The outcry sparked by the Washington Post’s launch of AI-generated podcasts has gone in one ear of the newspaper’s leadership and out the other, which isn’t too dissimilar to what might happen to anyone listening to the slop it’s trying to peddle.
On Monday, the Jeff Bezos-owned publication doubled down on its personalized podcasts push, brushing aside the criticism from readers and its own reporters alike.
“This is how products get built and developed in the digital age: ideation, research, design and prototyping, development, and then Beta,” a WaPo spokesperson told TheWrap. “Only if they prove to be successful for the customer do they then get launched. As stated clear on Your Personal Podcast, it is currently in Beta.”
WaPo launched its “Your Personal Podcast” feature last week, sparking immediate dissent in its ranks. Staffers were furious about the podcast AI inventing and misattributing quotes, Semafor reported Friday, and even sometimes editorializing on stories. Some staffers questioned the tech’s seemingly nonexistent guardrails, Status reported, while another described it as a “total disaster.”
Further underscoring the staggering incompetence on display, followup reporting from Semafor revealed that WaPo had conducted its own tests prior to launching which showed that up to 84 percent of the AI-generated podcasts scripts didn’t meet the newspaper’s standards — and were therefore unpublishable.
There’s a significant disconnect between the company’s newsroom and its product team in charge of the AI rollout. The podcast’s product team sees the errors as a normal part of rolling out a new and still experimental feature, and said it would “iterate through the remaining issues.”
Similar jargon was on display in the paper’s statement to TheWrap, and it’s equal parts telling and jarring that such tech-minded rhetoric is being deployed in the context of journalism. The AI may not be doing the work of an actual journalist, but it’s taking over the role of packaging the news to listeners. Would a human news anchor be given this much leeway, and be allowed to screw up more details than they get right because they’re learning on the job, or in tech parlance, “iterating through it”?
The Post isn’t the only newsroom deploying AI. The New York Times uses it to help generate headlines, and Bloomberg’s website features an AI that summarizes its articles. Many publications have some form of AI chatbot trained on their archives. But WaPo has been particularly AI-evangelistic, signifying its transformation under Jeff Bezos’ ownership. Along with deploying AI summaries and an AI chatbot, it also put forth a plan for letting non-professional writers submit articles written with AI.
The AI podcasts, however, seem to really have struck a nerve with its staff.
“It is truly astonishing that this was allowed to go forward at all,” one WaPo editor fumed on Slack, per Semafor. “Never would I have imagined that the Washington Post would deliberately warp its own journalism and then push these errors out to our audience at scale.”
“If we were serious, we would pull this tool immediately,” the editor added.
More on AI: Grok Is Making Wildly Contradictory Claims About Rob Reiner’s Death