How Museums Can Protect The Past By Shaping The Future Of AI

Museums, typically, deal almost exclusively with the past. They’re gateways to history, preserving the most important parts of it to educate and inspire the public. But while museums might be the conduits to the past, they can still play a very important part in how history is remembered in the future.

Who Teaches AI About The Past?

AI is already changing how many people interact with and understand their history. What was once a question reserved for experts or scholars can now be typed into a chatbot, with an answer generated in seconds as entire corpuses are distilled down into fragments and phrases. But a fundamental question still remains: how does AI know the information it has is correct?

An information’s accuracy is critical, especially when it comes to history, culture, and arts, as there is often so much context that goes around a piece. One small error or inaccuracy, or even a typo, can dramatically change what an AI understands, and misinformation can suddenly become presented as fact.

The development of large AI models has involved training on enormous quantities of web-based content, however the provenance of this content is questionable and datasets lack transparency. It’s also not guaranteed to be free of error or bias. And when millions of people rely on something to be correct, this presents a serious issue.

The Problem: Available Knowledge Isn’t Always Correct

Generative AI systems are typically trained on enormous quantities of information. This isn’t a problem for them, as there is hardly a lack of information online. Where there is a problem, though, is in finding credible, accurate, expert-reviewed information.

If an AI developer’s effective bar for quality is “anything that exists and the AI has read”, it can lead to an AI ingesting false, incorrect, or spurious data, which it then uses to create answers that lack the necessary context. Without oversight, it can also inadvertently train on harmful stereotypes, false accounts, or bias. This is especially true for historic and cultural subjects, as they can be particularly vulnerable to outdated information or missing context.

An AI-generated answer can sound authoritative, but it isn’t necessarily correct. The AI also then has the potential to perpetuate these inaccuracies indefinitely - eventually influencing and potentially altering what people believe things to be. If the public is going to ask AI about cultural heritage, it needs to have the correct information.

Museums Already Hold The Answer

Within the annals and archives of the UK’s museums lie millions of artifacts, exhibits, objects, and records that have been painstakingly preserved for future generations. But more than simply being the guardians of these collections, museums have the context and the knowledge to explain why they’re so meaningful.

In many cases, some of these artifacts are already digitised. They exist on servers and in the cloud, as well as on the museum floor, and hold within them the keys to humanity’s collective knowledge. But they aren’t being utilised.

Visitor numbers to museums are falling. The Department for Culture, Media and Sport (DCMS) released figures in 2026 showing that visits to the 15 DCMS-sponsored museums and galleries had dropped 16% in 2024/25 compared to 2018/19. Visits over the same periods by under-16s also fell by 9%. At the same time, museums face significant financial pressures. But the contents of those museums might hold the key for both teaching the younger generations about their history, and helping to manage a drop in finances. Licensing doesn’t necessarily bring these visitors back, but it can help to plug the gap.

New exhibits can help to counteract this, as seen with the British Museum’s stewardship of the Bayeux Tapestry. Approximately 100,000 tickets were sold on the first day of public sales and, in just its first month, it has welcomed over 100,000 visitors. But a constant stream of new exhibits can’t be relied upon to bring people back to museums and, what’s more, potential visitors aren’t always able to secure tickets. That’s even before considering travel to London or lengthy queues.

For the Creative Content Exchange (The CCE) , this illustrates a potential use for cultural content that goes beyond training AI models. Bringing together images, rich metadata, expert commentary, and clear permissions could help developers create educational and accessible services that can be conveniently viewed and accessed for many more visitors than the million expected to visit by next summer.

Licensing Can Unlock Value For Museums And AI Alike

AI developers desperately need high-quality data with which to train. Museums often have collections that are already digitised, and which would be suitable for licensing. This creates a unique opportunity for both, where AI developers are able to source high-quality, rights-checked, factually-accurate data with which to train their software, while museums are able to provide licenses for data that ensures they stay in control. Not only does this ensure that museums are paid a fair amount for their collections, but it helps to multiply the power of those collections - allowing far more people to benefit from them than would otherwise have been possible.

Licensing Can Create A Chain Of Provenance For AI

The issue for AI companies isn’t in finding data, it’s increasingly about verifying its accuracy. Data coming from a museum increases the inherent trust and accountability a company can have in it, making it a far more valuable resource and something that AI companies are more willing to pay a premium for.

Not only does the AI developer benefit from this resource, but the public do, too. The greater depth of information provided to the AI by a museum means that the AI can give a much stronger explanation which, critically, includes context. Everyone - the museum, the public, and the AI - wins in this situation.

Context Is King In The AI Age

An object doesn’t exist in isolation. While an AI application may be able to recognise a painting and understand what it is, that doesn’t necessarily translate into understanding the network of relationships that give the painting meaning. To truly understand a painting, you might need to know who painted it, what its subject is, and when and where it was created. You might also want to know where the painter was from, the period in which they were active, and where that work sits in the painter’s catalogue.

This is often what’s missing from AI, and is what a museum is so good at communicating. These extra pieces of context help to give an object a much richer cultural history.

Protecting The Past By Funding The Future

Licensing provides museums not just with a potential new revenue stream, but with the chance to help shape the future.

Just the cost of conservation, storage, and staffing can be difficult to manage, and if a museum closes then some exhibits may end up in private hands - removing them from the public’s benefit. By choosing to license exhibits, museums are making sure that AI developers have access to more authoritative and traceable sources of information. Combined with the appropriate retrieval, attribution, and evaluation systems, this could help AI to provide more accurate and contextually-grounded answers.

A Museum’s Job Remains The Same, But How They Do It Has Changed

Museums have always been about preserving objects of cultural, historical, creative, or scientific importance to help educate and inspire the public. In that way, the role of a museum hasn’t really changed. What has changed is the way in which they can do it.

The internet gave people access to information wherever they were, while AI has given people a way to interact with information in a much more organic way where they can ask questions and receive answers in much the same way that they would if they were at a museum. These conversations are already taking place, and are being based on, at times, incorrect information. Licensing offers museums a way to not only ensure the public receives the correct, contextually-relevant information, but to see some revenue for their hard work as well.

The future of AI will be shaped by what’s put into it. Licensing offers museums the opportunity to not only protect the past, but to help ensure that the machines recounting it to future generations get it right. And, what’s more, that the museum can continue to provide that benefit for years to come.

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

Previous
Previous

Why British Data Is Crucial To British AI

Next
Next

Invitation to Tender