In an ironic turn of events, amid widespread alarm over AI’s role in disinformation through image fakery and hallucinatory fabrications the technology has become what many traditional TV news broadcasters see as a vital linchpin to their survival as bastions of journalistic integrity.
Adding to the irony is the fact that some broadcasters are strengthening their newscasting reach by taking advantage of AI’s ability to distort reality by turning monolinguistic on-air personalities into fluent multilinguists. This is accomplished through real-time audio translation platforms that can generate multilingual news, sports and other live commentary by replicating the voices and reshaping lip movements of presenters speaking in one language for simultaneous output in one or more alternative languages.
While this is a bridge too far for some of the enthusiastic AI backers who are taking lead roles guiding station groups’ use of the technology, others see it as just another benefit that, with public transparency, allows them to do a better job reporting the news. One of the latter cases in point is Sinclair, Inc., which has gone further than most station groups by using live AI translation technology from Deeptune to test delivery of news in Spanish over live YouTube feeds from English-speaking newscasters at stations in Baltimore, San Antonio, West Palm Beach and Las Vegas.
What If Viewers Say It’s Okay?
Scott Ehrlich, chief innovation officer, Sinclair
As described by Sinclair chief innovation officer Scott Ehrlich, this step follows Sinclair’s successful use of generative AI last year when the company began using AI-enabled language translation with “Petko Unfiltered,” a Tennis Channel series and in newscasts for use delivering critical information to Spanish populations during hurricanes in southern states. Looking at its successes in those instances, Ehrlich says, “We thought streaming our news with Spanish translation on YouTube would be good for our Spanish-speaking communities, especially with providing coverage of news events directly impacting them in their native language.”
He notes a key question addressed by the test is whether, if the audience is clearly informed that AI is being used with the potential for errors in translation, people will react negatively or say, “‘We understand you’re using AI, and it’s okay.’” Having found strong support among viewers, Ehrlich says Sinclair is now weighing whether to put live translation to use in over-the-air broadcasts.
“We have great confidence in the technology,” he says. “It’s not like it can’t get better, but it’s reasonably mature at this stage.”
E.W. Scripps is another TV broadcaster taking the plunge with live broadcast translation technology. Having already gone the OTA route with translations of NBC affiliate KRIS-TV newscasts in Corpus Chrisi for delivery in Spanish over local Telemundo affiliate KAJA, Scripps “is working to roll out near [real]-time translation systems that would enable our news and sales teams to reach new audiences,” according to a written response to our queries attributed to Christina Hartman, vice president of emerging technology operations, and Keith St. Peter, director of newsroom AI. Without revealing details, they say that “by leveraging AI, it’s now possible to deliver content and advertising tailored for specific audiences that we would not have had the resources to produce even a few months ago.”
Other newsroom apps like script conversions between digital and broadcast content and summarization of lengthy meeting and other reports “are designed to free up our journalists to do more in-depth reporting from the field,” they note. Overall, they add, “Using our secure and proprietary platform, Engine Room, we have internally developed more than 50 applications that help our teams deliver accurate and human fact-checked content faster, serve more clients and overall operate more efficiently.”
No Idea Where It’s Going
Jeff Zellmer, EVP, digital operations, Fox Television Stations
Live language translation technology is getting a lot of attention across the broadcast industry, says Jeff Zellmer, executive vice president of digital operations at Fox Television Stations, who is quick to say his company is no fan of the practice when it’s applied whether people like it or not. “We did test it but didn’t see anything significant to justify moving forward,” Zellmer says, although Fox is considering creating a “language feature in an app for people to choose rather than deciding for them.”
While acknowledging he, like everyone else, has “no idea” of where AI is going, Zellmer voices the widely held perspective that “this is an area of growth we need to understand and be involved in.” Fox has no compunctions about trying things out, he adds, noting that about a year and a half ago the company even went so far as to try using AI avatars in a couple of overnight newscasts going out on all platforms, which quickly proved to be a bad idea. “It shows we’re willing to experiment, but if it doesn’t help viewers with better experiences, it’s not right for us,” Zellmer says.
