last week, when Washington Post announced A new service that uses AI to generate podcasts immediately warned readers to prepare for disaster.
As expected, the service – billed as “your personal podcast” – proved to be a flawed disaster from the start. As semaphore informed On Friday, several staffers at the embattled newspaper were angry over the AI inventing and misrepresenting quotes and misinterpreting basic facts, some of the hugely well-known and documented shortcomings of large language models.
Then the story got even worse. In follow-up reporting over the weekend, semaphore informed He WaPo The leadership had launched the controversial feature despite knowing that it had serious flaws. Internal tests found that the AI was producing podcasts full of errors and biases, with 68 to 84 percent of scripts deemed unscriptable by evaluators across three rounds of testing.
Despite dire results, the company’s product review team insisted on releasing the half-baked feature, arguing that it would “replicate remaining issues.”
This is a perfect example of the troubled relationship between generative AI technology and journalism, from news writers publishing fake, AI-generated articles to Google discouraging users from visiting trusted news outlets by offering error-laden AI summaries.
This interaction not only spreads misinformation and disinformation, but experts warn that it can further destroying trust Among the once prestigious publications, including WaPo,
A spokesman for the newspaper said semaphore The feature is “currently in beta” and new features are “only launched if they prove successful for customers.” In other words, there’s still a chance WaPo Can abandon this idea.
But it’s likely the Jeff Bezos-owned newspaper will continue to slim down its journalism in the near future as part of a broader effort to cut costs. earlier this year, WaPo launched an AI chatbot that was found making a lot of mistakesOne AI writing coach named AmberThe purpose of which is to help in writing opinions, work on this is also going on.
Meanwhile, the newspaper’s frustrated staff are forced to watch as chaos spreads.
“It’s really surprising that this was allowed to go ahead,” A. WaPo the editor wrote in a Slack message following the release of the latest AI Podcast feature, as quoted semaphore“I never imagined this Washington Post Will knowingly distort its journalism and then broadcast these errors to our audience at large.
The editor said, “If we were serious, we would remove this tool immediately.”
“This is a complete disaster,” another employee wrote in a separate Slack message. recipient Situation“I think the newsroom is embarrassed,”
WaPo Standards chief Karen Pensiero called the situation “disappointing for all of us” in an internal message shared semaphore,
A Washington Post guild union representative told npr A statement said the organization is “concerned about this new product and its rollout,” adding that it undermines the mission of the publication.
Journalists from other outlets agreed with that overall sentiment. the fact that WaPoThe product team was well aware that “Your Personal Podcast” was seriously flawed and that launching it to the public was harmful, to say the least.
“Concluding that people don’t need their newspapers to be accurate is professional self-harm done for no particular reason other than to justify layoffs and satisfy AI lords,” toronto star Columnist Justin Ling wrote in a bluesky post,
More WaPo, The Washington Post’s AI-generated podcasts are already a disaster full of errors
