Artificial Intelligence now curates shows, and people lose their minds. This is only the beginning of this idea, though, and we will get good at it. And, friendly reminder, photography did not replace painting.
The New York Times released an article last week about how the Duke University Nasher Museum of Art prompted ChatGPT to curate its latest show. The experiment predictably outraged a comment section intent on misunderstanding the experiment and thus proved the experiment’s worth.
Is this not always the case for new things? One valuable response, however, pointed to this New Yorker article written by a computer scientist that sits us down to explain the birds and the bees of A.I.
The curator of the future?
My interest, naturally, lies in the dataset used to curate the show. In other words, the algorithm bloodhounds the internet for traces of digitized collection information from cataloged objects relevant to the prompt entered into ChatGPT. In other words, the catalog information of the collection available online. In other words, the work of registrars and collections managers.
In thinking about the implications of this, I decided to appropriately ask AI to develop ideas about this topic (I did not edit a single word of the generated ideas). Of course, I had ideas of my own but the list that it compiled went deeper due to its more complete understanding of artificial intelligence (a.k.a. itself). I will look at each.
Idea 1: Utilizing AI to digitize and catalog art and artifacts in museums, creating a comprehensive database for curating exhibitions.
Currently, AI uses the available data to curate these imperfect exhibitions. Soon, we will begin to document these objects in a way that anticipates its use by AI to curate shows.
Idea 2: Developing AI algorithms that can analyze and categorize art and artifacts based on various criteria such as style, period, artist, and cultural significance.
On the surface, this already happens by registrars, but what if AI uses data already out in the world to do it for us? In other words, it can use already airborne images and information about Roman ceramics to suggest to the registrar cataloging the work that not only is this a Roman vase but it likely originates from x period. Similarly, it can suggest a work derives from the brush of Monet from x period.
Idea 3: Implementing AI-powered image recognition technology to identify and categorize art and artifacts based on visual features.
Expanding upon the previous idea, image recognition will anticipate the collections of objects to input. Take the case of the article discussed last week of the British Museum and their 4 million uncatalogued objects that posed a security risk. Perhaps a large portion of the AI could create basic entries for those objects. Yes, they would require and initial photograph and further review and refinement by a human, but they would get catalogued.
Idea 4: Creating AI algorithms that can automatically generate exhibit themes and connections between different artworks and artifacts based on their cataloged information.
To riff on a previous idea, not only will we catalog in a way that encourages AI use, we will artfully rewrite algorithms to more eruditely interact with the data to produce even more interesting results. I will go one step further and suggest that future curators role will requiring an expertise with the algorithm, prompts, and and overall ability to interact with the data in new and interesting ways.
Idea 5: Using AI to analyze visitor data and feedback to better understand their preferences and interests, and tailoring exhibitions accordingly.
This idea does not derive from collections management work nor does it inspire must of a collections response beyond prioritizing work on or with the most popular or sought after objects.
Idea 6: Employing AI to assist curators in identifying gaps or missing pieces in the museum's collection and suggesting potential acquisitions or loans.
This is an interesting idea. Again, it will require a level of nuanced skill with the technology in order to interact with the objects and data in interesting ways. The interpretation of existing collections will generate huge amounts of useful information.
Idea 7: Developing AI systems that can analyze historical and cultural contexts to provide deeper insights into the significance and interpretation of exhibited art and artifacts.
See above regarding the curator’s abilities to plug into the matrix of information. We will learn so much about ourselves through forced interpretations and juxtapositions we could not have conceived on our own.
Idea 8: Utilizing AI-powered natural language processing to generate informative and engaging descriptions for each artwork or artifact in an exhibition.
Not only will AI interact with factual data, but it will also smoke the cigarette of interactive text and exhale wild interpretations of objects from the lungs of its algorithms.
Idea 9: Integrating AI chatbots or virtual guides that can interact with visitors, providing additional information and answering questions about the artworks and artifacts.
The public will have easy access to all of the information compiled, explained, interpreted, and anticipated from the above interactions.
Idea 10: Using AI algorithms to analyze social media and online platforms to identify emerging trends and artists, informing curators' decisions for future exhibitions.
Lifting the hatch on the controlled space of the museum data bomb shelter, the river of factual and interpreted institutional object data will reach the ocean of internet text, ideas, and thought which will inspire AI to infinitely build and rebuild ideas with Legos of thought.
