This is a good example of an AI use case for a task that likely wouldn't be done otherwise (listening to loads of manosphere podcasts and summarizing them every week forever). There's a job no one...
This is a good example of an AI use case for a task that likely wouldn't be done otherwise (listening to loads of manosphere podcasts and summarizing them every week forever). There's a job no one at the NYT wants.
It replaces a lot of human hours but they're hours that likely wouldn't have been spent in the first place because it's not worth the resources (or pain). No doubt they have summaries setup for other "spaces" too.
Really smart use of LLM tools. This isn't the first time I've been surprised by the NYT's tech savvy
Yes very much so. There's so many terabytes of media generated every day it's not possible to consume even a fraction, having machine summary that one can then verify (very important) is a great...
Yes very much so. There's so many terabytes of media generated every day it's not possible to consume even a fraction, having machine summary that one can then verify (very important) is a great investigative tool
As with the Manosphere Report, Cheatsheet is rooted in a philosophy that creating new text and images for publication is not the most effective use case for generative AI in a newsroom like the Times. Rather, Seward sees the technology as a way to amplify the newsroom’s existing investigative power
I hope that it can have broader usage, and perhaps even be a way to break through Echo chambers: if it's not painful and heated and time consuming to hear that the other side is saying, we might be able to hear a bit more of it
For example, can you imagine how different an election prediction could be if think tanks and media used this for a pulse check instead of relying on polls, especially for views people know is kinda sketchy? I have no idea how Cheatsheet is weighted, but I imagine number of views/listens and engagement is part of it. Tracking what's controversial even within an echo chamber could be so important for a party trying to come up with something that would resonate
That's a good point, finding out what the demographic is talking about without the mental and emotional load of wading through the toxicity could be useful outside of journalism too. I've been...
if it's not painful and heated and time consuming to hear that the other side is saying
That's a good point, finding out what the demographic is talking about without the mental and emotional load of wading through the toxicity could be useful outside of journalism too.
Tracking what's controversial even within an echo chamber could be so important for a party trying to come up with something that would resonate
I've been getting a handful of updates in various fields that are AI summarized digests of what people are saying in online spaces (Reddit, Twitter, Discord, etc.) that I'd never otherwise spend enough time on to have a sense of the pulse. Complete with source links in case my faith in the human race is feeling strong enough to read the actual comments section. I could see a larger scale version of that being really useful politically.
It seems like a good thing, but a side effect is that the New York Times becomes more influenced by media that they wouldn’t otherwise watch. That is, these AI tools are a way to pay more...
It seems like a good thing, but a side effect is that the New York Times becomes more influenced by media that they wouldn’t otherwise watch. That is, these AI tools are a way to pay more attention to certain people. But who else should they be paying attention to?
Anyone looking to influence the media more should probably take having an AI tool devoted to them as a sign of success.
In July 2025, the Justice Department announced it would not make any additional files public from its investigation into child sex trafficker Jeffrey Epstein. The backlash against the decision was swift — and came from some unexpected corners of the internet.
A chorus of right-wing commentators and influencers openly criticized President Donald Trump and his administration for failing to follow through on their campaign promise to release the federal documents. Political podcasters who had embraced Trump during his reelection campaign were up in arms, with social media figures like Joe Rogan and Andrew Schulz publicly pressuring the administration to reverse course.
The New York Times trackedthisgrowingdiscontentacross the GOP base closely for months, culminating with the near-unanimous passage of the Epstein Files Transparency Act by Congress last November. An AI-generated report, delivered directly to the email inboxes of journalists, was an essential tool in the Times’ coverage. It was also one of the first signals that conservative media was turning against the administration, according to Zach Seward, editorial director for AI initiatives at the Times. (Seward was once an associate editor at Nieman Lab.)
[...]
Built in-house and known internally as the “Manosphere Report,” the tool uses large language models (LLMs) to transcribe and summarize new episodes of dozens of podcasts.
[...]
“In order to adequately cover this administration — among many other sources — it seemed crucial to have an eye on influencers, largely conservative young male influencers,” Seward told me. “It turned out there were enough specific requests and enough broad interest [in the newsroom] that it made sense to automate sending that out.”
Launched a year ago, the Manosphere Report now follows about 80 podcasts hand-selected by reporters at the Times on desks covering politics, public health, and internet culture. That includes right-wing podcasts like The Ben Shapiro Show, Red Scare with “Dimes Square” shock jocks Dasha Nekrasova and Anna Khachiyan, and The Clay Travis & Buck Sexton Show, a successor to Rush Limbaugh’s talk radio show. It also keeps tabs on Huberman Lab, a podcast hosted by Stanford neuroscientist Andrew Huberman that has been criticized for spreading health misinformation. Seward notes the report also includes some liberal-leaning shows, like MeidasTouch, an anti-Trump podcast with a largely male audience.
