The Scoop
The New York Times has begun testing its first generative AI tool for readers, a step the publication hopes will improve its troubled search feature but which has also alarmed some of the paper’s journalists.
In recent weeks, the paper has quietly rolled out a new search page for a small subset of visitors that responds to queries with excerpts and relevant links to Times stories, as well as AI-generated summaries of Times reporting. It’s the news arm’s first test of AI-generated text that is unmediated by a Times journalist or editor.
In a statement to Semafor, Times spokesperson Graham James said the new AI tool was intended to improve reader queries.
“We are always testing new ways for our users to discover and engage with Times journalism,” he said. “This experimental feature is an example of how we are using the latest technology to create a better search experience.”
The paper has experimented with other AI features in recent months across its non-news sections. Wirecutter, the Times’ review site, recently began testing an AI search feature called Wirecutter Finder, which points users to AI-generated summaries and tips based on the site’s recommendations.
In this article:
The View From Times journalists
The new AI-generated summaries have been met with skepticism among some rank-and-file journalists in the Times newsroom.
The media newsletter Breaker reported earlier this month that Times management had informed its union that it would test AI summaries in search. The announcement, which came during a contract bargaining session in July, led to frustrations within the company’s editorial union, which says the Times is violating its own editorial standards around AI use.
The union has proposed a series of rules around AI, including the creation of an employee payment pool in which 22.5% of AI licensing deals for training would be split up among unionized employees. It’s also pushing for stipulations that use of AI must have human oversight, a clause that would render the new search tool impractical.
In a statement to Semafor, Times Guild unit chair and senior staff editor Jim Luttrell said the paper’s current AI policies are “meaningless.”
“Management can change and even violate the company’s policy in any way they want, as they are doing with the summarization tool,” he told Semafor. “Every time AI summaries have been deployed, they have made embarrassing mistakes. That’s why we need strong guardrails on AI in the contract — to protect readers’ trust in our journalism.”
Max’s view
The Times is playing catch-up to search and summarization tools that have become a common feature of many websites, apps, inboxes, and even other news media organizations.
It’s a small step that, according to several Times employees I spoke to last week, is necessary to improve the paper’s internal search tools, which have long lagged its otherwise best-in-class publishing products.
A several-sentence summary of existing Times journalism based solely on the Times’ own work is not the most ambitious implementation of AI in news media. But figuring out how to implement generative AI is a trickier task for the Times in particular, a news institution that is obsessed with accuracy and attention to detail. While the paper regularly makes mistakes and appends corrections, it also works hard to avoid them, with layers of painstaking fact-checking and editing.
Embracing AI poses obvious risks to accuracy, particularly as the Times’ political enemies look to discredit the paper’s journalism. What’s to stop a malicious actor from spending all day trying to trip up the paper’s AI summarization tools to get it to say something stupid?
But the paper’s emphasis on AI tools also could potentially undermine a larger point the Times has made to readers. In an era where deepfakes, bots, and hallucinations have undermined the fabric of reality, news media organizations like the Times have positioned themselves as the remaining arbiters of truth, bulwarks against AI-fueled disinformation. Removing humans from the equation, even in a small way, cuts against the fundamental value of their journalism in the AI era.
Notable
- The Times has made internal AI tools available to its journalists to use in their own work for a while now.
- The Washington Post trialed an AI-generated podcast feature, with less-than-accurate results.