Dead Poet Talking: Polish radio experiment bares pitfalls of AI


Demski, a presenter who lost his job, in Krakow, Poland. A radio station in Poland fired its on-air talent and brought in AI-generated presenters. An outcry over a chat with a purported Nobel laureate quickly ended that experiment. — The New York Times

KRAKOW, Poland: When a state-funded Polish radio station cancelled a weekly show featuring interviews with theater directors and writers, the host of the programme went quietly, resigned to media industry realities of cost-cutting and shifting tastes away from highbrow culture.

But his resignation turned to fury in late October after his former employer, Off Radio Krakow, aired what it billed as a “unique interview” with an icon of Polish culture, Wislawa Szymborska, winner of the 1996 Nobel Prize for literature.

The terminated radio host, Lukasz Zaleski, said he would have invited Szymborska on his morning show himself, but never did for a simple reason: She died in 2012.

The station used artificial intelligence to generate the recent interview – a dramatic and, to many, outrageous example of technology replacing humans, even dead ones.

Zaleski conceded that the computer-generated version of the poet’s distinctive voice was convincing. “It was very, very good,” he said, but “I went to her funeral, so I know for sure that she is dead.”

The technology-enabled resurrection of the dead poet was part of a novel experiment by Off Radio Krakow, an arm of Poland’s public broadcasting system in the southern city of Krakow. The aim was to test whether AI could revive a moribund local station that had “close to zero” listeners, according to the head of public radio in Krakow.

The station also planned from-the-grave interviews with other dead people, including Jozef Pilsudski, Poland’s leader when it regained its independence in 1918.

Novelty value – and a storm of public outrage – worked to bolster Off Radio Krakow’s audience, which the head of Radio Krakow said grew to 8,000 overnight from just a handful of people after the introduction of three AI-generated Generation Z presenters – Emilia, 20, Jakub, 22, and Alex, 23, each of whom had a computer-generated photograph and biography on the station’s website.

Less welcome than the audience surge, however, has been a barrage of abuse directed at the public broadcasting system and accusations that it was sacrificing humans on the altar of technology.

“I have been turned into a job-killing monster who wants to replace real people with avatars,” said Mariusz Marcin Pulit, editor-in-chief of Radio Krakow and of niche stations operating under its umbrella, like Off Radio Krakow.

He insisted that it was never his intention to replace people with machines, and that his only goal was to revive Off Radio Krakow, make it more appealing to younger listeners and stir debate about AI as Poland’s parliament discusses new legislation to regulate its use.

The technology used to generate the fake interview with Szymborska and other programing, he added, has been widely used: Open AI’s ChatGPT, speech synthesis software developed by ElevenLabs, and the image-generating programs of Leonardo.Ai.

But Pulit’s assurances have done nothing to calm public anger – and alarm that humans are being written out of the script.

Among those outraged by Pulit’s experiment was Jaroslaw Juszkiewicz, a radio journalist whose voice was used for more than a decade to guide drivers using the Polish version of Google Maps. His replacement by a metallic computer-generated voice in 2020 stirred fury on social media, prompting Google to restore him, at least for a time.

He announced this past week that he had been yanked again, lamenting that AI was “sweeping through the world of human voice work like a giant steamroller.” Pulit continued, “And I can, in my own human voice, say, probably for the last time: ‘Smile beautifully and head south.’”

In a Facebook post, he said the use of AI to fake an interview with the dead Nobel Prize winner had left him speechless. “If that is not a breach of journalistic ethics,” he said, “I don’t know what is.”

The National Radio and Television Council, a regulatory body stacked with supporters of Poland’s previous right-wing government, assailed Pulit, who was appointed by a new center-left administration formed in December.

He was “eliminating the human factor” and forcing media to obey “unethical commands and ideas serving, for example, strictly political interests,” a council member, Marzena Paczuska, wrote in a letter to the culture minister.

A member of the government also expressed alarm. The minister of digitalization, Krzysztof Gawkowski, complained on the social media platform X that “although I am a fan of AI development, I believe that certain boundaries are being crossed more and more.” He added: “The widespread use of AI must be done for people, not against them!”

Tired of being accused of wanting to make humans redundant, Pulit recently pulled the plug on his AI experiment.

“We are pioneers, and the fate of pioneers can be difficult,” he said in a recent message to staff members announcing an abrupt termination of AI presenters and their replacement by music created and performed by humans.

Among the AI presenters removed from Off Radio Krakow was Alex Szulc, a nonexistent person who had been presented as a nonbinary progressive “full of social commitment.”

A biography on the station’s website was later rewritten to delete any mention of the presenter’s sexual orientation after LGBTQ+ activists angrily complained that they needed a real person to speak for them, not a computer-generated one.

Also gone is Emilia Nowak, the station’s computer-generated “pop culture expert,” who conducted the “interview” with the dead poet. The station first announced the conversation as if it were a real interview, but later clarified that it had been fabricated by a machine.

Michal Rusinek, head of a foundation that manages the late Nobel Prize winner’s literary estate, said he had given Off Radio Krakow permission to use Szymborska’s voice for the segment because the poet “had a sense of humor and would have found it funny.”

But he said the interview “was horrible” and put words in the poet’s mouth that she would never have used, making her sound “bland,” “naive” and of “no interest whatsoever.” But that, he added, was heartening because “it shows that AI does not yet work” as well as humans. “If the interview had been really good,” he said, “it would be terrifying.”

Felix Simon, author of a report published in February on the effect of AI on journalism, said the Polish experiment had not altered his view that technology “aids news workers rather than replaces them.” For the moment, he added, “there is still reason to believe it will not bring the big jobs wipeout some people fear.”

For the many in Poland who criticize Off Radio Krakow’s flirtation with AI, the station’s use of computer-generated presenters, though now suspended, has highlighted a grave and immediate danger.

An online petition drafted by Zaleski and Mateusz Demski, a fellow presenter who also lost his job, warned that “the case of Off Radio Krakow is an important reminder for the entire industry” and a “dangerous precedent that hits us all.”

The use of AI-generated presenters, the petition warned, “is opening the door to a world in which experienced employees associated for years with the media and people employed in creative industries will be replaced by machines.”

Pulit dismissed that as “fake news,” noting that none of the people who had lost their jobs at the radio were full-time employees.

Zaleski said most of his income had always come from work as a theater director, so he was not particularly upset when he lost his weekly slot on Off Radio Krakow, which paid US$62 per show.

But he said he was appalled at being replaced by a machine-generated substitute. “I was very angry that real, deep talks and real interviews with real people were replaced with something totally fake.” – ©2024 The New York Times Company

Follow us on our official WhatsApp channel for breaking news alerts and key updates!

   

Next In Tech News

Scientists use AI to help track penguins in Antarctica
Windows 10 users will soon have to pay to keep getting security updates
Wrong claims by Musk on US election got 2 billion views on X in 2024, report says
OpenAI in talks with California to become for-profit company, Bloomberg News reports
Meta to extend ban on new political ads after U.S. election
Crypto firms including Robinhood, Kraken launch global stablecoin network
Chipmaker NXP forecasts quarterly revenue below estimates on macroeconomic concerns
Exclusive-Silver Lake, Bain prepare to bid for multi-billion stake in Intel's Altera unit, sources say
Cryptoverse: U.S. election speculators play the prediction markets
Coordinated efforts, shared learning among OIC states essential for digital transformation, says Fahmi

Others Also Read