More generally, Fox is taking advantage of AI wherever it makes sense as a way to add efficiencies to workflows, such as vendor solutions that employ AI to expand content metadata and expedite searches across massive archives, which is “a huge area that can help us.” Fox is also using AI to help developers with coding new apps.
For Fox and everyone else we’ve spoken to, preserving jobs remains a top priority, but the long view casts a different light on the issue. “Honestly, if we aren’t working on learning how to use AI every day, we won’t be ready to use it for something that’s absolutely critical for us,” Zellmer says. “Not taking advantage of that could have a big impact on jobs and our company’s fortunes over time.”
In growing numbers, station owners across the country are implementing internal expertise and positions in the org chart aimed at vetting solutions and coming up with new ways to use the technology in whatever modes are deemed to be responsible by the people in charge. For example, in Sinclair’s case, everything is unfolding under the watchful gaze of the Sinclair AI Forum, an internal, interdepartmental AI workgroup the company created to facilitate knowledge sharing, governance, compliance, and strategic guidance.
“We have to be really sharp about what we do with the efficiencies we find,” Ehrlich says. “It’s not just about the availability of functionality; it’s about how people actually use it. So it’s one step at a time.”
Reliability All Over the Lot
Lee Zurik, SVP, news strategy & innovation, Gray Media.
That’s the same attitude shaping the approach to AI taken by Gray Media, another top-tier station group owner. “We’ve been very cautious with AI as a company, because viewer trust is our North Star,” says Lee Zurik, senior vice president of news strategy and innovation at Gray Media. “We’ve been smart and methodical in how we’ve looked at AI, and now we’re ready to roll out some applications.”
No matter how it’s used, AI reliability is a major issue, which means that where things go from here and how fast largely depends on where the rubber meets the road as stations find reliable suppliers of whatever they’re looking for. AI “is an extremely exciting opportunity,” Zurik says, but the viability of products touting AI support “is all over the lot.” Talking with “dozens and dozens of vendors,” Zurik’s team is finding “some products are really good and can help us, but others still have work to do on accuracy and other issues.”
When it comes to journalistic applications, he adds, “The way we look at it is, AI can’t replace humans on the street doing reporting, but if there’s something it can do from a workflow basis to speed up processes, free up people’s time and allow us to do more things, that’s what we’re looking for.” He cites things like “helping investigative reporters to dig through thousands of pages of documents or converting broadcast script to digital script.”
Preserving a Pillar of Democracy with Integrity
So where will a growing stampede toward cutting costs in a competitive marketplace lead? As worries about the downsides, including the impact on jobs, give way to pragmatic recognition of the risks that come with being left behind, the likelihood grows that once an AI application lends a competitive edge to its users, other stations groups will jump on board.
Nowhere is what unfolds more critical to setting boundaries that matter than how AI impacts the standing of TV news as a Fourth Estate pillar of democracy. The outcome will come down to how the industry as a whole resolves the tension between keeping broadcast news alive and maintaining its integrity.
“I came here because I believe in what journalism means to the communities we serve,” says Michael Newman, who leads one of the industry’s new AI-oriented development teams as director of transformation at Graham Media Group. “If we do AI the right way, it allows us to amplify our journalistic message and community voice and helps us to bring journalism into the modern world in ways we desperately need.”
Nobody is more transparent about how it uses AI than Detroit-based Graham Media, which operates seven large broadcast network-affiliated stations in four states. Uniformly, the stations’ newscast websites carry this statement at the bottom of their home pages: “In our commitment to covering our communities with innovation and excellence, we incorporate Artificial Intelligence (AI) technologies to enhance our news gathering, reporting, and presentation processes.”
The text goes on to refer readers to a policy statement that lets people know that while the stations don’t “anthropomorphize AI,” they are putting AI to use in “generating news summaries, identifying trending topics, and curating personalized content for our readers” and “to produce accurate and immediate coverage for specific types of data-driven stories, such as financial reports or sports results.”
“I think we’re trying to be very transparent to our communities,” Newman says. “We are actively experimenting and have several production workflows that leverage AI.”
