OFF THE TOP
You might have heard: New York Times, CNN and Australia’s ABC block OpenAI’s GPTBot web crawler from accessing content (The Guardian)
But did you know: News group says A.I. chatbots heavily rely on news content (The New York Times)
According to new research from The News/Media Alliance, AI chatbots disproportionately rely on news content for training. They argue that this is a violation of copyright law and an extension of big tech companies not compensating news organizations for their work. The study looked at how the curated large language models that are believed to train chatbots like ChatGPT compare with open-source datasets that pull generic content from the internet. The curated lists are between five and 100 times more likely to use news content than the generic set. [newsletter align=right]
+ Noted: Israeli govt. denies Rolling Stone a press credential after critical coverage (Rolling Stone)
Trust Tip: Ask if your news is really worth paying for (Trusting News)
Journalists in newsrooms with paywalls seem to forget that a lot of news ISN’T behind a paywall. Unless you’re the only newsroom in town, if you have a story about a traffic accident, a new restaurant opening or who won the high school football game, there’s likely another newsroom providing that information. The same is true for national outlets.
A lot of community updates are probably shared on social media. Posts in a local Facebook group might not have the context and detail a reporter would provide, but they might be enough to satisfy folks’ immediate information needs.
Is what you’re offering significantly better than what people can get for free? Perhaps most importantly, how can you convey your unique value within the text of the blocker people see when they’re told they need to pay?
TRY THIS AT HOME
For the Chicago-based Investigative Project on Race and Equity, training the next generation matters as much as the reporting (Nieman Lab)
The Investigative Project on Race and Equity is focused on adding capacity to Chicago-area newsrooms to help them do investigative reporting around racial equity. These stories will then be published by media partners. The Project pays part-time “apprentices” who work with professional reporters and editors on each of the stories. The goal, its founders say, is to train the next generation of investigative journalists on a project-by-project basis, without having to conform to a news publishing schedule.
Kyiv Independent shares lessons for growing a media startup during a war (INMA)
The Kyiv Independent was founded in November of 2021, meaning it had to build much of its infrastructure while its country was at war. They’ve been sustained largely through reader revenue, even as the war added new costs like security equipment. They’ve brought their reader-revenue platform in-house and invested in integrating technology to make it easier to connect with donors and predict monthly revenue. That has allowed them to build connections with their supporters. “We find people from all walks of life who understand our story could one day become their own if Russian atrocities aren’t held to account,” write COO Zakhar Protsiuk and advisor Matthew Harrison.
The Guardian rips Microsoft for distasteful generative AI poll about death (Axios)
A Guardian article about a teacher who was found dead at a school in Australia was featured in Microsoft Start, a curated news aggregator platform, alongside an AI-generated poll that asked readers what they thought was the cause of her death. Microsoft eventually took down the poll, which had upset readers and commenters who assumed The Guardian had written it. In a letter to Microsoft, Guardian CEO Anna Bateson said that they had previously asked Microsoft not to use their experimental AI on Guardian content for “exactly the sort of instance.”
A multilingual, statewide reporting collaboration (Medium, Center for Cooperative Media)
A group of New Jersey journalists and news outlets have built the state’s first collaborative voter guide, which focuses on the 120 state senate and assembly races. The collaborative — coordinated by the Center for Cooperative Media at Montclair State University and NJ Spotlight News — had a goal of obtaining questionnaire answers from all 246 candidates vying for state election. In the end, they had a 41% response rate. The final guide included stories and videos translated into Chinese, Turkish, Spanish, Urdu and Korean, and partner websites translated individual questionnaire answers into languages relevant to certain communities. [newsletter]