Recent empirical research unveils, in copyright infringement matters, people assign greater culpability to the allegedly infringing work when they believe the creator is AI.
Co-authored by Joseph Avery, assistant professor at the University of Miami Patti and Allan Herbert Business School, and associate professor Mike Schuster of the University of Georgia, "AI Artists on the Stand: Bias Against Artificial Intelligence-Generated Works in Copyright Law" is the first to examine how AI's presence distorts legal outcomes in copyright infringement lawsuits. The study, published in the UC Irvine Law Review, identifies a larger perceptual bias at play.
"If a human and an AI do the exact same thing, with the same input and output, people still react differently," said Avery. "It's as if our vision has changed."
Avery said that this reaction reveals a deeper puzzle: what we cannot see—the creative process—changes how we perceive what we can see—the resulting work. This perceptual bias, in which people judge an AI-created work differently than a human-created one, leads to attributing more fault to the AI-created work, producing what the researchers term an "AI litigation penalty."
"People are consistently tougher when an AI is involved," Avery said. "They view AI as more culpable and impose greater damages."
The litigation penalty transpired in the empirical study. Participants were shown an original, copyrighted work and asked to evaluate two identical, allegedly infringing works created under the same conditions: one attributed to a human creator and the other to AI. Ultimately, participants judged the AI-created work as less ethical and fair, and of lower quality. And when positioned as mock jury members, they deemed the work by AI as significantly more infringing or plagiaristic than the identical, human-created equivalent.
This AI litigation penalty isn't limited to copyright matters; Avery's forthcoming research shows it also persists in patent and trade secret legal disputes.
What causes the perceptual bias and resulting litigation penalty remains unclear, but central to Avery's research.
"Perhaps we want to reward what feels human, and sometimes that instinct gets tangled up in our legal judgments," Avery said. "I suspect there are manifold reasons, and I plan on discovering them."
This study's findings may serve as a legal cautionary tale to artists who use AI and the companies that hire them. More broadly, Avery suggests the very reason copyright exists could also be threatened.
"The point of copyright is to encourage production and dissemination of creative works," Avery said. "If we start punishing works simply because AI was part of the process, we may risk chilling innovation and limiting what people can imagine."
However, Avery notes that as quickly as AI evolves, so too could this perceptual bias—a hypothesis he is currently researching.
"As we get used to AI, our reactions could flip."
