So the LA Times replaced me with an AI that defends the KKK

The LA Times is laying off and buying out staff, while introducing highly dubious AI tools. This is what automation looks like in 2025.

Greetings all — I’m in Austin at SXSW for the weekend; if you’re around give a shout. Yesterday, I had a great chat with crypto skeptic Molly White; tomorrow I’ll be talking generative AI and labor at the Lighthouse. Monday, I’ll join the 404 Media folks to talk about AI content and the state of the internet. Thanks as always to readers and subscribers who make all of this work possible.

My former employer got into some (more) hot water this week, when its new AI tool came to the defense of, I shit you not, the Ku Klux Klan. The AI feature, called Insights, is, for some reason, designed to evaluate the political orientation of opinion articles and then artificially generate countervailing points for the reader’s consumption. As my former boss, LA Times owner Patrick Soon-Shiong, explains in his introduction to the feature, Insights “offers an annotated summary of the ideas expressed in the piece along with different views on the topic from a variety of sources.”

In this case, the AI responded to a piece by my former columnist colleague Gustavo Arellano, which argued that his hometown of Anaheim shouldn’t forget the KKK’s reign of terror there. The AI Insight then informed readers that, actually, as the New York Times’ Ryan Mac pointed out, the Klan may simply have been “responding to societal changes rather than an explicitly hate-driven movement,” which is an objectively insane thing to say about one of history’s most obviously hate-driven movements.

This feature had already been derided for automating one of journalists’ least-favorite editorial imperatives, both-sidesing a story. Now it was automating the revision of history and the dulling the realities of racism, too.

I admit that when I first heard about the project, I had just rolled my eyes at this thing. There’s a lot going on, and sure, it seemed quite dumb, and to embody a wide variety of AI tools’ goofier impulses—it’s an ‘innovation’ nobody asked for, its output presents the tool as an authority over subjects (including political ideology) on which it categorically is not, and it creates more slop content that people have to sort through. It’s dumb, in other words, in a classic kind of ‘upper management was asked to bring three ideas about how to use AI to the table each and this was the one the boss liked best’ kind of way. But at first blush it didn’t seem orders of magnitude dumber than, I don’t know, the way Pitchfork music reviews are designed to suggest scientific rigor in a decidedly subjective arena by handing out scores down to the decimal point.

Now it’s clear that it is in fact orders of magnitude dumber than that; as far as I know Pitchfork never awarded an 8.3 to the soundtrack of the Triumph of the Will. This should be disqualifying for the tool, which, again, nobody seems to want or like, and which is appearing in the pages of a newspaper, whose ostensible goal is to be provider of reliable and dependable information to the public.

But the debacle gnawed at me beyond even all that. The week before the AI’s KKK incident, the Times accepted the buyouts of 48 journalists and staffers. This a mere year after layoffs hit 120 or so people, including yours truly—nearly a third of the paper’s staff. Those layoffs followed even more, just the year before. Into this void comes the dumb auto-both-sidesing AI feature. The AI tool, in other words, follows the mass layoffs.

Now, no one at the paper’s management would ever make the case that it is explicitly trying to replace writers or staff with AI; they would say that it is a feature that’s supposed to make the paper feel like it’s on the cutting edge, like it’s not falling behind, and so on. (They would probably pointedly not say that it is an effort imposed by a conservative, Trump and Musk-supporting billionaire newspaper owner to blunt his staff’s liberal leaning opinions.) My former bosses did not sit down and say, “I bet we can replace our tech columnist Brian Merchant with this highly dubious AI technology.” But that AI *is* also supposed to do is increase the time spent on page at as little cost as possible. It’s generating value that formerly would have been generated by human writers and editors. It’s also normalizing the use of AI in spaces formerly dedicated entirely to human voices. Its mere existence advocating a technological solution to a human labor issue.

It’s a similar trend that I noticed and wrote about, perhaps ironically, in a column for the LA Times: When BuzzFeed announced that it would be using AI to create content, games and quizzes, around the same time that CEO Jonah Peretti laid off the entire News division. Peretti and I got into a long debate about what was going on there: He argued he wasn’t using AI to replace any working journalists, that his AI strategy was an entirely different beast. I argued that he was still essentially trying to make up for the value lost by human journalists with AI features.

The point is that it’s rarely a case of one-to-one replacement of human roles with AI automation. AI features can convince management that they can make do with fewer workers, that some if not all of their value or production can be made up for—even if it can’t, and even if that AI tool might defend the KKK. And AI and automation are especially positioned to impact troubled industries—like journalism and writing. Take the example of Sports Illustrated, which only started pumping out the AI-generated content (the subject of another column!) after decades of decline, ad revenues cannibalized by big tech, and mismanagement.

This is what galls me, then. Writers and journalists are being ‘replaced’ not wholesale, not entirely or directly but in kind, by automated gimmicks and features. And rather than confront the tech sector that’s landed most of the existential body blows to journalism, management is once again trying to cut deals with AI companies, to build engagement-generating features, in hopes of wringing out a slice of revenue or relevance in darkening skies; meanwhile more and more journalism jobs disappear every year. The Wall Street Journal just laid off its tech reporters!

To its credit, Times management did support the push to get Google and Meta to pay a share of the ad revenues in California—an effort centrist-pivoting Gavin Newsom apparently killed with a veto threat. But I can’t help but despair a little when I see mass layoffs and new AI features unleashed back to back. We absolutely need to be disabused of the notion that AI can provide any sort of solution for struggling newsrooms and journalism in general, and think long and hard about whether we want to live in a world with more reliable reporting—or more automated bots that defend the KKK.

On that note, if you would like to help support journalism that is 100% human powered and guaranteed AI-free, and to make more pro-human writing, reporting, and analysis possible, you can do so by becoming a paid subscriber. Just this week, I’ve reported on DOGE’s mass firing of 18F, the widely respected team of government tech workers, on a new study that demonstrates workers aren’t falling for Silicon Valley’s AI hype, and written about the great horror film that predicted our post-truth politics. All that’s made possible by paying subscribers—thanks so much.