Artificial Intelligence creates art
This study examines who is most recognized in the creation of AI art and what role humanization plays in this process.
Intelligent algorithms can be used to create art, poetry, and music. A study done by an international team of researchers at the Massachusetts Institute of Technology (MIT) and the Max Planck Institute for Human Development’s Center of Humans and Machines, found that people view artificial intelligence (AI), as either the creator of art or a tool used by artists. It all depends on how the information is presented about AI art. These results were published in iScience.
Artificial intelligence makes it easy to see the obvious. It was sold by Christies Auction House for over 430,000 US Dollars in 2018. The creator of the painting has been identified through a scientific investigation.
A work of art created by Edmond de Belamie with the aid of an intelligent algorithm was sold at Christie’s Auction House for 432,500 USD in October 2018. According to Christie’s auction advertisement the portrait was created using artificial intelligence (AI). This was often described by the media as the first piece of art that was not created by humans but by an autonomous machine. The machine was not paid for the proceeds, but Obvious, a collective of French artists. The collective fed the algorithm pictures of paintings created by humans and taught it how to create images on its own. The group then printed a particular picture, gave it a name and promoted it. The programmers of artificial neural networks and algorithms were not mentioned. They also did not receive any proceeds from the sale.
AI art involves many people: programmers, curators, artists and curators alike. There is also a tendency, especially in the media, to give AI human-like characteristics. The reports that you have read indicate that creative AI creates original works of art by itself. Ziv Epstein (a PhD student at MIT Media Lab, and first author of this study), explained that we wanted to see if there was a link between humanization of AI, and who gets credit for AI art.
Artists deserve recognition
The researchers interviewed 600 participants to learn how AI art was created, and then asked them who should be recognized for their work. They also determined how much each participant humanizes AIs. Each person’s answers were different. However, on average, AI was perceived as an art form by people who were more humane than the AI tool.
When asked who deserves the most recognition for creating AI art, the first person to be recognized was the artist who created the algorithms and trained them. The curators were then followed by the technicians who programed the algorithms. The “crowd”, i.e.,. The “crowd” (i.e., the large number of Internet users who create the data material with AIs is mentioned). Respondents who humanized AI gave greater recognition to the technicians and to the crowd, but less to artists. Similar results are seen when respondents are asked who is responsible for AI artworks that violate copyright. The AIs were also more responsible for this when humans were involved.
AI is influenced by language
One key finding from the study was that people can actively control whether AIs are humanized by changing the language used for reporting on AI systems in art. It is possible to describe the creative process by explaining that AI creates and supports new art works by working alone. The process can also be explained by explaining that AI is an artist who creates the artwork, and that the AI follows simple commands from the artist. These descriptions varied in their humanization, and so participants were able to determine who was responsible for AI art recognition and accountability.
“AI is rapidly affecting our society. We will need to pay closer attention to who is responsible. Every AI is ultimately a product of humans. This is especially true when an AI malfunctions or causes damage, such as in an accident involving an auto-piloted vehicle. It is important to recognize that AI’s perceptions are influenced by language. This is why it is so difficult to assign responsibility to AI,” says Iyad Rahwan (director of the Center for Humans and Machine, Max Planck Institute for Human Development) and co-author of this study.
