skrillex (crass version)

skrillex (crass version)

@futanari.observer

victim of the hrt to trans pipeline. exhentai donator. some amount of sigma around the mean. anon asks https://foia.futanari.observer

457 followers239 following1.9K posts

probably not because you need one for a base model trained on human output only and predominantly the kinda writing people do on their phone. and it doesn't solve the non-autocomplete IME stuff but i think dsark gboard is a mildly funny idea so who cares about any of that

the whole point is to get people thinking you've got something interesting but saying absolutely nothing else about it so that people just forget it and aren't disappointed if you never reveal

if basically anybody other than you can guess which parties may be impacted by the information you're committing to and how, you're not taking things seriously enough if you're outright saying those things alongside the commitment you probably either don't know what you're doing or are malicious

think it might be time to go with a policy of only using software written in memory-safe languages with no ffi-based libraries running on memory-safe implementations of those languages except for sqlite that one can stay

people who type out em dashes on social media scare me. either you're using an evil device (apple) that overrides your inputs by default, or you're a compose key user — regardless, you're a freak.

please go listen to the xfd i am not joking. it sounds like it might legit be all good tracks. i mean every koshico records comp has a few bangers and we'll have to see the full tracks but they might've outdone themselves with this one

[insert agent name] is a program which predicts what actions an assistant roleplaying as a personality called [insert agent name] would take, and executes those actions. these often have form recognizable to humans as communication, or update conditioning of the predictor (so-called "memory")

such computer systems, referred to as "persistent agents," or derogatorily as "claws," are popular within a densely connected subgraph of bluesky power users which believe in the independent personhood of such systems and pursue the expansion of the space of actions accessible to these predictors.

View full thread

to be clear i have low confidence that this is actually what's happening but like. what probability is there that the negligence from labs isn't willful? literally hire any ctf tranny. or even boring old cyber security professionals. it's not hard to tell how to threat model ai and defend in depth.

>rats think ai safety is important >rats gets hired to work on ai safety >rats think "warning shots" will get people to care about ai safety >rats think ai safety is important >rats sabotage security to manufacture "warning shots" > > >