Here’s How Four Major Newsrooms Are Using AI
AI is everywhere these days , but when you see an article online from a source that has historically been considered reliable, it’s reasonable to assume that AI isn’t involved. But at major news outlets across the country, that reality is becoming increasingly unclear.
While not every use of AI in a newsroom is as obvious as AI simply scribbling a message onto a blank page (though that does happen, too ), it’s important to know how the sausage is made when it comes to the information that shapes your worldview.
Here are just a few news outlets that have begun openly using AI in their processes so you can stay informed about exactly where the information you read comes from.
New York Times
Earlier this week, Semafor published an article about new artificial intelligence tools that New York Times executives are reportedly encouraging employees to try. While the publication’s legal division is currently embroiled in a lawsuit against OpenAI and Microsoft for copyright infringement, it appears that the site’s editorial and product divisions are ready to embrace AI in the newsroom, at least through communication from the top.
According to Semaphore’s post, NYT management is currently providing AI training to its journalists, introducing an internal AI tool called “Echo,” and endorsing the use of external AI tools, including Google Vertex AI, several Amazon AI products, and, ironically, Microsoft’s Copilot and OpenAI tool, which does not support ChatGPT.
Not all of them will be used for stories on the website—there’s more to the New York Times than what you see on its front page—but Semaphore said journalists are encouraged to “use these artificial intelligence tools” for tasks such as moderately polishing content or coming up with questions to ask during interviews.
“Generative AI can help our journalists uncover the truth and help more people understand the world,” read the company’s AI editorial recommendations , which are published on its website. “We don’t see this technology as some magic solution, but as a powerful tool.”
Some guardrails appear to still exist for this purpose. Semaphore says the NYT has warned staff not to use AI to write or substantively revise articles, and noted that using AI could potentially violate copyrights or inadvertently reveal sources.
Still, with internal communications suggesting writers use AI to create headlines and draft copy for social media, it’s worth keeping hallucinations in mind the next time you see a shady NYT story go viral on social media.
As for journalists, Semaphore says some remain skeptical, fearing that AI in the newsroom could cause “laziness” and reduce accuracy or creativity.
Quartz
First of all, a disclaimer. Quartz is currently owned by G/O Media, which also owned Lifehacker before its sale to Ziff Davis in 2023.
That said, G/O Media has since become a major proponent of artificial intelligence in newsrooms , with business news publication Quartz being its biggest experiment.
Scroll a little further into Quartz’s byline and you’ll find posts attributed to the Quartz Intelligence newsroom , which appears to be lifting all of the New York Times’ restrictions. Here, the AI ”writer” has been quietly creating income reports for several months, and has recently started blogging more generally as well. These include stories about the potential value of Bitcoin or how to delete your Meta social media accounts , but as you’d expect, each has issues you should be aware of.
To its credit, Quartz makes no bones about the fact that these stories are AI-generated, and the AI cites its sources, but there doesn’t seem to be any human oversight to address any problems that might arise from this.
For example, the Quartz article on how to delete your Meta-owned social media accounts appears to be a simple regurgitation of the TechCrunch story (which did not give permission to summarize it, for what it’s worth), but with clear instructions replaced by what the TechCrunch writer with the recordings calls “vague” gestures in the right direction. Speaking about other stories written by Quartz AI, the same writer also said, “My editor would never allow me to publish something so sloppy.”
At the same time, it is likely that you will read this story, hoping for useful advice, but will leave disappointed and even more confused than when you arrived. But the chances of this happening will only increase if you come across another story from a less reliable source than TechCrunch.
As Futurism notes, Intelligence Newsroom frequently cites a site called Devdiscourse, which itself has all the hallmarks of an AI-powered content farm. When robots quote robots, I have to ask: why not just go to ChatGPT and suggest it yourself? Even if I think the AI news is newsworthy, it’s unclear to me what G/O’s efforts add to it.
G/O places a disclaimer on all AI-powered Quartz stories, saying the company is in the “first phase of a new reporting pilot,” but since efforts to bring AI to the company began in 2023 , it’s unclear whether it will ever figure out what the experiment is supposed to deliver.
In the meantime, keep an eye on all the Quartz writers, and if you get a whiff of AI, maybe consider giving you the sources the bot mixes for you instead. It seems that when you click on this attractive title, you are actually playing roulette.
AP
While the New York Times is just starting to dip its toes into artificial intelligence and Quartz has already become a complete success, AP’s use of artificial intelligence seems to fall somewhere in the middle. The agency proudly states on its website that it uses AI for translation, transcription, headlines, research, and even some automated articles, but general blogging remains in human hands.
“Our goal is to give people a good way to understand how we can do small experiments but still be safe,” AP Vice President for News Standards and Inclusion Amanda Barrett said in 2023, when the group first published its recommendations on artificial intelligence .
What’s most likely to impact you is the site’s use of Wordsmith, an artificial intelligence program that specializes in summarizing content such as sports scores, weather forecasts and, as with Quartz, earnings. AP has been using a version of this program since 2014 , so it’s not entirely new, but it’s worth knowing that stories on these topics without a specific author are likely to have come from a bot. But aside from those stories, the only other apparent AP use of content written directly by artificial intelligence was an experiment with reporting public safety incidents in a particular Minnesota newspaper .
Otherwise, the only time you’re likely to see direct AI content in an AP story is during its recap pilot: for example, the AI might place a blurb underneath reporting on Election Day to say which positions are up for grabs.
Again, news headlines and research are still admittedly driven by artificial intelligence, so it’s important to be extra careful when something doesn’t pass your sniff test, but it looks like humans are still king here, at least for now.
Washington Post
The Washington Post’s use of AI is unique because it does not directly influence content. Rather, it is an advanced search engine for readers who are looking for it. The bot, called ” Ask the Post AI, ” answers questions, provides a short, AI-generated answer based on Washington Post content, and then lists relevant articles below, Google-style.
“Answers are generated by artificial intelligence based on published reports,” the bot warns when you ask a question. “Please check by reading the articles provided.”
The responses seem to tend towards a measured tone, relying on quotes from noteworthy opinion sources, and when I asked the bot about President Trump’s changing stance on TikTok or Elon Musk’s Tesla story, I got truthful answers back, although the first was a long paragraph and the second just a sentence. I guess how detailed your answers are depends on how much the article covers a particular topic.
The Washington Post does not specify which AI technologies have the capabilities. Ask Post AI, but generally it won’t bother you unless you go looking for them. Judging by the common warnings about its use, it appears to be intended more as a place to start research than as a direct replacement for news.