Artificial Intelligence Trained to Draw Inspiration From Images, Not Copy Them

May 17, 2024 • by Karen Davidson

Researchers are using corrupted data to help generative AI models avoid the misuse of images under copyright.

Three rows of similarly themed illustrations—earnest dogs, scientist pandas and robot graffiti—differ in each of five iterations per row.

A new framework, led by a University of Texas at Austin research team, helps a type of generative artificial intelligence that creates images to avoid replicating existing works. Credit: The University of Texas at Austin


Share


A bar graph of blue bars, with a growing trend from left to right

School of Computing

The Long View: New Study Explores a Decade of AI Going Mainstream

A Longhorn silhouette is marked up with calculations and illustrated robots are adjacent

UT News

2 UT Faculty Selected for Inaugural Cohort of Navy Research Fellows

Peter Stone has been tapped by Sony Corp. to head up the U.S. branch of its new global artificial intelligence research division, called Sony AI.

School of Computing

Getting to Know Peter Stone: Curiosity, AI and the Future of Computing