I remember a couple years ago - probably more than it feels, with the way time passes these days - that the internet was finally having a conversation about unintentional algorithmic harm. I think that's what people were deciding to call it at the time. The way Facebook would offer you a photo collage of pictures with you and a dead relative or a bad ex, in the bright exuberance of Corporate Memphis. There actually was a moment of a movement to actually consider at a product design level: "would this feature be horrific to offer someone on the worst day of their life?"
It's darkly gratifying, as a skeptic of mainstream tech, to see that the industry briefly considered incorporating that question into its decisionmaking, and then roundly decided "nahhhhhh, fuck that actually." The only choice they were equipped to make, really. But it was all a bit of pleasant kayfabe that Big Tech could behave responsibly, while the moment lasted.