When one of the shows publishes a new episode, the tool automatically downloads it, transcribes it, and summarizes the transcript. Every 24 hours the tool collates those summaries and generates a meta-summary with shared talking points and other notable daily trends. The final report is automatically emailed to journalists each morning at 8 a.m. ET. The Times is exploring how to use this workflow to launch similar AI-generated summary reports for other beats.
[...]
The Times is not the first newsroom to turn to LLMs to parse through the mountains of audio and video material on the internet that journalists are expected to consume to keep on top of their beats. Local news outlets across the country have been using LLMs to keep tabs on school board and town hall meeting livestreams through email summaries. Last year, my colleague Neel covered “Roganbot,” a tool created by AI consulting lab Verso to generate searchable transcripts of The Joe Rogan Experience podcast. Among several features, the tool suggests potentially controversial or false statements to fact-check.
[...]
Seward said that the Manosphere Report was an outgrowth of one of those existing tools, called Cheatsheet.
[...]
As with the Manosphere Report, Cheatsheet is rooted in a philosophy that creating new text and images for publication is not the most effective use case for generative AI in a newsroom like the Times. Rather, Seward sees the technology as a way to amplify the newsroom’s existing investigative power.
Don't fall for the hype this "tool" is a magical AI miracle worker. It's simply a series of Python scripts that make API calls to ChatGPT, Claude, Gemini, or a combination of the three. The Times...
Built in-house and known internally as the “Manosphere Report,” the tool uses large language models (LLMs) to transcribe and summarize new episodes of dozens of podcasts.
Don't fall for the hype this "tool" is a magical AI miracle worker. It's simply a series of Python scripts that make API calls to ChatGPT, Claude, Gemini, or a combination of the three. The Times has enough money for its AI team to use nearly unlimited token for its work.
A simple tool can still be quite useful. Also there's a lot of activity around figuring out how best to build tools that use LLM calls to do interesting things. These tools are cheap to build so...
A simple tool can still be quite useful.
Also there's a lot of activity around figuring out how best to build tools that use LLM calls to do interesting things. These tools are cheap to build so there's a zillion of them. People are building their own.
It reminds me a bit of when seemingly everyone was building their own websites. A lot of them will be slop. Most will be ignored. But I expect there will be hits, too, like that crazy OpenClaw thing. Well, hopefully better than that next time.
This is a good example of an AI use case for a task that likely wouldn't be done otherwise (listening to loads of manosphere podcasts and summarizing them every week forever). There's a job no one at the NYT wants.
It replaces a lot of human hours but they're hours that likely wouldn't have been spent in the first place because it's not worth the resources (or pain). No doubt they have summaries setup for other "spaces" too.
Really smart use of LLM tools. This isn't the first time I've been surprised by the NYT's tech savvy
Yes very much so. There's so many terabytes of media generated every day it's not possible to consume even a fraction, having machine summary that one can then verify (very important) is a great investigative tool
I hope that it can have broader usage, and perhaps even be a way to break through Echo chambers: if it's not painful and heated and time consuming to hear that the other side is saying, we might be able to hear a bit more of it
For example, can you imagine how different an election prediction could be if think tanks and media used this for a pulse check instead of relying on polls, especially for views people know is kinda sketchy? I have no idea how Cheatsheet is weighted, but I imagine number of views/listens and engagement is part of it. Tracking what's controversial even within an echo chamber could be so important for a party trying to come up with something that would resonate
That's a good point, finding out what the demographic is talking about without the mental and emotional load of wading through the toxicity could be useful outside of journalism too.
I've been getting a handful of updates in various fields that are AI summarized digests of what people are saying in online spaces (Reddit, Twitter, Discord, etc.) that I'd never otherwise spend enough time on to have a sense of the pulse. Complete with source links in case my faith in the human race is feeling strong enough to read the actual comments section. I could see a larger scale version of that being really useful politically.
It seems like a good thing, but a side effect is that the New York Times becomes more influenced by media that they wouldn’t otherwise watch. That is, these AI tools are a way to pay more attention to certain people. But who else should they be paying attention to?
Anyone looking to influence the media more should probably take having an AI tool devoted to them as a sign of success.
From the article:
[...]
[...]
[...]
[...]
[...]
Don't fall for the hype this "tool" is a magical AI miracle worker. It's simply a series of Python scripts that make API calls to ChatGPT, Claude, Gemini, or a combination of the three. The Times has enough money for its AI team to use nearly unlimited token for its work.
A simple tool can still be quite useful.
Also there's a lot of activity around figuring out how best to build tools that use LLM calls to do interesting things. These tools are cheap to build so there's a zillion of them. People are building their own.
It reminds me a bit of when seemingly everyone was building their own websites. A lot of them will be slop. Most will be ignored. But I expect there will be hits, too, like that crazy OpenClaw thing. Well, hopefully better than that next time.