AI Safety or lack thereof, to the extent that it emerges as a real threat, will be a matter for law enforcement. Just like “terrorism” and “drugs” and all the

AI Safety or lack thereof, to the extent that it emerges as a real threat, will be a matter for law enforcement. Just like “terrorism” and “drugs” and all the other vaguely menacing things that ended up being worth a pile of money to some incumbent interest, the real story will be boring: some crooks used a fancy-ass LLM to take a bank apart or something. The bank calls the cops, this happens about twice before the cops have a division for this sort of thing, and sooner or later someone is in a courtroom. It would be nauseatingly boring to get into how many times this has happened. Also predictably depressing, the real story, which is yet another brick in the wall of “we’re handing the reigns to people who make movie Zuck look flat fucking normal”, goes largely unreported. We used to complain about dual-class share structures. Sam just eviscerated a board of an ostensible charity with the full backing of MSFT and got himself installed as Ungovernable God Emperor of Arrakis for life. Still doing the eyeball scanner thing. Still has never shown the public he can handle numpy to the tune of MNIST. This is all in broad daylight and the Gemini launch gets clobbered because the 4-series comparable one comes out in January. Sam’s a smart guy, I’m quite sure he’d pay attention if the braintrust (this site) complained loudly. But the message is: “more, not less”.