Putting the “Arts” in “Artificial Intelligence”

Artificial Intelligence (AI), or intelligence exhibited by machines that can mimic human cognitive functions, has become a common phrase in the tech world. AI has allowed computers to complete advanced tasks that were not possible before, and the applications of AI continues to increase and diversify. In fact, AI has even found a place in the arts. Technology professionals and arts managers are exploring a broad range of possibilities for AI within areas of the arts world.

For example, Baidu, an Asian Internet giant, has used AI to translate famous pieces of art into original melodies. Inspired by Robert Rauschenberg’s The 1/4 Mile or 2 Furlong Piece, the AI composer created two songs that were displayed at Ullens Center for Contemporary Art (UCCA) in Beijing in 2016 from July to August. 

On July 19, 2016, Baidu and UCCA co-hosted the “AI & Art Night” event to introduce Baidu AI composer to the public. (Photo Courtesy of Ullens Center for Contemporary Art)

On July 19, 2016, Baidu and UCCA co-hosted the “AI & Art Night” event to introduce Baidu AI composer to the public. (Photo Courtesy of Ullens Center for Contemporary Art)

Baidu’s promotional video introduced how paintings and pictures could arouse certain musical emotions and moods. For instance, a rainy day might evoke a melancholic melody; however, when a bright and beautiful beach scene appears, some light music might play in your heart. Inspired by the seemingly hard-to-explain connection between moods and pictures, Baidu trained its AI to identify the objects, colors, and setting, as each corresponds to certain moods and emotions. Then, the AI gets access to a vast database of musical scores that are broken down into musical units by moods and emotions, and composes a piece of melody using musical units corresponds with the visual elements it has analyzed in a certain image.

The two melodies based on Rauschenberg’s paintings are not the only achievements by Baidu AI composer. As shown in the promotional video, there are two other beautiful piano melodies inspired by Vincent Van Gogh’s The Starry Night and Xu Beihong’s Galloping Horses.  

Ullens Center for Contemporary Art. (Photo Courtesy of Ullens Center for Contemporary Art)

Ullens Center for Contemporary Art. (Photo Courtesy of Ullens Center for Contemporary Art)

When asked whether Baidu had any future plans on applying AI to the arts world, the PR representative stated, “The Baidu AI Composer program marks our emergence into the arts world. It is still at a preliminary stage. In the future, we will continue to do amazing things that initiate interactions between technology and other areas.”

Baidu is not the only Internet company that has taken a crack at discovering the possibilities between AI and the arts. In June earlier this year, Google, one of the most influential Internet companies in the world, announced a new research project called Magenta, developed by a small team of researchers from Google Brain team. The main goal of Magenta is to use machine learning to create compelling art and music, and the team plans on setting up a community of artists, coders and machine learning researchers.

A 83-second piano song, which is the first piece created by machine learning under the efforts of Magenta project team, has already been unveiled. According to a blog post written by Douglas Eck, a researcher working on the Magenta project, one of the biggest challenges for Magenta is not creating art, but actually telling a compelling story.

Ultimately, Google doesn’t want to use AI to only produce art by learning from samples, but to provide artists tools by bringing machine learning to the artists’ own creations. Magenta is a bold attempt by the creative Google Brain team, and it is not Google’s first attempt to collide AI with the arts. In June 2015, Google showed how computers interpret objects in images with an art generator called DeepDream. Google called this experiment “Inceptionism”, a tribute to the movie Inception, which tells a story of humans inserted into each other’s dreams through novel technology.  

A Google DeepDream picture. A group of 29 paintings created by Google DeepDream was sold at a charitable auction in San Francisco in February 2016. (Photo Courtesy of Google)

A Google DeepDream picture. A group of 29 paintings created by Google DeepDream was sold at a charitable auction in San Francisco in February 2016. (Photo Courtesy of Google)

Microsoft, another Internet giant, has also entered the arts world through AI technology. In September 2016, Microsoft announced its partnership with Tate Britain and used AI to bring Tate’s art collection to life. Microsoft’s Cognitive Services identifies expressions, themes, contexts and similarities in composition between pictures in the news and Tate’s art collection, making interesting comparisons between modern and historical styles. A video by The Drum gives insight into this AI and art partnership, and showed Microsoft’s commitment to applying AI technology to the arts.

Other interesting AI and art experiments done by Microsoft include CaptionBot, a bot that automatically creates captions for any uploaded photos. An article published by Business Insider talked about some fun experiments with CaptionBot by asking it to interpret some famous work of art, and it proved to be a very entertaining “art history class”.

Industry leaders see a bright future for AI. Microsoft considers it to be one of the year's definitive technologies, going so far as to dub 2016 "the year of AI". As programmers make leaps and bounds in AI, arts managers are leveraging aspects of AI technology to redefine the way we experience art. Artificial intelligence might be the key to bridging the gap between art and technology, heralding a new age of creativity. Perhaps art and science are not opposites, but two sides of the same coin.