Is the Washington Post's new AI podcast a symbol of the future?: NPR

It's not your mother's podcast – it's not your father's, or anyone else's. washington post's new service, Your Personal Podcast, uses artificial intelligence to customize podcasts for you, blending algorithms like those found in your news feed with the convenience of portable audio.

Podcasts are “automatically personalized based on your reading history.” post The paper's help page says it has an article. Listeners also have some control. You can also change the combination of topics in your podcast or swap out the computer-generated “host” with the click of a button.

The AI ​​podcast quickly made headlines and drew criticism from people who questioned its accuracy and the motives behind it.

Nicholas Quar, critic and staff writer at Vulture; new york magazine who He writes a newsletter about podcasts and says AI Podcasts is one example. post's extensive digital experimentation did not go well.

He says this is one of many technological and digital-oriented experiments they are conducting with the aim of “reaching a larger audience and tapping into new demographics.” These wide-ranging efforts range from generative AI tools for readers to digital publishing platforms. But in this case, “it feels like the core idea of ​​what a news product is is being undermined,” Quah added.

The newspaper emphasizes on its help page that the podcast is in early beta and “not a traditional editorial podcast.”

Bailey Kattleman, Head of Product and Design postcalls this an “AI-powered audio briefing experience.” This is something that listeners will be able to hear back to right away.

“In future releases, they'll be able to actually interact and ask follow-up questions to dig deeper into what they've heard,” Cattleman said in an interview with NPR.

Although it sounds technically sophisticated, there are many questions about the new podcast's accuracy and even its ability to pronounce the name correctly. post Quoting statements from journalists. Semafor reported the errors identified by staff. post, The paper's own position included “inserting explanations such as misattributing or fabricating quotations, and interpreting source quotations.”

The paper's app advises listeners to “verify information” by checking the podcast against the source material.

The Washington Post Guild, which represents newsroom employees and other staffers, told NPR in a statement that “we are concerned about this new product and its rollout,” arguing it would harm public services. postmission and the work of journalists.

The guild cited the newspaper's traditional practice of issuing corrections when articles are incorrect, adding: “Why should we support technology that adheres to a different, lower standard?”

So why? post Will you roll out AI podcasts? And will other news and audio news organizations follow suit?

Below are some questions and answers.

Isn’t AI podcasting already popular?

” post Andrew Deck added to NPR that it's not the first time newspapers have experimented with AI-generated podcasts in the broader news industry.

Deck, who writes about journalism and AI at Harvard's Nieman Lab, points to examples such as the BBC's My Club Daily, an AI-generated soccer podcast that allows users to listen to content related to their favorite clubs. In 2023, “Switzerland's public broadcaster used voice clones of real radio hosts during broadcasts,” he added.

News organizations have also long offered automated capabilities to convert text articles into computer-generated audio.

Even outside of the news industry, AI tools for creating podcasts and other audio have never been more accessible. Some promise to streamline the editing process, while others let you synthesize documents and websites into podcast conversations.

Why would publishers want to experiment with AI podcasts?

“It's cost-effective,” said Gabriel Soto, senior director of research at Edison Research, which tracks the podcast industry. “We have eliminated many of the resources and people needed to produce a podcast: the studio, writers, editors, and hosts themselves.”

And in today's competitive podcasting market, Soto adds, if brands can create successful AI virtual podcasts, it could become valuable intellectual property in the future.

Dec said: postIf the experiment is successful, the paper could “significantly expand and expand its audio journalism offerings without investing the effort normally required to scale.”

In an interview, Cattleman emphasized that the new product is not meant to replace traditional podcasts, saying, “We believe podcasts have a unique and enduring role, and in Post, that's not going away.”

What makes it unique? post AI podcast?

For Deck, the level of customization promised is revolutionary. The ability to customize a podcast to a specific person is “probably beyond what journalism podcast teams can create manually right now,” he says.

In the example, post Once published, listeners can choose audio options with names like “Charlie and Lucy” or “Bert and Ernie.”

Cattleman said her team worked with the idea that there is no “one size fits all” tool for audiences when it comes to AI and journalism.

“Some people want a really straightforward briefing style, and others like a more conversational, more vocal style,” she says.

Quah said the addition of AI podcasts is to make stories more accessible to a wider audience.

In his podcast, he post It seems that they are trying to reach young people who say, “I don't want to read books anymore, I just want to listen to the news.''

A key goal, Cattleman said, is to make podcasts more flexible and appeal to younger listeners on the go.

Overview of the process behind it postIn his AI podcast, Cattleman said: “Everything is washington post Journalism. ”

LLMs (large-scale language models) convert stories into short audio scripts, she says. A second LLM then checks the script for correctness. Cattleman added that audio will narrate the episodes after the final script is pieced together.

Will listeners embrace an AI news podcast?

Edison Research's Soto says one in five podcasters has listened to an AI-narrated podcast.

But when it comes to podcast listeners, “many people prefer the human connection and are receptive to AI tools that help create content, but not running or hosting a podcast,” he added.

The new AI podcast reminded Deck of the highly personalized choices that TikTok and other social media offer users.

“There is a certain degree of familiarity.
And there's definitely a growing comfort with algorithmic curation among younger audiences,” he says.

But while younger audiences tend to be more tech-savvy, many are also thoughtful about authenticity and connection.

“Community is at the heart of why people listen to podcasts,” Soto said.

There's also the idea of ​​host and creator personality that drives engagement on TikTok and other platforms.

“Even if these creators don't talk to their sources, they're building relationships and maybe even trust with their audiences,” Deck says. “This kind of news content is very different from the disembodied banter of an AI podcast host.”

What are the potential drawbacks of AI podcasts?

One of the major potential impacts is the loss of jobs and, for businesses, the loss of talent.

“Automation is kind of killing the entire voice performance industry,” Quah said. “There are people who make a living doing this. They can produce higher quality versions of these recordings,” he added.

There are also concerns that if AI chooses stories and controls how they are presented, it could create echo chambers and omit the context and skepticism that journalists might provide.

“AI-based news personalization tends to be firmly entrenched in delivering what viewers want to hear,” Deck says.

Deck says he would be happy to donate. postLet's take a little look at how the AI ​​podcast will play out. But Deck has major concerns. “You could say that pure white generative AI models are hallucinating.”

And when AI models are wrong, he says, they are often confidently wrong.

Blurring the lines between human and AI voices can also raise issues of trust, which is a critical factor for news organizations.

As Soto says, “What happens when your audience is expecting content from the real you and instead finds an AI?”