18 September 2026

^{Still from 700 Jaar Enschede - De Film. The Dutch city of Enschede celebrated 700 years of city rights in 2025 with a 27-minute AI-assisted film travelling from the medieval settlement to the modern city.}

 

Put on a virtual-reality headset and the modern world disappears. In its place is Amsterdam in 1652: a city of narrow streets, wooden jetties and ships gathered around a busy waterfront.

You are not watching it through a screen. You can walk through the reconstructed city and explore it alongside other visitors. They inhabit the same virtual Amsterdam, turning what could have been a solitary encounter with a headset into a shared journey.

Amsterdam 1652: The Hands of the City is the centrepiece of ENTR, which describes itself as the Netherlands’ first cultural VR museum. The 45-minute experience takes visitors from the IJ waterfront to Dam Square, showing seventeenth-century Amsterdam through the eyes of its inhabitants.

It was also shown in the official competition at Venice Immersive, part of the 2026 Venice International Film Festival. The experience belongs to a new generation of cultural productions that do more than place digital information beside a historical object. They attempt to put visitors inside history itself.

Museums and cultural organisations are using virtual reality, generative AI, 3D modelling and conversational systems to reconstruct lost places and make collections more interactive. These technologies can help people explore a vanished city, watch centuries of change or ask a historical object questions.

That power brings its own complication. The more completely a historical world surrounds us, the easier it becomes to forget that somebody, or something, had to invent the missing pieces.

 

How virtual reality rebuilds a lost city

 

Creating a virtual version of seventeenth-century Amsterdam involves considerably more than asking an image generator for some attractive canal houses.

Historical maps, drawings, archaeological findings and architectural research establish what is known about the city. Artists and 3D modellers turn that evidence into streets, buildings, ships and everyday objects. A game engine renders the environment as visitors move through it, while tracking technology places them in the virtual world.

Sound completes much of the illusion. Voices, footsteps, creaking timber and activity around the docks can make an immaculate digital model feel inhabited.

But no archive contains everything. A surviving map might establish the curve of a street. An archaeological find might reveal the shape of a roof tile. Neither tells the designers who stood outside the baker’s shop on a rainy Tuesday morning.

Every historical reconstruction therefore contains gaps. Its creators must decide how to fill them.

 

How Enschede used AI to reconstruct 700 years

 

My own city took a different route into its history. Enschede celebrated 700 years of city rights in 2025 with a 27-minute AI-assisted film travelling from the medieval settlement to the modern city.

You can watch the trailer on YouTube here.

I have walked through today’s Enschede countless times. It is strange to see versions of the same place before its textile factories, shopping streets and familiar landmarks existed.

The production was made by Enschede filmmaker Jan Eppink and his team at Say Yeah Motion Media. But it began in classrooms across the city.

More than 1,800 young people participated in the 700 Jaar AI Challenge, which involved over 40 classes from eight schools. Groups worked on different episodes in Enschede’s history and created short AI films about periods stretching from the Middle Ages to the present. Their work helped provide inspiration and input for the final production.

Say Yeah then developed a complete script, which the company says was verified by several historical organisations in Enschede. A specially assembled collection of generative AI systems produced more than 4,000 images.

These did not simply travel from an AI model to the cinema screen. The filmmakers assessed the output, selected suitable images and wrote targeted prompts when particular scenes or details needed more work. Approximately 400 resulting video fragments were combined with composed music, sound effects and many hours of human creative work.

Those figures reveal both the speed and the limitations of AI filmmaking. A machine can generate thousands of possible images, but producing more material is not the same as making a film. People still had to research, verify, reject, refine and assemble it.

Amsterdam uses virtual reality to place visitors inside one moment in history. Enschede uses generative AI and film to carry an audience across seven centuries. One makes history spatial; the other makes it cinematic.

 

When an AI reconstruction meets a real memory

 

The Enschede film also shows that reconstructed history can affect people who remember part of it.

