] MuseumNext — How Museums Are Using Artificial Intelligence, 2024
[2] American Alliance of Museums — Transforming Museum Workflows with AI, 2024
Independent AI & IT advisory for cultural institutions. Based in Berlin. Working internationally.
Links
Stay in touch
© 2026 Artorythm