Graham Media, like other station groups, is especially focused on taking advantage of the public’s reliance on digital outlets. “Our station’s news teams do amazing work that doesn’t always get the reach it deserves,” Newman says.
While TV news still serves vast audiences, “it doesn’t have the relevance, the scale and necessarily the best business model when it gets into the digital space,” he adds. The company created the Graham Digital unit he oversees to address the issues on both the news gathering and monetization sides of the business.
It’s a massive undertaking where a primary goal is to take advantage of the capacity for more news coverage in the digital space to bring more information that matters to local communities. AI helps to enable expansion of coverage cost effectively with more news gathering by “people on the ground providing first-hand coverage out in the field” abetted by automated transformation of their live reports into textual output for the digital feeds.
For example, Newman notes, with stations staffed by meteorologists delivering extensive local weather reports, Graham Digital has worked with stations to develop a process that “takes long-form videos and runs them through AI” using rules that “turn broadcast speak into written content.” By taking that responsibility off the backs of the weather report staffs, “we’re actually creating articles that are better in the first draft than they were when people were turning them in between broadcasts.”
These are only first drafts, which must be parsed by the meteorologists before release. But there’s a significant amount of time saved on the order of a half hour per report, which “is really a boon to us,” he says.
Notwithstanding Graham Media’s eagerness to maximize the AI benefits, Newman isn’t signing off on live broadcast translation. To preserve “the journalistic integrity of the team, I very much want a trained editor who’s fluent in the language to be editing content before it goes out,” he says. Making sure “all AI output is reviewed by a human set of eyes before it’s published is very important to us.”
The Wild West Revisited
But Newman acknowledges Sinclair’s move to live translation in newscasts is “really cool and a great example of how AI and original reporting by Sinclair can be accessed by more people.” Whether or not the industry sets guidelines to sort out whether such things are acceptable remains to be seen.
While the industry hasn’t formalized standards around AI usage, Sinclair’s Ehrlich says there are groups moving the industry in that direction. One cited by Ehrlich is the Coalition for Content Provenance and Authenticity (C2PA), an ad hoc group charged in Sinclair’s words with “working to protect content provenance and address misinformation.” Ehrlich says this is “one of the most important with important industry players,” including Sinclair, which is also a member of the Content Authenticity Initiative (CAI).
“We have to develop consumer trust,” he says. “What we saw when the world went digital is still a consistent thread. We have to respect privacy and intellectual property. It’s table stakes.”
With things still a long way from being set in digital generally, let alone AI, “moving responsibly now falls to companies like ours,” he adds. “It’s too soon to expect industry-wide standards. By necessity it’s still formative, because the technology is still formative. It’s hard to standardize when you’re dealing with a moving target.”
But, as Ehrlich also notes, “It’s not like we’ll get to the end of development of AI tools. There is no end,” although he expects there’ll come a time when the environment lends itself to some level of standards setting. Meanwhile, all we can hope for in a scenario akin to the Wild West battle between fence-building sheep herders and open-range cowboys is that, eventually, there’ll be some fences reining everybody in.
Adding to the irony is the fact that some broadcasters are strengthening their newscasting reach by taking advantage of AI’s ability to distort reality by turning monolinguistic on-air personalities into fluent multilinguists. This is accomplished through real-time audio translation platforms that can generate multilingual news, sports and other live commentary by replicating the voices and reshaping lip movements of presenters speaking in one language for simultaneous output in one or more alternative languages.
While this is a bridge too far for some of the enthusiastic AI backers who are taking lead roles guiding station groups’ use of the technology, others see it as just another benefit that, with public transparency, allows them to do a better job reporting the news. One of the latter cases in point is Sinclair, Inc., which has gone further than most station groups by using live AI translation technology from Deeptune to test delivery of news in Spanish over live YouTube feeds from English-speaking newscasters at stations in Baltimore, San Antonio, West Palm Beach and Las Vegas.
He notes a key question addressed by the test is whether, if the audience is clearly informed that AI is being used with the potential for errors in translation, people will react negatively or say, “‘We understand you’re using AI, and it’s okay.’” Having found strong support among viewers, Ehrlich says Sinclair is now weighing whether to put live translation to use in over-the-air broadcasts.
