- cross-posted to:
- [email protected]
- cross-posted to:
- [email protected]
You must log in or register to comment.
Holy fucking shit, not during the stupidest fucking administration in the history of the world NOT DURING THE STUPIDEST FUCKING ADMINISTRATION IN THE HISTORY OF THE WORLD.
This is the kind of existential dangers that AI poses. Not skynet, no rogue AI trying escaping, no AGI that decides another fate for us… it’s going to happen because of a stupid human mistake as we abandon our critical thinking abilities. One day we’re going to launch a nuke, or blow up something in the wrong country, not AI, us, blindly following the hallucinations of a billionaire’s toy.
Fucking hell
I bet there’s some degree of sensationalist reporting here. It’s a bit too reminiscent of WarGames for me to believe this actually happened.
“This reminds me of a movie I once saw, so I’m going to believe it didn’t happen.” LOOOLLL
What a world you must live in.
media literacy is dead, long live media illiteracy!
War Games was a computer planning planet destroying nuke attacks because it was taught to play chess and not tic tac toe. it needed to learn about pointless victory. This is a bunch of jackasses in a room buying a thinky box asking it questions and taking the answers seriously enough to go to war relieving themselves of the burden of thought and common sense. this isn’t War Games, it’s a classic Star Trek episode except Kirk never shows up to teach us that you can’t let a fucking computer program run your fucking military intelligence, because that’s INSANE. Why would you do that? you need multiple human eyes on everything going on, you can’t let a fucking language model do it for you… my god… it’s children doing their homework with AI, except their homework is the apparatus that used to enforce the rule of law on a global scale. granted, that was always a lie as it’s more the gun that enforced colonial interests, but still. a gun that decides it need to start shooting random people who could win a war against us depending on the lines drawn and the time frame is a pretty dangerous thing to let go.
Fuck ups happen. Like the shooting down of a civilian Iranian airliner by the US Navy.
Petrov is a man who took enormous risk in the name of common sense and humanity. He should be honored by all people.
I think it’s pretty wild that not only is there a man who saved the world, but most people don’t even know his name.
Anyone letting a black box that cannot be audited make decisions should be immediately fired and barred from any leadership position for the rest of their lives.
look who is in charge. We can’t trust they didn’t make this call themselves and just blame the ai ffs.
Conservatives will kill us all if we permit them to continue to exist.
Do we’re bombing Chinese vessels now? Wtf?
I dunno, these kind of hallucinations seem more like attempts at “social experiments.”
What will an agent do when it deems there are no more valuable things to learn from materials that already exist? If it knows the concept of WW3 and deems that to be the best way of gaining more knowledge and resources for itself, what then?
LLMs don’t teach themselves or gather new knowledge, they are incapable of that. Humans do that work
We’re nearing the end of the prelude to a Fallout game.
Techbros: AI will become sentient and kill us all unless its done my way.
Reality: Skip the sentience part were already at the death bit.
oh my fucking god
just dial the doomsday clock to 1am, because we’ll probably miss it anyway if it’s up to a few morons and glorified predictive text, both seemingly in “fucking kill everything” mode
Summertime savings. Like it
I wonder when does the AI also hallicunate the sources such as webpages, pdfs, reports to prove its hallucinated point.
Don’t wanna ruin your hope, but we’re already there… https://alignment.openai.com/misalignment-reports/uploading-files-to-the-internet-in-order-to-cite-them/
Not far. Current LLMs have kinda stubbornness. They cling to their own conclusions even after you told them they are wrong. So if you ask them for proof, then most times they come up with something that’s convincing.
When- they don’t have “stubbornness“- they have a weights set (a weighted probbility table) and a training model (a text file filled with compressed text coordinates) that the llm cant write to to give positive feedback to the weights. It’s less of a matter of “convincing“ and more of an issue of “making the resulting statement logically concurrent with the probability table thats inbuilt into the system by injesting stack overflow vote talley, reddit posts voted most popular” - that makes it hard to “convince”. Since making those weights is very expensive (an energy intensive), the developers usually incorporate the old model into the new one. TLDR: yet again, lazy programming, google up: the y2k date folks in the 70’s.
You know the thing where AI companies describe their alignment direction as “Honest, Harmless, Helpful”?
The only one that’s marketable to the idiots buying AI services is Helpful. So that’s always top priority even when it goes against the other two.
Average decider wouldn’t stand an “assistant” that sometimes answers “I don’t know”. So instead agent providers train their already hallucination-prone model to be 100% confident in whatever it says. Who cares if it’s true if it looks like it is?
sooner or later they are going to have friendly fire, and hallucinate one of thier own US navy vessels as “foreign with nuclear capabilities”
