Narbo

@garbonarbo.bsky.social

You know, in the world where people were building specialized AI models for limited use cases on things that actually improve people's lives and happiness, AI wouldn't be meeting the resistance that it has and would probably be celebrated as the achievement it is. This is not that world.

Meredith Rose (from the grave) 👻@mrose.ink · 3d ago

Consumer-facing LLMs have a ton of limitations! They're advertised for all kinds of crap they can't do! There's a whole huge *but largely distinct* set of policy nightmares associated with them! Open-weight, specially-trained AI models do exist and are being used to do really wild research stuff!

as i said from the jump, the anti hasan ads aren't working. just because the gop consultant thought this would be their most effective strategy, doesn't mean it would work. they simply have nothing to run on, and now they know it because the polls show it. libs should take note

Bild

my initial read on the Cornell 7 case is that we’re likely to have an overdue national conversation about the voluntary intoxication loophole.

They’re fucking up the social issues too, homie. You cannot ethically use a tool built on the stolen collective works of humanity and then also claim to understand any kind of social issue in good faith. Also, getting the technical things wrong results in blood.

hikikomorphism@hikikomorphism.bsky.social · last wk.

A lot of journalists use LLMs as a jumping off point so they can justify talking about social issues, because editors love AI right now. Sometimes they get technical details wrong, and maybe in that world it's understood as the polite thing to do to ignore that and focus on the social-issue point?

Watching racists rage after they got banned from the WoW Forever beta for being flagrant genocidal pieces of shit has been very satisfying this morning.