This is the web version of dot.LA’s daily newsletter. Sign up to get the latest news on Southern California’s tech, startup and venture capital scene.
Recently, Von Raees, founder of publishing group HeySoCal and tech startup Heywire AI, decided to conduct an experiment with his newsroom staff. Raees asked a human reporter and an AI reporter to write the same stories, and without knowing the bylines he said his own staff couldn’t tell the difference.
This sort of Turing test proved to Raees and his staff that AI newsgathering and reporting tools are advancing fast, and could serve as a catalyst for changing how the age-old business operates at a time when the industry is struggling.
In 2020, NPR reported that over 2,000 American newspapers have folded since 2004. That number sharply increased during the pandemic with more than 360 local papers – about two a week — went out of business since the pandemic began. What’s more, journalists are leaving the industry rapidly; between 2008 and 2020 the number of newspaper journalists nationwide more than halved.
This is a problem that Heywire AI hopes to solve.
“On the web, volume translates into dollars and profit,” Raaes told me. “If you want to make volume you need fast, regurgitated news that’s out there… many journalists today aren’t doing original pieces; most people who want to be journalists, that’s not what they live for.”
That’s why Raaes thinks his AI named Wells (after trailblazing reporter Ida B. Wells-Barnett) could step in to help get local newsrooms back on track.
Heywire, which launched last week, is affiliated with HeySoCal, the publishing group founded by Raees in 1996. The group publishes 21 local newspapers spanning the San Gabriel and San Fernando Valleys, Santa Monica and Long Beach that, according to Raaes, serve about 20 million readers in the Southland.
But could asking an AI to write news in place of reporters perpetuate this very problem of declining quality news coverage? According to Raees, not if the AI is tasked with writing the more mundane, daily copy that keeps a paper and its website running.
“This is meant to be a tool to augment the experience and capabilities of true human journalists,” Raees told me. “Our number one intention with this is to allow journalists to be journalists by taking away the mundane everyday sort of things that an AI can handle.”
According to Raees those include earnings reports, or formulaic stories about breaking news. This is a common sentiment shared by major news outlets. In 2020, Reuters began using AI to automate video reporting after developing an internal AI that would suggest story ideas, analyze data and write copy two years prior.
In fact, though AI still feels like a new trend, as of this week, the World Association of News Publishers found that 39% of newserooms are currently using generative AI tools, while only 20% reported having guidelines for such tools in place.
Raees believes that the HeySoCal news group can prove through its use of AI that it is worthy of use at a larger scale. The business model will eventually see Heywire license out its software in a SaaS model to other newsrooms of varying size. He told me Heywire is working on integrations with different content management systems to add the AI directly into the software newsrooms use to produce and edit content.
Editing and fact-checking are top of mind when building this AI, Raees said. He noted Heywire recruited former Los Angeles Times senior executive Jeffrey S. Klein as an advisor and is actively consulting with its reporting staff during development.
There’s also a concern that the AI could hallucinate false events or quotes. To prevent those issues, Raees said the AI draws from numerous sources of data and is never allowed to self-publish. Adding that, the AI is programmed to constantly scrub the internet for trusted information: “We are training Wells on an ongoing basis, with real time news content,” he said.
Right now Heywire’s target customers are small and medium-size news publishers. Raees said he’s in the process of raising $10 million to continue developing the platform, and also indicated there’s interest from two “multinational organizations” he wouldn’t disclose.
“All AI can do is deal with news that’s currently on the internet,” Raees said. “This is our way of contributing and being able to provide an avenue for small local news organizations to become viable again… And then taking their few journalists that they have left, and allowing them to go back to work doing what they're passionate about and what they love doing.”
- USC Expands AI Education With New Research Center, Amazon Partnership ›
- How Long Will the Public's Interest in AI Endure? ›
- Could Lack of Access to Training Data Block Lightning-Fast AI Development? ›