Google’s own argument would also “significantly diminish” the benefit of the feature, the court noted, if the overview were “generally recognized as unreliable.”

That’s an important detail, and I’m glad the court caught it up — Google tries to eat the cake and have it too.

The model was tweaked to vomit certainty. Its output is in a page you’re hitting as you’re looking for answers, and this implies the output contains an accurate answer. Everything screams “trust me” = “be gullible” from a distance. Except of course when the model gets something wrong, then it’s your fault for being gullible/trusting.

And this is not just Google, mind you. Every single corporation behind large “language” models does the same shit: “believe me”, then “lol you stupid you believe me lmao haha”.

Reminder: fooling suckers is still fooling people.