Artificial Intelligence (AI) is blooming across all sectors, with arts and culture being no exception. In museums, visitors can now talk to an AI to learn more about a work of art and medieval manuscripts are digitized by AI. Artists can even use AI to create new art. But AI’s development also raises fears among artists and others who make their living from culture. Are those fears justified and what will happen next?
AI at the museum
French startup Ask Mona designs chatbots to provide practical information to visitors at cultural institutions. The Louis Vuitton Foundation in Paris has adopted Twelvy, a bilingual conversational AI that provides personalized practical information to visitors, such as opening hours, tickets rates, programming, itinerary, and even a game based on current exhibitions. The Louvre Abu Dhabi and the Centre Pompidou (Paris) also employed a trilingual virtual assistant. The Centre Pompidou’s tool, designed by Ask Mona, offers the option of scanning in an image of an artwork to receive video, and podcast analysis of it.
Along with the exhibition “Van Gogh in Auvers-sur-Oise, the Musée d'Orsay developed a virtual reality experience.” Going by the name “Van Gogh’s Palette,” it immerses the visitor in Van Gogh’s work and allows them to explore it from within as they would a landscape. One can mix Van Gogh’s favorite colors while sitting in front of Marguerite Gachet (Van Gogh doctor’s daughter as portrayed by the painter) and listening to her recount her memories.
This VR experience is presented along with “Hello Vincent,” an application developed by Jumbo Mana, a start-up specializing in generative AI related to human behavior. At the end of the exhibition, visitors are invited to speak to a virtual Vincent Van Gogh on a monitor. “What did Auvers-sur-Oise represent for you?” a journalist asks him. With a Dutch accent, the virtual Van Gogh replies: “It’s the place where I found refuge and my inspiration at the end of my life.” The French start-up developed the character in collaboration with Dr. Wouter van der Veen, an art historian and expert on Van Gogh who trained “Hello Vincent” using Van Gogh’s letters.
Like Musée d'Orsay, the Louvre Abu Dhabi will introduce the first mediation tool generated with AI since its opening. Sebastien Cotte, Digital Manager at France Muséums, who coordinates this tool, explained to Tamooda how it works. “The device has been set up with Chat GPT,” he said. “The visitor will be able to generate a tale in three languages (Arabic, French, and English), based on three parameters: animals, pointed moral and spatial framework. They then need to choose the tale’s style between Ibn Battûta and Jean de la Fontaine.” This experience will be available with an exhibition coming soon.
AI as a competitor of cultural professionals?
In a debate organized by the International Council of Museums (ICOM), entitled “Looking to the future? AI and museums” ICOM France President Emilie Girard pointed that AI could potentially cover many areas of museum expertise: “producing labels, texts, exhibition proposals, assisting with collections management, documentation, content development, communication, financial and administrative management…”
Without doubt, Ask Mona’s chatbot saves time for museum reception teams. The start-up emphasizes the fact that employees “can focus on other tasks with added value for the institution.” Yet employees are also right to be wary that AI may pose a threat to their livelihoods.
Institutions like Bibliothèque Nationale de France (BNF) have used Optical Character Recognition (OCR) technology to digitize their collections. After being scanned, old manuscripts are converted to digitized texts and added to BNF’s online resources. The tool can establish connections between words or groups of words and write indexes, a task that is usually carried out manually.
Despite this, an archivist who wishes to remain anonymous told Tamooda that he is not worried about the development of AI tools in his institution. He pointed out that the tool still requires systematic human intervention (such as paleographers and archivists) to train the model with samples and to proofread after its work is done. In other words, the machine requires examples of manual manuscript retranscriptions. He stresses that the tool is not very effective with handwritten material, i.e administrative documents, writers’ manuscripts and correspondence. The proofread afterwards sometimes takes more time than doing it manually. For him, AI is “an idol with feet of clay.”
There is also the matter of copyright. In attending a charity gala event organized by the nonprofit association “La Source”, I was invited to use an AI model to generate a portrait based on photos taken earlier at the event. I opted for the style of Gerard Garouste, a famous painter and the founder of the association. The AI tool then portrayed me as a chess player. The portrait was then sent to the guests with the possibility to buy. I wondered whether, when reselling the artwork, one would be able to assign credit to Gerard Garouste.
And what if the AI-generated artwork were submitted for an Art Prize? This is not a dystopia: in September 2022, Jason Allen submitted his work, designed with MidJourney, an AI system that creates images based on texts, to a fine-arts competition. The artwork, “Théâtre D’opéra Spatial,” won first place. In the days that followed, many artists expressed their alarm at Allen’s victory.
The ability of AI tools to write and edit texts and scripts has also caused a major stir among authors. ChatGPT processes vast datasets including websites, articles, and books to generate its responses. The technology makes no distinction between works that are copyrighted and those in the public domain. At the same time, artwork can be generated based on a person's original idea without their being given any. George RR Martin, author of A Song of Ice and Fire, which was adapted into the hit HBO series Game of Thrones, is one of the increasing number of authors who have sued AI companies (Chat GPT or META). Authors allege that “their copyright has been infringed to train the system,” according to the BBC.
In their five-month strike, Hollywood writers managed to obtain strong limits to the way AI can be used in the cinema industry. In the future, “AI-generated material cannot be used to undermine a writer’s credit or separated rights” reads to the new Writers Guild contract. In the rest of the world, AI regulation is still in its infancy.
AI in the MENA region
Unsurprisingly, Saudi Arabia and UAE are the MENA region’s leaders in AI. However, as Sabriya El Mengad reported, AI tools’ performance in Arabic is far below that in English and a number of other languages. Abu Dhabi announced plans to launch its own large language models (LLM), Falcon 40B and NOOR, to compete with the likes of OpenAI, DeepMind, and Google. The UAE also now has a Ministry for Artificial Intelligence, so watch this space.