“We have great confidence in the technology,” he says. “It’s not like it can’t get better, but it’s reasonably mature at this stage.”
E.W. Scripps is another TV broadcaster taking the plunge with live broadcast translation technology. Having already gone the OTA route with translations of NBC affiliate KRIS-TV newscasts in Corpus Chrisi for delivery in Spanish over local Telemundo affiliate KAJA, Scripps “is working to roll out near [real]-time translation systems that would enable our news and sales teams to reach new audiences,” according to a written response to our queries attributed to Christina Hartman, vice president of emerging technology operations, and Keith St. Peter, director of newsroom AI. Without revealing details, they say that “by leveraging AI, it’s now possible to deliver content and advertising tailored for specific audiences that we would not have had the resources to produce even a few months ago.”
Other newsroom apps like script conversions between digital and broadcast content and summarization of lengthy meeting and other reports “are designed to free up our journalists to do more in-depth reporting from the field,” they note. Overall, they add, “Using our secure and proprietary platform, Engine Room, we have internally developed more than 50 applications that help our teams deliver accurate and human fact-checked content faster, serve more clients and overall operate more efficiently.”
While acknowledging he, like everyone else, has “no idea” of where AI is going, Zellmer voices the widely held perspective that “this is an area of growth we need to understand and be involved in.” Fox has no compunctions about trying things out, he adds, noting that about a year and a half ago the company even went so far as to try using AI avatars in a couple of overnight newscasts going out on all platforms, which quickly proved to be a bad idea. “It shows we’re willing to experiment, but if it doesn’t help viewers with better experiences, it’s not right for us,” Zellmer says.
More generally, Fox is taking advantage of AI wherever it makes sense as a way to add efficiencies to workflows, such as vendor solutions that employ AI to expand content metadata and expedite searches across massive archives, which is “a huge area that can help us.” Fox is also using AI to help developers with coding new apps.
For Fox and everyone else we’ve spoken to, preserving jobs remains a top priority, but the long view casts a different light on the issue. “Honestly, if we aren’t working on learning how to use AI every day, we won’t be ready to use it for something that’s absolutely critical for us,” Zellmer says. “Not taking advantage of that could have a big impact on jobs and our company’s fortunes over time.”
In growing numbers, station owners across the country are implementing internal expertise and positions in the org chart aimed at vetting solutions and coming up with new ways to use the technology in whatever modes are deemed to be responsible by the people in charge. For example, in Sinclair’s case, everything is unfolding under the watchful gaze of the Sinclair AI Forum, an internal, interdepartmental AI workgroup the company created to facilitate knowledge sharing, governance, compliance, and strategic guidance.
“We have to be really sharp about what we do with the efficiencies we find,” Ehrlich says. “It’s not just about the availability of functionality; it’s about how people actually use it. So it’s one step at a time.”
No matter how it’s used, AI reliability is a major issue, which means that where things go from here and how fast largely depends on where the rubber meets the road as stations find reliable suppliers of whatever they’re looking for. AI “is an extremely exciting opportunity,” Zurik says, but the viability of products touting AI support “is all over the lot.” Talking with “dozens and dozens of vendors,” Zurik’s team is finding “some products are really good and can help us, but others still have work to do on accuracy and other issues.”
When it comes to journalistic applications, he adds, “The way we look at it is, AI can’t replace humans on the street doing reporting, but if there’s something it can do from a workflow basis to speed up processes, free up people’s time and allow us to do more things, that’s what we’re looking for.” He cites things like “helping investigative reporters to dig through thousands of pages of documents or converting broadcast script to digital script.”
So where will a growing stampede toward cutting costs in a competitive marketplace lead? As worries about the downsides, including the impact on jobs, give way to pragmatic recognition of the risks that come with being left behind, the likelihood grows that once an AI application lends a competitive edge to its users, other stations groups will jump on board.
Nowhere is what unfolds more critical to setting boundaries that matter than how AI impacts the standing of TV news as a Fourth Estate pillar of democracy. The outcome will come down to how the industry as a whole resolves the tension between keeping broadcast news alive and maintaining its integrity.
