LA Times’ AI Bias Analysis Faces Early Challenges

Billionaire Los Angeles Times owner Patrick Soon-Shiong announced a new AI-driven initiative to label articles that take a stance or have a personal perspective with a “Voices” tag. These articles may also receive AI-generated “Insights” at the bottom, including alternative viewpoints.

Soon-Shiong explained in a letter to readers that “Voices” applies beyond the Opinion section, covering news commentary, reviews, and other pieces with a subjective tone. He believes this feature will support journalistic integrity by offering diverse perspectives.

However, the move has been met with criticism from the LA Times Guild. Vice Chair Matt Hamilton stated that while the union supports efforts to distinguish opinion from reporting, using AI to generate analysis without editorial oversight is unlikely to build trust in the media.

Concerns about accuracy arose almost immediately. The Guardian highlighted a March 1st opinion piece on AI’s role in historical documentaries. The AI tool classified it as “Center Left” and suggested that AI “democratizes historical storytelling.”

Another example involved a February 25th article on California cities electing KKK members in the 1920s. The AI-generated insights included a now-removed point suggesting the Klan was a reaction to societal changes rather than an explicitly hate-driven movement. This framing was widely criticized.

The LA Times’ experience reflects broader concerns about AI’s role in journalism. Poor oversight has led to past AI blunders, such as MSN recommending an Ottawa food bank as a tourist destination and Apple Intelligence misrepresenting a BBC headline.

Major news outlets, including Bloomberg, The Wall Street Journal, and The New York Times, use AI, but typically not for editorial assessments. The LA Times’ approach underscores the risks of automated content analysis without human oversight.