Research Shows How Language Humanizes AI
Intelligent algorithms are used to make paintings, write poems & compose music. Consistent with a study by an international team of researchers from Massachusetts Institute of Technology (MIT) and the Center of Humans & Machines at the Planck Institute for Human Development, whether people perceive Artificial Intelligence (AI) as ingenious creator of art or just another tool used by artists depends on how information about AI art presented. The results published in the journal iScience.
In October 2018, a piece of art by Edmond de Belamie, which was created with the help of an intelligent algorithm was auctioned for $432,500 at the Christie’s Auction House. Consistent with Christie’s auction advertisement, the portrait was created through artificial intelligence (AI). The media described this as the first work of art that was not created by a human but rather autonomously by a machine. The proceeds weren’t given to the machine but rather to the French artists’ collective Obvious. This collective had fed an algorithm with pictures of real-paintings by human painters & trained it to make images autonomously. They then selected a particular picture, printed it, gave it a name & marketed it. However, the programmers who developed the artificial neural networks & algorithms used weren’t mentioned, nor they did receive any of the proceeds from the sale of painting.
“Many people are involved in AI’s art: artists, curators & programmers alike. At the same time, there’s ability, especially in the media to endow AI with human-like characteristics. Consistent with the reports you read, creative AI autonomously creates ingenious works of art. We wanted to understand whether there’s a connection between this humanization of AI & the question of who gets credit for AI art,” Ziv Epstein, a PhD student at MIT Media Lab & first author of the study, explained.
To this end, researchers informed almost 600 participants about how AI art is made & asked who should receive recognition for the work of art. At the same time, they determined the extent to which each participant humanizes AI. The individual’s answers varied greatly. But on average, people who humanized AI and didn’t perceive it merely as a tool also felt that AI should receive recognition for the AI art & not the people involved in the creation process.
When asked which individuals deserve the most recognition in the process of making AI art, recognition was initially given to the artists who provided the learning algorithms with data & trained them. Only then were curators named, followed by the technicians who programmed the algorithms. And finally, the “crowd” (mass of Internet users who produce data material with which AIs are trained) was mentioned. Respondents who humanized the AI gave more recognition to the technicians & the crowd, but proportionally less to the artists. A similar picture emerges when respondents are asked about who is responsible, for instance, when an AI artwork violates copyright. Here too, those who humanized the AIs placed more responsibility on the AIs.
A main finding of the study is that it’s possible to actively-manipulate whether people humanize AIs by changing the language used to report on AI systems in art. The creative process can be often described by explaining the very fact that AI supported only by an artistic collaborator, conceives & creates new works of art. Alternatively, the process can be often described by explaining the very fact that an artist conceives the artwork which the AI executes simple commands given by the artist. The various descriptions changed the degree of humanization and thus also to whom the participants attributed recognition & responsibility for AI art from among the human actors.
“Because AI is increasingly penetrating our society, we’ll have to pay more attention to who is responsible for what’s created with AI. In the end, there are humans behind every AI. this is often particularly relevant when the AI malfunctions & causes damage, for instance, in an accident involving an autonomous vehicle. It’s important to understand that language influences our view of AI and which a humanization of AI results in problems in assigning responsibility,” says Iyad Rahwan, director of the Center for Humans & Machine at the Planck Institute for Human Development & co-author of the study.