The Washington Post’s AI podcast became controversial for its repeated mistakes

The chief standards secretary of the Washington Post criticized the widespread errors in the artificial intelligence podcasts of this newspaper and announced that the publication of these podcasts has aroused great concern among journalists.

The podcasts, which were released last week to users of the Washington Post mobile app, were supposed to provide a personalized experience for users, allowing them to choose the topic and host of their choice. But less than 48 hours after publication, the newspaper’s staff and editors realized that these podcasts were full of serious errors; From simple mistakes in pronunciation to distorting the news, creating fake quotes, misattributing information and adding personal opinions that sometimes presented the statements of sources as the official position of the newspaper and threatened the professional credibility of reporters.

Backlash to The Washington Post’s AI podcasts

According to the Semafor report, this situation has caused great anger and frustration among Washington Post reporters and editors. In an internal Slack message, one of the newspaper’s editors wrote: “It’s really surprising that a project like this was allowed to go ahead. “I never imagined that the Washington Post would intentionally distort its reporting and spread these errors to a wide audience.” Senior Secretary of Standards Karen Penciro also described the situation as “frustrating and frustrating” for staff. Other staff members also criticized the lack of adequate safeguards to ensure the accuracy of information, and some admitted that the situation embarrassed the news team.

Readers have also noticed these mistakes. Jane Rosenzweig, a technology reporter on Bluesky, said that Washington Post’s AI podcasts have confused audiences by bringing up topics such as the execution of mentally disabled people without providing sufficient background or explanation. Besides this, there is considerable tension between the news team and the product department of the newspaper; The product department sees these errors as a natural part of testing new features, but journalists see them as a humiliation to their profession and a threat to journalistic standards.

These podcasts, produced in collaboration with Eleven Labs and with voice cloning technology, are part of The Washington Post’s efforts to make extensive use of artificial intelligence under Jeff Bezos’ ownership. Previously, The Washington Post used artificial intelligence for news summaries, accepting articles from non-professional writers, and an “Ask The Post AI” page. However, the experience of AI podcasts showed that without human supervision, content errors and poor editorial quality can jeopardize the professional credibility of journalists. In addition to news mistakes, these podcasts create an unnatural and artificial listening experience for the audience with artificial speech simulations such as “em” and “ah”.