Witnesses Assumed Charlie Kirk Shooter Was Just Ordinary Gunman On School Campus OREM, UT—As law enforcement officials search for a person of interest in the assassination of 31-year-old Turning Point USA co-founder Charlie Kirk, witnesses on the scene at Utah Valley University ad...

Witnesses Assumed Charlie Kirk Shooter Was Just Ordinary Gunman On School Campus

every time i use an all-gender restroom i get nervous that someone will figure out that i'm not really all genders

I'm not sure the feature engineering exist to make that work on real data.

It’s going to be okay because we’re going to make it okay

Used to be and still is. A billion dollars simply isnt that large in the housing market.

oh never mind this is actually true in Britain

I guess the answer to the Democratic Party's Young Men Problem is to, like, run even one guy under 50 who doesn't clearly suck?

the race for the twenty-first century is a race to see who can stop hitting themselves first

Remember, the Online Safety Act was passed by a Tory government in 2023. It only just went into effect. If Labour wanted, they could safely wash their hands of the whole mess and blame conservatives.

Except now they *are* the conservatives.

I’m deadly serious by the way.

The UK fiscal and economic situation is fucking dire right now and that is in large part because government policy is permanently tilted towards the needs and preferences of retirees.

Pedocon theory appears to be the real engine of history

It's... complex. But to sumarise, the "centeral" group likes usaid, but its diffuse, and usually avoids dissavowing. In general, its about what *you* do, getting others to agree is optional.

why am I paying someone to fetch me a bucket of water when I can get as much as I want at the free tap it came from

China is cracking down on young women who write gay erotica At least 30 writers have been arrested across the country since February, a lawyer tells the BBC.

I think people not in online fandom world are liable to under estimate how significant an expression of women's sexuality this is nowadays. It'd be like banning books with topless Scotsmen called The Highlander's Hidden Heart or whatever in the 80s/90s.

www.bbc.com/news/article...

I dont really understand the differences between this and universal MOEs arxiv.org/pdf/2405.16039 arxiv.org/pdf/2310.07096

I thought that going mad from microsoft copilot (as opposed to something like 4o, with its known sycophacy issues(also the fact that github copilot, the other github copilot, microsoft copilot, 4o and o4 are named that way is stupid)) was silly but then i remebered sydney bing.

(If the “future scary paradigm” is so bad, should we push on LLMs instead? I offer some concerns there too—e.g., the reasons to feel good about alignment of today’s LLMs seem to be getting less and less applicable over time!) (9/10)

in this sense trump is basically the id of every american voter

I mean their aggenda mostly consists of "Loosen zoning restrictions" and "stop having so many meetings" so its like half of their aggenda. and from what I can tell, seemingly the controversial one with voters.

"Greedy physicians expand clinic, want to treat and cure another 21 patients."

That doesnt really make sense to me for EG a train. Good to counterbalance the natural priority of who would win in a colision though, otherwise you end up with just a bunch of big cars.

AI alignment folks: Is there a consensus on the relative difficulty/importance of:

(1) Making AI understand human values
(2) Making AI care about human values
(3) Making AI succeed at what it tries to do

(Or... is there a consensus that this is the wrong way to think about it?)

So yeah, its less about the technical difficulty and more about the stakes. in general its assumed that and ASI will have a high quality model of the world and thus of humanity, so 1 and 3 are basicaly the same. or rather, as a model is not an agent, 1 is a subset of 3.

My understanding is that 2 is the hardest for superintelegent AIs, though this is assumed to be because you cant trial and error it. For other AIs, the ordering is less strict.

On the bit torrent twitter wishing it was the signal twitter.

Well they are predicitive models. The base models closely match the proabilities of humans for any given answer (Note, not the chatbots, as they are given futher training)

something that really irritates me is that the same corner of the Internet puts out these two messages without really understanding the contradiction between

"just don't be an asshole/be a good person and you'll get ___"

"no one owes you anything/life owes you nothing"

it's insane how good LLMs are given how bad they are

I tried to get Gemini to make this for me but I guess it can't do that. Usually it just changed the pfp and this time just gave me an unrelated image and I'm too lazy to edit the image myself