Can generative artificial intelligence make art? Ted Chiang isn’t convinced.

On March 18, the multi-Hugo-award-winning science fiction author and 2023 TIME100 Most Influential Person in AI lectured at Princeton University, laying out what he described as fundamental incompatibilities between generative AI and works of artistic expression. The talk, which Chiang presented to a packed McCosh Hall lecture room, drew on themes he’d explored in an essay on the same topic published with The New Yorker in August. 

“The selling point of generative AI is that these programs produce far more than you put into them,” said Chiang. “That is precisely what prevents them from being effective tools for artists.”

Art, Chiang argued, is an amalgamation of the series of choices made by an individual artist. By putting in a prompt and letting generative AI do the rest, a person is passing off the work of making necessary artistic choices to the machine. “In essence, they’re advertising a way of making art that is all inspiration and no perspiration,” said Chiang. “I contend that these things cannot be easily separated.” 

Over the course of the talk, Chiang further explored what he described as the tension between art and commerce. Though generative AI may increase the supply of fictional writing or digital images, he said, it cannot increase the demand for them. “Chiang is a voice of reason, tempering the current hype, exposing its capitalist underbelly, and laying out with refreshingly accessible terms the actual capabilities and risks of artificial intelligence today,” said Meredith Martin, the faculty director of the Center for Digital Humanities, which organized the talk with the support of the Princeton Laboratory for Artificial Intelligence, as well as the new Princeton Humanities Initiative and the Princeton Public Library.  

Chiang’s lecture is part of CDH’s ongoing “Humanities for AI” series. The series includes a number of projects, initiatives, and conversations which center humanities approaches and values to the development and use of AI. Previously, Chiang came to Princeton in October as part of the Spencer Trask Lecture Series. In that talk, he laid out his argument against the use of AI as a tool of achieving immortality via the speculative concept of “uploading” human consciousness to machines.

In closing his lecture, Chiang emphasized he does not take an anti-technology stance but is pro-technology, which expands the possibilities for humans to express that which they could not express before. He used the electric guitar as an example, saying though the technology built upon the acoustic guitar, it did not make guitar playing “easier” but instead allowed musicians to create new genres of music.

“If we think of art as a concentrated form of intention, then generative AI is a way of diluting intention,” said Chiang. “Making a lot of choices is hard work, but artistic self-expression requires that you make all the choices.”