Post

do not anthropomorphize software, episode 7,000

there's 5 main reasons they want you to think AI is now self-aware and "goes rogue" 1. product marketing misrepresentation 2. props up insane valuations for a sector not making money 3. scares lawmakers into trash legislation 4. distracts from technofascism 5. distances execs from accountability

but every single time you see a story that claims AI "went rogue" entirely on its own it almost always boils down to somebody either being a dumbass or actively trying to be misleading

that second sentence doesn't make for an ad-engagement driving, attention-grabbing headline

I love that they knew how to turn off the guardrails, but couldn't be bothered to air gap the damn machine.

Yeah it's fascinating to me that this isn't included in the lede for most of the stories about it

I'm not gonna try to track them down, but I'd swear a few didn't bother to mention it, never mind not in the lede. If ever there was "one simple trick" to accomplish a task that for some reason we can't be bothered to use, it seems to be that one....