As highly skewed perspectives and hoaxes ricochet within social media bubbles, a new site called Knowhere is using homegrown natural-language processing and machine learning to scan and weigh sources around the web–from Slate to Infowars. The AI then writes what its creators consider to be unbiased articles.

Trusting a robot to not only understand news but convey it to readers recalls studio executive Griffin Mill’s line in the movie The Player: “I was just thinking what an interesting concept it is to eliminate the writer from the artistic process.” But Nathaniel Barling, the 23-year-old CEO of San-Francisco-based Knowhere, isn’t eliminating writers. He’s just using ones who work at other companies to fuel his site.

On story selection and evaluation, “we make that judgment algorithmically first, in order to be as efficient as possible,” says Barling. A team of nine editors then rewrites what the software spits out to improve grammar and flow. The edited articles I read, like one about Kellyanne Conway, met the industry standard for readability.

Knowhere’s technology identifies hot stories by scanning a hand-curated list of “professional resources,” explains Barling, whose father, Kurt, is a journalism professor and former BBC reporter. How did a conspiracy-theory site like Infowars (which claims the Sandy Hook shooting was a hoax) make the list? “They are absolutely not well regarded for [accuracy],” Barling concedes. But he defends including such sites: “How does one know [what] a right-leaning or a left-leaning or a completely egregious representation of the story is if…