“I came here because I believe in what journalism means to the communities we serve,” says Michael Newman, who leads one of the industry’s new AI-oriented development teams as director of transformation at Graham Media Group. “If we do AI the right way, it allows us to amplify our journalistic message and community voice and helps us to bring journalism into the modern world in ways we desperately need.”
Nobody is more transparent about how it uses AI than Detroit-based Graham Media, which operates seven large broadcast network-affiliated stations in four states. Uniformly, the stations’ newscast websites carry this statement at the bottom of their home pages: “In our commitment to covering our communities with innovation and excellence, we incorporate Artificial Intelligence (AI) technologies to enhance our news gathering, reporting, and presentation processes.”
The text goes on to refer readers to a policy statement that lets people know that while the stations don’t “anthropomorphize AI,” they are putting AI to use in “generating news summaries, identifying trending topics, and curating personalized content for our readers” and “to produce accurate and immediate coverage for specific types of data-driven stories, such as financial reports or sports results.”
“I think we’re trying to be very transparent to our communities,” Newman says. “We are actively experimenting and have several production workflows that leverage AI.”
Graham Media, like other station groups, is especially focused on taking advantage of the public’s reliance on digital outlets. “Our station’s news teams do amazing work that doesn’t always get the reach it deserves,” Newman says.
While TV news still serves vast audiences, “it doesn’t have the relevance, the scale and necessarily the best business model when it gets into the digital space,” he adds. The company created the Graham Digital unit he oversees to address the issues on both the news gathering and monetization sides of the business.
It’s a massive undertaking where a primary goal is to take advantage of the capacity for more news coverage in the digital space to bring more information that matters to local communities. AI helps to enable expansion of coverage cost effectively with more news gathering by “people on the ground providing first-hand coverage out in the field” abetted by automated transformation of their live reports into textual output for the digital feeds.
For example, Newman notes, with stations staffed by meteorologists delivering extensive local weather reports, Graham Digital has worked with stations to develop a process that “takes long-form videos and runs them through AI” using rules that “turn broadcast speak into written content.” By taking that responsibility off the backs of the weather report staffs, “we’re actually creating articles that are better in the first draft than they were when people were turning them in between broadcasts.”
These are only first drafts, which must be parsed by the meteorologists before release. But there’s a significant amount of time saved on the order of a half hour per report, which “is really a boon to us,” he says.
Notwithstanding Graham Media’s eagerness to maximize the AI benefits, Newman isn’t signing off on live broadcast translation. To preserve “the journalistic integrity of the team, I very much want a trained editor who’s fluent in the language to be editing content before it goes out,” he says. Making sure “all AI output is reviewed by a human set of eyes before it’s published is very important to us.”
But Newman acknowledges Sinclair’s move to live translation in newscasts is “really cool and a great example of how AI and original reporting by Sinclair can be accessed by more people.” Whether or not the industry sets guidelines to sort out whether such things are acceptable remains to be seen.
While the industry hasn’t formalized standards around AI usage, Sinclair’s Ehrlich says there are groups moving the industry in that direction. One cited by Ehrlich is the Coalition for Content Provenance and Authenticity (C2PA), an ad hoc group charged in Sinclair’s words with “working to protect content provenance and address misinformation.” Ehrlich says this is “one of the most important with important industry players,” including Sinclair, which is also a member of the Content Authenticity Initiative (CAI).
“We have to develop consumer trust,” he says. “What we saw when the world went digital is still a consistent thread. We have to respect privacy and intellectual property. It’s table stakes.”
With things still a long way from being set in digital generally, let alone AI, “moving responsibly now falls to companies like ours,” he adds. “It’s too soon to expect industry-wide standards. By necessity it’s still formative, because the technology is still formative. It’s hard to standardize when you’re dealing with a moving target.”
But, as Ehrlich also notes, “It’s not like we’ll get to the end of development of AI tools. There is no end,” although he expects there’ll come a time when the environment lends itself to some level of standards setting. Meanwhile, all we can hope for in a scenario akin to the Wild West battle between fence-building sheep herders and open-range cowboys is that, eventually, there’ll be some fences reining everybody in.