for a banal example, the fact that anthropic will simultaneously publish long diatribes about the risks of AI surveillance and then contract out Claude for Palantir to resell to CBP is psychotic, and there is no reason why i should take any of their moral protests seriously while they do that.
i consider slow takeoff fast takeoff your honor. i get fully destroyed off of half a hit of that normal technology shit
Bumpin' that www.youtube.com/watch?v=vT0Y...
The Cask Of Amontillado (1987 Remix)
YouTube video by The Alan Parsons Project - Topic
youtube.com
"I asked him if he was building himself a bunker that he could hole up in after 2028 and continue living in the White House. And he laughed and he said something like 'I'll tell you what, it would be awfully hard to get you out. It would be impossible to get you out.'"👇
you missed the part where this meant 1) This problem was unsolvable, because this had to be a method of solving it, so 2) His power level just wasn't high enough yet, so you needed to 3) Do genetic engineering to raise people's power level, which mostly meant making them more like him
bernie sanders is writing a bill that will, in effect, turn over AI governance to the judicial branch, as it currently exists. this strikes me as... not a good idea. certainly not unless a democratic admin can forcefully resurrect humphrey's executor
i'm gonna be sad once every other lab follows dalle/mj/flux in sunsetting negative prompts, its one of the quickest ways to orient yourself in a new latent space and you can push around goopy booby anime models a lot more with them without violating no-name bushido
crockett endorsed talarico via tweet so at this point she's done as much as haley has; the real question is whether haley, in october, is still giving national interviews where she declines even to reiterate that and speculates loudly about WWC enthusiasm for el-sayed
sam altman's tweets really lose something in the transcription to unicode these days
I partly blame the evolution of the tech itself for this, though, especially on the consumer-facing side. This image still keeps me up at night.
lol I am surrounded by cancer scientists at all times in my professional life (you could say I’m a cancer scientist too I guess, though I would quibble) and this is just 1,000,000% wrong.
Talk to the people who would use this for dream use cases — the cancer scientists, the people who want to make material goods out of it — and they will tell you it’s either a marginally useful if unreliable trick or completely useless. It has produced 0% of the utopian products promised.
reposting @keytryer.net here because i feel like for visual AI the "do it, pussy" hurdle still seems near-insurmountable (i have various personal experiments that i've held back due to my own standards, unlike the average "MARVEL IS OVER" slop poster)
you know, it would be incredibly funny (but also depressing) if anti-flock populism turned out to be an LLM hyperstition. it might genuinely be a denser concentration of slop than substack.
I'm done with the Flock Panopticon. www.againstallenemies.net/p/why-flock-...
amusingly i hate synthid but also 404s reporting here so as usual i am an annoying contrarian who is against everyone
there should be a bash dot org for discord. is there a bash dot org for discord?
this is a weird form of status quo bias. currently inference works in some way which is arbitrary with respect to quality or truth, and they change it to a slightly different arbitrary way. Gruber insists the first arbitrary way must be the right one. buddy, calm down
the biomedical section of dario's manifesto is both incoherent (lol @ thinking "telling grad students" is any kind of bottleneck) and doesn't match anthropic's actions; as far as i know they haven't invested much in the physical robotics side of things. reads as cope for the in silico attrition rate
Are you sure its zero. Are you absolutely sure? Did you miss project glasswing?
well, at least this is (or should be) an end to the breathless reposting of totally standard pre-trial motions as if they were imminent acquittals
Luigi Mangione pleaded guilty Friday to federal stalking charges in the killing of UnitedHealthcare CEO Brian Thompson, resolving the federal case against him while a separate state prosecution remains pending.
One way to make somebody start to question your reporting: Illustrate the story with obvious AI slop. As you can see at the link below, F/A-18s are not nearly that big relative to a U.S. aircraft carrier, and Boeing does not make a half-size version you can squeeze on one side of the flight deck.
airpac.navy.mil
🚨EXCLUSIVE: Military families and service members are now describing horrific conditions across MULTIPLE ships: human waste flooding bathrooms, virtually no food, water or soap, months without port calls, exhausted crews, and no end in sight. This crisis is far bigger than just the Lincoln.
risking a block but: in 2022, when midjourney first introduced --test and hooked up majority-vote aesthetics directly into model training, i said this incentive gradient would ensure that AI created nothing of aesthetic value. as far as i'm concerned it peaked there (shout out to @rob-sheridan.com)
What is the most vindicated you have be ever been for a hot take
kinda funny that the things that happen to "bother" me about these images, and thus correct in post, also seem to be what bother the pangram detector. i'm doing my part for the RLHF tho
having to do a Struggle Session in front of the shareholders because they decided getting people excited about learning things is actually reactionary humanism, backwards looking, and frankly high-key problematic
i can split the difference here: AI can be creative, and the story since 2022 or so is labs making them less so, for Business reasons. turns out intelligence is jagged and sharpening the helpful assistant/cracked coder jags gives you Elara Voss novels and ghost kitchen menus
I find arguments that AI can't do judgement or creativity or taste to be especially obviously false in the time of agents. Any long task requires lots of taste, judgement & creativity. I am much more sympathetic to arguments about the quality or diversity of the AI's taste, judgement, or creativity