What Happens When AI Becomes The New Tour Guide?

The way that people interact with history is changing, and it begs the question: what are museums doing about it?

Extraordinary Knowledge, Limited Ways To Communicate It

Museums have an incredible amount of data at their fingertips, but often struggle to convey just how significant an object is. That’s because space is limited, and choices have to be made about both what gets exhibited and how much information is given about each entry. Oftentimes a small plaque bearing the name and year of an item might be all the information provided to a visitor, while behind the scenes the museum itself is sitting on a wealth of background and context that would educate and excite a visitor given the chance.

Audio guides can be a useful way to impart more knowledge, but they offer no way for a visitor to ask questions. Human tour guides can provide some back and forth, but are limited to the scope of their knowledge on an individual piece. Both create predetermined experiences, putting visitors on an invisible conveyor belt that provides a very similar or identical experience each time.

This creates a situation where the museum has far more high-quality, contextually-relevant information about the pieces within its collection, but simply has no way to give that to its visitors. It’s an interesting paradox: museums have more data than ever before, visitors have more ways to ask questions, yet the two remain unconnected.

The Tour Guide Is Becoming Conversational

There’s no single type of museum visitor. One could be an expert in the field, another might be a child, while a third could be a tourist who doesn’t speak English particularly well. But all want to know what the object is, and why it’s important. A single plaque or label isn’t going to be able to convey the type of information that each of them want given their different backgrounds, and it certainly won’t have the extent of the museum’s knowledge on it.

At the same time, curators want people to understand why a chosen piece is so important, and for them to leave feeling educated. But with British Museum data suggesting that older permanent gallery displays can sometimes only receive a few minutes of dwell time, and most visitors only stopping at a small number of objects in each display, it presents the need for a more conversational tour - and one dictated by the visitor themselves.

Once again, human tour guides can help here, but may have a fixed interpretation or history to recount that doesn’t have room for a variety of questions. The tour itself might not have the time to dedicate to a single piece, or an object which is of interest to some might not be included as part of the tour. On the other hand, an AI can answer follow-up questions and generate different experiences for these different types of visitors, but how does the museum ensure that what the AI is saying is correct? And, what’s more, do museums have a duty to ensure their collections are properly interpreted by these AI systems?

What Makes A Museum’s Knowledge Valuable To AI?

Museums already possess an enormous amount of knowledge about things both on display and in their collections. Within their systems and databases will exist things like:

  • Details and descriptions on collections

  • Curatorial research

  • Exhibition texts

  • Audio guides

  • Educational resources

  • Biographies

  • Bodies of work

  • Object metadata

  • Scholarly publications

  • Interviews

  • And more

The issue facing AI isn’t a lack of knowledge. It already understands what, for example, The Fighting Temeraire by J.M.W. Turner looks like, along with the painter, when it was created, what the method was, and so on. But what it typically doesn’t understand are the nuances and deliberate choices portrayed by Turner, and their impact on the painting itself.

AI might understand that in reality, the boat was brought in by two tugs, while the painting shows one. The painting also shows the ship with its masts, whereas these had actually been taken down prior to the voyage. These are both conscious choices made by Turner in his painting rather than factual inaccuracies, and are things that an art historian or tour guide can explain. They can help to give a richer and more contextual interpretation of the painting, while an AI might ignore or gloss over these facts to the detriment of a visitor’s experience.

Particularly in the art and cultural world, a machine’s understanding of innately human decisions doesn’t always translate into ones and zeros. Feelings and emotions that the work is intended to invoke in a person can’t be communicated or understood. This is where the museum comes in.

Turning “Training Data” Into Licensed, Monetised Knowledge

Though they work for the benefit of the public, museums do typically monetise commercial access to their knowledge. This could be in the form of TV productions, films, consulting, archives, and more, as it allows them to fund their ongoing works. AI should be no different.

A museum licensing their knowledge for AI use ensures that the AI is able to train on high-quality, factual, contextually-relevant information, while the museum can generate additional revenue through its pre-existing data. Museums remain in control, defining what is licensable, how it can be used, what rights and restrictions it has on the data, and what responsible use looks like, while creating a stronger relationship with the AI companies while providing a benefit to the public as well.

This gives museums the opportunity to connect with far more visitors than they would otherwise. Rather than having to wait for a tour or commit to an audio guide, visitors can explore objects in their own way and at their own pace, creating an experience driven by themselves. Utilising the museum’s knowledge, they tailor their own experience.

Museums Remain The Authorities

AI companies need the institutional expertise and cultural authority that museums have worked hard to cultivate, but licensing doesn’t mean a museum giving up its place as the authority. Museums remain the owner of the data, simply providing a license for AI companies to use it. The amount, the subject(s), and the way in which it is used are all the choice of the museum, while the AI company benefits from the relationship by having proven provenance to their data.

The AI acts as a conduit between the museum and the visitor, as all three see value from the situation: the museum connects with the visitors more while increasing revenue, the visitor receives accurate, contextual, culturally-sensitive information, and the AI company provides answers that are more cognisant of that knowledge.

What Could The AI-Enhanced Museum Experience Look Like?

Museums already understand what’s important to history, but the individual visitor decides what’s important to them. There’s no way to know what piece or pieces will pique a person’s interest, and which they’ll want to find out more about. This is where AI can be helpful.

A patron can find an exhibit that speaks to them, and have a conversation with the AI about what they find interesting. Through proper licensing, the AI would then be able to provide the information owned by the museum, which the visitor would otherwise have had no access to. This enables pieces to become more than just an object behind a piece of glass, and into something far more educational. Follow-up questions can deepen a visitor’s understanding, giving them a bespoke experience that otherwise wouldn’t have been possible.

In some instances, like the Teddy Roosevelt experience at the Presidential Library, or Eleanor of Castile at Leeds Castle, AI is used to power an immersive experience where a visitor can speak directly to a historical figure. This further helps a visitor to engage with the museum and history itself.

Then there are the experiences that use a museum’s data to create experiences that take place outside of the museum itself. The Living Museum, for example, uses the British Museum’s collection database and a large language model to create AI-powered chatbots for its collections.

AI Presents A New Distribution Channel For Museum Expertise

Museums have always searched for new and exciting ways in which to make their data accessible. UK institutions including the National Railway Museum, the Science Museum, and the Birmingham Museum & Art Gallery all use VR in some displays to immerse visitors into a different world, allowing them to experience things in a much more personal way. AI could be the next iteration of this, personalising a patron’s experience and giving them what they want to know, rather than what a curator thinks they should. Each visitor’s questions can be answered succinctly and with the museum’s data in mind, applying its staff’s extensive knowledge in a new, innovative way.

The Museum Of The AI Era

AI doesn’t have to spell the end of museums, or the end of visitors. Far from it. It actually presents the greatest opportunity for truly personalised visits that museums have seen. But how well that works is up to the museums themselves.

Licensing allows museums a way to ensure that the information given to visitors is credible, accurate, and properly attributed in a way that only they can do. It doesn’t cut museums out of the picture, but instead puts a far higher premium on the knowledge that they’ve worked hard to preserve. The question isn’t simply “should we use AI?”, but “how much do we want our expertise to shape the experiences of our audience?”. It’s an opportunity to protect the past while preparing for the future. And that’s a much more interesting proposition.

Interested in finding out more about the Creative Content Exchange? Get in touch with us today.

Previous
Previous

A Museum’s Guide To Licensing Data And Content

Next
Next

Why British Data Is Crucial To British AI