From ukiyo-e woodcut paintings that have fading over time in Japan or Song Dynasty calligraphy that has become colored in China, to ancient Buddhist works in Korea that have been effectively restored by artisans thanks to the "powerful assistant" of AI. However, the intervention of AI in heritage also raises controversial questions: What is the line between recovery and regeneration? Is AI contributing to preserving or is it just reenacting the past in its own way?
AI joins heritage conservation
In Japan, a famous country that respects tradition, the technology of removing ancient paintings using AI is gradually being accepted in the museum community. According to an article in Asahi Shimbun, a group of researchers have collaborated with art museums to use deep learning algorithms to restore roll paintings and especially ukiyo-e woodcut paintings. ukiyo-e paintings are one of the oldest art forms of Japan that developed strongly during the Edo period (1603-1868).
These algorithms are trained by reading tens of thousands of original works to learn how to coordinate colors, drawings and layouts. Thanks to that, AI has the ability to "imagize" the painting that has been blurred or completely lost - something that cannot be done by traditional restoration techniques. There are works that have been almost completely restored, helping modern viewers access and admire the most classic beauty that seems to have been erased by time.
Meanwhile, Korea is investing heavily in the field of cultural preservation technology. According to The Korea Herald, a number of technology startups cooperate with national museums and national cultural heritage agencies to digitize and restore folk paintings and Buddhist paintings from the Joseon Dynasty (1392-1897). Technology here does not stop at cleaning digital paintings, but also uses AI to recreate primitive colors, identify paper and fabric materials and analyze the painting techniques of ancient painters.
In Seoul, another project using AI in heritage conservation is also being implemented with the goal of restoring paintings using a 3D model, in which AI analyzes the physical structure of silk paintings to propose a restoration plan without touching the original artifacts - which is extremely important for fragile or seriously degraded works.
The story in China has a more ambitious scale. According to the Huan rang Post Office, scientists at the Chinese Academy of Sciences used deep learning tools to restore murals in Mac Cao cave (Don Hoang) located in Gansu province, which preserves thousands of Buddhist paintings from the 4th to 14th centuries. These paintings have been severely damaged by environmental conditions and time. AI technology is used to analyze colors, patterns, and layouts from intact works, thereby "re-creating" lost parts with high accuracy.
Technology is even integrated with virtual reality (AR), allowing visitors to wear AR glasses when coming to the museum to see a blurred painting clearly like in its heyday. It is a unique combination of classic visual tradition and modern technology experience.
Is technology really a bridge between the past and the future?
Although it has brought many breakthroughs, the use of AI in ancient art restoration is not uncompromising. A big question arises in the heritage conservation community: Is the part that AI recovers really a real reflection, or is it just a product of machine judgment? And more importantly, who owns the right to re-create the past - humans, or algorithms?
In Japan, many scholars are still skeptical. According to them, restoring ancient paintings is not simply "making beautiful paintings return", but a process of absorbing knowledge, emotions and historical judgment. Some Kyoto researchers believe that AI's automatic repainting of lost paintings can make viewers mistakenly believe that it is the original, thereby overshadowing the natural historical process of the work.
In China, although the technology is widely supported, some experts still recommend that AI should only be a support tool, not the final decision maker. An expert in the field of art history of this country once shared: "We cannot let machines tell ancient stories in their modern language without cultural verification and historical context".
There are also concerns about the abuse of AI in the exhibition space. When "virtual restoration" paintings are shown publicly in eye-catching colors, there is a risk that viewers will lose respect for the real work - a place where time has left more valuable traces than any technology.
However, it is undeniable that in the context of a shortage of conservation human resources, limited budgets and an increasing number of heritages, AI is an extremely effective support tool. It allows museums to digitize huge archives, restore and display quickly, thereby expanding public access to cultural heritage, especially young people.
AI is not a human replacement for performance art - but a means for people to better fulfill their role in heritage conservation. With proper control and monitoring, AI can become a bridge between technology and tradition, between present and history.
We can imagine a future where AI is not only used to restore ancient paintings, but also to teach art, simulate the creative processes of ancient masters, or even open up global virtual exhibitions where people can "enter" 12th-century paintings without having to leave their seats.
Importantly, when technology is used with understanding and respect, it does not overshadow the past - but makes the past more vivid than ever.
