Here's a decision that probably sounded smart in a boardroom and did not survive contact with reality. LA Times owner Patrick Soon-Shiong wanted a way to prove his paper wasn't biased, so the Times built an AI tool called Insights that reads opinion pieces and automatically generates a "counterpoint" -- the other side, instantly, no human required. The very first big test came on a February column by Times writer Gustavo Arellano, about the Ku Klux Klan's real, documented history in Anaheim, California. The AI's generated counterpoint didn't just offer "another angle." It described the Klan as a response to a "white Protestant culture" reacting to societal change -- leaving out, you know, the part where it was a violent hate group. CNN's own headline on the fallout: "The LA Times' new AI tool sympathized with the KKK. Its owner wasn't aware until hours later." The note got pulled from that column. The bigger "bias meter" -- a public-facing score meant to rate every article's political lean -- quietly never launched at all, even though Soon-Shiong had announced it. The Times' own union wasn't shy about why: Guild vice chair Matt Hamilton said flatly that "AI-generated analysis unvetted by editorial staff" wouldn't build trust, it would erode it -- and pointed out the money spent building a bias-detecting robot could've gone toward the reporters who hadn't seen a cost-of-living raise since 2021. We don't think noticing your own blind spots is a bad instinct -- every newsroom should want that, us included. We just think there's a difference between asking your reporters to be more careful and asking a chatbot to grade their homework. One of those has an editor in the room. The other one, on its very first assignment, needed one.