When the production was screened for several weeks at Concordia theatre, one visitor was a woman in her eighties who had experienced the wartime bombing of Enschede. The film depicts the devastating Allied raid of 10 October 1943, in which bombs fell on the city and killed 151 people.

According to a reaction reported by regional broadcaster RTV Oost, the reconstructed scenes stirred such powerful emotions that the woman was moved to tears.

AI cannot recover her experience of that day. It cannot recreate the sounds, fear or confusion stored in one person’s memory. But it can place an image of the event before someone who was there—and before younger residents for whom 1943 might otherwise remain a date in a history book.

That encounter raises the stakes. A fictional medieval pedestrian cannot object to the colour of his coat. A living witness can recognise when a reconstruction approaches something truthful, or when it does not.

 

AI-generated history on social media

 

Reconstructed history is no longer confined to museums and commemorative films. Instagram, TikTok and YouTube are filled with AI-generated journeys through ancient Rome, medieval villages and Victorian streets.

Their appeal is obvious. In seconds, we can apparently look through a window into an age that existed before photography or film.

Online, the medieval past is remarkably clean. Peasants have excellent teeth, clothes arrive in tasteful earth tones and even the mud appears art-directed. Buildings from different centuries share the same street, while everyday life acquires the polished lighting of a prestige television drama.

Generative AI does not know what the past looked like in the way a historian does. It generates images by learning statistical patterns from large amounts of training data. Unless carefully directed and checked, it may blend archaeological evidence, old paintings, film sets, fantasy illustrations and other synthetic images into something that feels authentic without being accurate.

There is an important difference between inventing a past and doing justice to it. The first begins with the desired spectacle: show me daily life in a medieval city. The second begins with evidence, establishes what is known and only then uses AI to visualise it.

That does not mean every historical AI film must qualify as an academic reconstruction. Imaginative work can be enjoyable and worthwhile, provided it is presented as interpretation. The trouble begins when visual confidence is mistaken for historical certainty.

 

AI is making museum objects talk

 

Some cultural institutions are making history interactive in another way.

In the gardens of Versailles, visitors can scan a QR code and speak to 20 statues and fountains. A system developed with Ask Mona and OpenAI allows each one to answer spoken questions about its history, symbolism and mythology. According to the palace, its curators supplied and checked the material used by the system to improve the reliability of its answers.

It resembles a personalised audio guide, except that visitors determine where the conversation goes. One might ask how a sculpture was made; another may want to hear the myth behind it.

Not every interactive museum experiment produces deeper understanding, however. In 2018, the Smithsonian began testing Pepper humanoid robots at several of its museums and research centres. Visitors enjoyed them, but a subsequent evaluation found that crowds, background noise and unfavourable lighting limited their reliability as information sources. The robots proved more effective at attracting attention, broadcasting simple information and making visitors feel welcome.

A technological spectacle may persuade someone to stop. That does not automatically mean they leave knowing more.

 

Historical accuracy in AI reconstructions

 

Historical films, paintings and museum models have always combined evidence with interpretation. Even a carefully researched diorama contains decisions about colour, atmosphere and human activity.

Immersive technology makes those decisions easier to overlook. A caption signals that someone is explaining the past. A world surrounding us can feel like the past itself.

Creators therefore need to distinguish between documented fact, probable reconstruction and visual invention. A historically checked script, as used for the Enschede production, supplies a firm foundation. Curator-checked information, as at Versailles, can reduce the risk of conversational AI freely improvising.

Transparency need not spoil the experience. It can make it more interesting by revealing that reconstructing history is itself an investigation.

Used carefully, AI and virtual reality can return an excavated object to its original surroundings, rebuild a street lost to fire or redevelopment and allow people with limited mobility to experience an inaccessible site. They can also make us look differently at places we thought we knew.

Much of medieval Enschede has disappeared, although the oldest part of the Grote Kerk still survives at the Oude Markt. AI cannot recover the world that vanished around it. But it encouraged 1,800 young residents to explore that world, and may persuade them to look at those familiar streets and wonder who walked them first.

 

^{Still from Amsterdam 1652: The Hands of the City (credit: ENTR)}