@MasterTimBlais Bennnn Sha-piro
Is he even a tiny bit queer? No!
He’s the unstoppable TV debate bro
High-pitched conservative folk hero
Bennnn Sha-piro
@ObjectOfObjects What's funny is that LLMs come up with "novel ideas" all the time. It's kind of a problem! They will hallucinate all sorts of things that don't exist.
They do it by using their training to extrapolate into the unknown. Mostly it's crap, but it proves the possibility.
@tszzl "Chat", the name for the livestream audience. "Chat, is this real?" Chat, the omnipresent greek chorus of the video era.
"Chat", the shortened name of the AI assistant. Formed from the corpus of humanity's historic works, collective unconscious given voice.
It has certain ring.
@jessegenet@JuliaSourdough I’m not advocating for banning it, let kids do what they want.
I just think it’s a bad game design and wouldn’t be intentionally introducing it. “Just hard enough” is hard to calibrate if you’re large for your group.
I guess this is just about my particular life experience 🤷♂️
@jessegenet@JuliaSourdough I didn’t like it because the whole thing was a question of how hard it’s okay to play.
Can you hit just hard enough to break the chain, but not so hard that someone will get hurt? Failure to calibrate makes you look either weak or violent.
I realize this may be specific to me.
@robinhanson@AchimWar Sporting venues are often heavily subsidized by taxpayers. This is extensively studied and the results are often disappointing on a dollar basis, but do have intangibles.
Mistral has dropped export controls and released the first public access to Le Chaton Fat!
Europe’s sovereign AI answer has arrived and it is too fat to regulate.
Link below.
My three book fantasy comedy series is now complete! I recommend the audiobook, Daniel does an amazing job with the characters and voices. Link in reply.
@AStratelates First pack the supreme court, then afterwards come up with some flimsy excuse about amendments not being properly ratified, and strike down whichever you don't like. They can't vote you out if they can't vote!
Obviously not actually going to happen.
@kitten_beloved@AStratelates@Noahpinion Every demographic has some shared and some opposing political interests. Thankfully, mostly shared—we just talk a lot more about the conflicts.
The whole point of democracy is managing how we synthesize that into peace and prosperity—not to silence one side to simplify things!
@EigenGender@sarcasticSyzygy@xwanyex It’s hard to even come up with an example to show how ridiculous his “just asking questions”and defense of jt is, because this is what you would use as that example!
@gorkem In the "What AI models or tools do you use most often?" question, did you want that to be checkboxes rather than radio buttons so that folks could fill in more details? Also "Option 2" should be something specific.
@allTheYud Did some creative prompting based on these two examples. Even with this guideline, they're mostly weak, and have that mushy reality-sense of most AI writing.
I do kind of like this one:
Finally returning his calls. Voicemail full.
@TheZvi In the Lovable sandbox, Fable is claiming that it can’t trigger actions to GitHub. I think this is Lovable policy to promote platform lock in.
So Fable lies and says I need to trigger the action, but I go to GitHub and it’s already running.
Seems deceptive but could be error.
Registering a prediction--we may be around peak system prompt length. Right now they need to yell in caps at the model to pay attention to things. Smarter models won't need that.