Los Angeles Times introduces AI tags and challenges media trust

Recently, Patrick Soon-Shiong, a billionaire owner of the Los Angeles Times, posted a letter to readers, announcing that the newspaper will use artificial intelligence technology to add a "sound" label to some articles. These articles will be identified as "voice" if they hold "stance" or are written from a "personal perspective". At the same time, AI will generate a set of content called "Insights" that will be displayed in key points at the bottom of the article, including some annotated "different views on the topic".

In the letter, Matsuo said that the "voice" label is not limited to the content of the opinion column, but also includes news comments, criticisms, reviews, etc. If an article holds a position or takes a personal perspective, it may be marked as "voice". He believes that providing more diverse perspectives can help support the media’s journalistic mission and help readers better understand the problems facing the country today.

However, the change was not welcomed by members of the Los Angeles Times union. Matt Hamilton, vice chairman of the union, said that while the union supports some initiatives to help readers differentiate between news reports and comment articles, they do not believe that AI-generating analytics that have not been reviewed by the editorial team will increase the trust in the media.

Shortly after the change was implemented, some suspicious results had appeared. The Guardian noted that at the bottom of a commentary article about the dangers of AI's unregulated use in historical documentaries, the AI ​​tool claims that the article "is generally consistent with the center-left perspective" and suggests that "AI democratizes historical narratives." Furthermore, in a report about California cities electing members of the Klan to serve as city council in the 1920s, an AI-generated view claimed that local historical records sometimes describe the Klan as "a product of the 'white Protestant culture' in response to social changes" rather than a clearly hatred-driven movement that downplays its ideological threat. Although this statement is correct to some extent, it seems to be presented in a clumsy manner, which is somewhat opposite to the subject of the article.

Ideally, if AI tools are used, some editing supervision should be performed to prevent problems such as the Los Angeles Times currently facing. Lack of supervision can lead to various errors, such as MSN's AI News Aggregator's mistaken recommendation of attractions, or Apple's misinterpretation of the BBC title in the latest notification summary.

Although other media organizations also use AI technology in news operations, they are not usually used to generate editorial evaluations. Bloomberg, U.S. Today, Wall Street Journal, The New York Times and The Washington Post are using this technology in different ways.