Visiting a Museum With Gemini or ChatGPT in Your Headphones
A practical guide to using AI as your on-demand, real-time museum companion
Most museums still rely on the traditional model: brief labels, fixed audio guides, and docents who deliver one narrative to a large group. This works reasonably well for the highlights of a collection, but it becomes limiting as soon as you want deeper context, comparisons, or answers to questions the museum didn’t plan for.
AI—Gemini, ChatGPT, or any capable model—changes this dynamic. Used properly, it makes museum visits more flexible, more efficient, and more responsive to your own curiosity.
While the examples below draw from MoMA (because the collection is especially varied and well-documented), the method applies to any museum in the world.
Docents follow set narratives; AI follows your questions
Museum tours, even at major institutions like MoMA, usually follow a predetermined storyline. Guides typically focus on key interpretive points: for example, that Les Demoiselles d’Avignon shattered traditional representation through its fractured planes and African mask influences, that Pollock’s One: Number 31 exemplifies the physicality and improvisation of action painting, or that Van Gogh’s The Starry Night conveys the psychological intensity of his late period. While accurate, these explanations are necessarily high-level and curated for the entire group.
Many questions that a curious visitor might have fall outside this script. You might wonder whether the subtle surface cracking in Brâncuși’s Bird in Space is a deliberate effect of his polishing technique, or whether it results from metal fatigue over time. You might notice a small photogram in MoMA’s photography collection and want to understand the technical difference between a photogram, which is made without a camera, and a contact print, which reproduces a negative. These kinds of detailed, material- or process-oriented questions often exceed what a docent can cover on a standard tour, either because of time constraints or the guide’s specific expertise.
AI in your headphones can answer these questions immediately. You can follow up with additional queries, ask for comparisons, or request clarifications in real time. For instance, after asking about Brâncuși’s polishing, you could ask how similar techniques were used by contemporary sculptors like Giacometti or Moore, and the AI can provide context while you continue observing the work itself.
This functionality is not limited to MoMA. Whether you are at the Rijksmuseum exploring Rembrandt’s etchings, at Tate Modern examining Rothko’s abstractions, at the Metropolitan Museum of Art comparing different Renaissance altarpieces, or in a small regional museum in Finland studying local modernist works, AI enables you to pursue your own line of inquiry instead of being constrained by the museum’s pre-defined narrative. It transforms the visit from a passive reception of curated information into an interactive exploration driven by your curiosity.
AI lets you keep your eyes on the art rather than your phone
The whole point of a museum is to be present with the work—to slow down, reflect, and let new understanding emerge. Too often, visitors interrupt this experience with screen time: looking up artists, scrolling through pages, or trying to find context online. This turns a visit into a digital research session rather than a personal engagement with the art.
An AI audio companion allows you to maintain that focus while deepening your understanding. For example, standing in front of a Roberto Matta abstraction at MoMA, you could ask how his construction of abstract space might have influenced Francis Bacon’s later figurative paintings. The AI can explain that Matta’s fluid, biomorphic spatial compositions prefigure the psychological tension and distorted interiors seen in Bacon’s work. This kind of insight is not on the plaque, not in the audio guide, and not part of the docent tour.
The information comes verbally, keeping your attention on the object. This preserves the meditative rhythm of a museum visit while allowing you to notice details and think critically about connections. AI does not replace reflection; it enhances it, helping you see more, think more deeply, and leave with a richer understanding of the art before you.
AI reveals the “quiet” parts of any museum collection
Most museums, by necessity, highlight only a small fraction of their holdings. The rest—minor paintings, small sculptures, experimental photographs, textiles, and one-off prototypes—often has minimal signage and no audio commentary, due to limited funding, scholarship, and other institutional constraints. These works can easily be overlooked, but they often contain rich visual and historical information.
AI is particularly useful for these less-documented areas. Even when there isn’t much information available online, an AI can help you generate ideas for further research, suggest connections between works, and identify questions to explore in more depth later with a curator or scholar. You can ask about when a work was produced, the movement it belongs to, its significance, and other contextual details. Even if the AI doesn’t provide definitive answers, the interaction helps you think critically and prepare informed questions for deeper study.
By guiding your curiosity, AI turns under-documented sections of the collection into spaces where you can explore, investigate, and develop your own understanding.
Dealing with imperfect audio recognition: how to ask questions clearly
Voice recognition is imperfect—this is not a minor concern but a defining part of using AI in a museum. Galleries are noisy, you are speaking quietly, and names in art history are often foreign, difficult to pronounce, or spelled in inconsistent ways. The model might not understand what you said on the first attempt.
There are straightforward strategies to avoid errors and wasted queries:
1. State the name of the artwork, the artist, and the year
If the AI mishears one element, it can still match the others.
Instead of saying: “Tell me more about “Demoiselles d’Avignon”,” try, “Tell me about Les Demoiselles d’Avignon, Picasso, 1907.” Most museums post the year clearly. Use it. Years reduce error dramatically.
2. When you can’t pronounce something, spell it
This works far better than guessing the pronunciation. If you don’t know how to say Brâncuși, Richter, Lichtenstein don’t worry neither does the AI, but the AI recognizes letters more reliably than imprecise phonetics.
3. Identify the object by location if necessary
If you’re at MoMA, you can say: “I’m in Gallery 210, the one with the Pollock. Explain the smaller Judd aluminum piece to the left.”
AI doesn’t know your location, but it can use your description as a proxy.
4. If all else fails, type
If all else fails, you can type your question instead. You can ask AI to write an essay by typing “write an essay on this topic,” and when it’s done, use the read-aloud feature so you can put your phone away again and return to looking at the art.
Managing hallucinations: a procedural solution
AI occasionally generates incorrect facts. This is normal. Unlike a docent, the model will always attempt an answer, even when uncertain.
The solution is to adopt a simple verification protocol:
Ask your question.
Then ask: “List what you know is documented vs. what you are inferring.”
Ask for sources, publications, or museum catalog references.
If the answer seems too confident, request limitations:
“Give me only information confirmed by MoMA or multiple independent sources.”
This reduces hallucination risk to a manageable level. Think of it as treating the model like a research assistant who needs occasional correction.
A more efficient and customizable way to visit any museum
AI does not replace curation. It supplements it. It gives you a more flexible way to move, think, and focus during a museum visit.
You avoid staring at your phone.
You follow your own interests.
You can investigate obscure works as easily as famous ones.
You get dependable context on demand.
You develop better questions and clearer ways to ask them.
This approach works in MoMA, the Louvre, a science museum, a modern design museum, or a one-room local collection. AI doesn’t make museums less human—it makes the experience more aligned with your actual curiosity.
You might also find that using AI in this way changes your relationship to museums. Instead of treating a visit as a one‑time event, you might start returning to the same institutions because each pass through the galleries becomes a new conversation. You notice more, ask different questions, and make connections that weren’t apparent before. Over time, you might become a member, not out of obligation, but because the museum becomes a place you actually engage with. AI may deepen your experience, encourage repeat visits, and more importantly end up indirectly supporting the institution. And this is something we should all get behind!