shambibble

you've heard of the bell curve, now get ready for the HELL curve

ah, right on time www.cnn.com/2026/10/25/p...

made my biennial futile gesture towards blexas, long live gay communist james mohammed talarico

Engineer them to do what? They’re not even trying to be practical anymore. They’re just selling the idea of being a Sci Fi villain to other VC people.

Argon is AGI confirmed

made some modestly slutty scallion noodles with peanuts and chili crisp

kind of insane that the same company that makes plexamp has gone 10+ years without patching in the ability for its music database to match multiple artists on collaborative/split albums

i'll give it to opus 5.5; it's at least better at comedically lampshading and recovering from these

“The French never rioted or burned things down before!” (I mean, if anything, this shows that immigrants have assimilated to French history and culture).

keep an ear open for the name "rachit dubey." they are slimy and shameless enough to go very far in the current information environment, and i reckon we'll be hearing from them more in the future, claiming never to have sold stinger missiles to the butlerian jihad

It really says something deep when support staff do their job (and here it seems well) and faculty feel comfortable to throw them under the bus.

Of course I never claimed or suggested that you are a bad actor, only I found the press release non-credible. If you have no control over your own press office, surely you can’t expect me to have any control over the posts of various random people who follow me on Bluesky.

If we want to regulate AI, why not start with a simple law forcing LLMs to use "Meesa" instead of "I"? Easy to build, easy to verify and deflates a hype while still allowing people to use chatbots (so no one can claim we are limiting anything)

what do burglars even steal now? any TV one guy can run out with probably costs <$500. do they grab desktop computers and hope for recent parts? raid the bathrooms for meds? grab the keys and pivot to grand theft auto?

shame you can't self-play cancer and verify cures in lean

those "decompiled" Adobe apps are bugged to shit and don't work. please test the thing someone is claiming before you boost what is effectively a higgsfield-tier AI slop post on here of all places

This is basically exactly correct. The problem with LLM output is usually code quality, and asking to merge garbage code is rude. If you and your tools can do good work, fine. If you can't, get good. Managing to make a hack that kind of works is not enough to contribute to sustainable development!

the murdoch papers have been posting unflattering stories about bari weiss all year, because she's explicitly trying to poach fox's audience. she sucks, but this isn't some kind of trending indicator

increasingly I think the way to look at a lot of these guys is LLMs. this is not something Wayne "believes". it's just the output you get if you prompt him about indians doing better than white people

i am still curious to what extent curricular learning matters in AI training, for image models especially. does front-loading scanned drawing exercises and public-domain old masters improve the model's "fundamentals" at all? based off novelAI i would say "yes," but that's only my power user eye test

a funny thing i am noticing about opus 5.5 is that if you ask it for a translation task from language A to language B, the chain of thought will incredibly frequently detour into languages C, D, and E

cracking up at the idea of shooting a scene where the car is "blasting NPR." what is recognizably NPR? are they gonna set this to the all things considered riff?

at this rate, i'm gonna end up voting for the dem primary candidate who sounds the most like ed zitron, because "up against the wall for technically incorrect reasons" will be preferable to returning to any kind of business as usual with these oligarchs

pangram is the only AI detector that has earned "trust me bro" status, and it does not appear in the article. applying the eye test to graphs of unicode points w/o a control can only ever be suggestive. what is also suggestive: the friendly interaction in benjamin ryan's inevitable first post

crockett more or less confirming that this post-primary tweet will be her sole gesture of support and she will not utter a single positive word about talarico (this has been obvious for so long that frankly continued silence would've been preferable to whatever the fuck this is)

"attempting lazy misandry and doing transphobia instead" is a failure mode i see from more and more big woke 1 accounts lately and i feel like there should be more learning happening

i should note that much of the time this is a feint. scott alexander is not a "high-decoupler" on race; he views flippant racism in the sailer/hanania style as such a positive epistemic signal that a schizophrenic nut permabanned from his own discord for slurs is liable to turn up on his links page

6' vs 5'11

Never in doubt

as someone who has been really critical of a lot of the initial breathless "teehee, weird sex" reportage, i really appreciate this thread for nailing the actual fucked up immoral thing that is happening; you CANNOT consent if your employment or job success are on the line

honorable mentions to utopia and full tilt poker as they are edge cases i am not sure qualify in the same sense as the others here

mfw anthropic hypes up another in silico bioinformatics result that would get a politely encouraging nod from your average PI

it don't make no damn sense

this is wholesome pet care, the slurs less so

should be pretty easy to write a national security pretext for a law that says Congresspeople are only allowed to post with their own fingers