Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

White House Secures Commitments from Leading AI Companies to Manage AI Risks

whitehouse.gov

Post reply on HN
Print view View original

91–100 of 105 posts

Re: White House Secures Commitments from Leading AI Companies to Manage AI Risks

#91

Post by throwaway72762 »

kalkin wrote: post #90
throwaway72762 wrote: post #89

Earlier quoted context omitted.

Why should the tone have changed? The billionaires and AI godfathers talking about it now all were inspired by LW and scifi, and it's the same narrative: no need to care about people living today or real problems in the world when there's a scifi story about trillions of people in the future for whom we must sacrifice anything (except the billionaires and AI godfathers comfort, status, or money) in the present to pro…

> "all were inspired by LW" Evidence? > "no need to care about people living today or real problems in the world" Nobody says this. > "trillions of people in the future" Weird longtermists who talk about this are a small minority even among AI doomers. > "AI godfathers comfort, status, or money" I'm not suggesting anyone feel sorry for Geoff Hinton but quitting your Google job does seem like a pretty obvious sacrific…

Eh, one can dislike the longtermists and the non-longtermist plutocrats. I'd say there's good reason to dislike both, but for different reasons.

Re: White House Secures Commitments from Leading AI Companies to Manage AI Risks

#92

Post by LatteLazy »

kalkin wrote: post #66
LatteLazy wrote: post #5

Earlier quoted context omitted.

I think you also need to draw a line between different types of danger: An actual AGI, that can think and has it's own motives and initiative is an entirely different danger to GPT-4, -5 etc which cannot. Those might post a danger in terms of fake news, or worker displacement, but they're just social issues. No rate improvement in chat GPT will make it "decide to kill all the humans" etc.

People have literally tried to hack together an agent on top of GTP-4 and given it an explicit goal of "kill all humans" for the lulz (ChaosGPT). I don't think that particular case is going anywhere, but assuming that nothing roughly GPT-shaped can acquire harmful goals seems naive.

Respectfully, you're mixing things again: you have described 2 separate unrelated things:

* A system like chat gpt that can do things a human cannot (find a way to kill all humans that's achievable by a small group).

* A system different to chatgpt that has its own motives and initiative.

Both need to be achieved.

Re: White House Secures Commitments from Leading AI Companies to Manage AI Risks

#93

Post by lesuorac »

LatteLazy wrote: post #18
lesuorac wrote: post #8

Earlier quoted context omitted.

I mean if you wire the output of GPT-x to a nuclear reactor it can cause all kinds of danger. The thing that keeps us safe from computers is generally we don't let them have access to anything dangerous and when we do theres typically an evaluation of what can go wrong and how we'll prevent the computer from doing that (often in the other direction, we only program the computer to do X in Y situations). But sometimes…

If (and it's a huge if that we are no where near) we created a super intelligent, general ai, how would you keep it from accessing things it shouldn't? It would be smarter than the world's best hacking team and work 24/7. That's sort of the deus ex machina of AI...

Of course if you create some super intelligent, general ai, and put it into a robotic humanoid its going to have a lot of agency.

However, Chat-GPT is only able to interact with people because it was hooked up to the internet not because it figured out how to access the internet. It also didn't create a worm and infect the entire world, the second you leave the webpage you no longer have Chat-GPT.

The way to keep the AI from accessing it shouldn't is literally just don't hook it up to those things in the first place.

Re: White House Secures Commitments from Leading AI Companies to Manage AI Risks

#94

Post by mrguyorama »

kalkin wrote: post #82
mrguyorama wrote: post #77

Earlier quoted context omitted.

Pretending to play along with "requests" to "keep AI safe" while ignoring the harms it will cause in other ways as the companies that own it try and gobble up literally all the money is 100% in line with the Sam Altman style of "I deserve to be the richest motherfucker on the planet" thinking

Do you think AI is capable of gobbling up "literally all the money" for its owners? If so, doesn't that suggest it's kinda dangerous? FWIW I don't think these voluntary commitments are sufficient or address all of the important harms. But that's different than saying any government action is just "regulatory capture" and an attempt to keep open source models down. I'm only attempting to argue against the latter here,…

>AI is capable of gobbling up "literally all the money" for its owners?

No, I think wealthy and capital rich companies can leverage AI to extract value from places that already extract value simply by underpaying people and use stupid rhetoric and other methods that they are very familiar with to continue buying up basically everything as most normal people struggle to survive.

I explicitly do not think "AI" or even AGI has any inherent danger in itself, and only has danger in the same way any automation has in a capitalist system that also gives richer people more power in the justice system and the political system: Rich will get richer by squeezing the non-rich even harder, while politicians don't care because most of them genuinely believe in capitalism as an unabashed and incorruptible good and the misinformation and discourse control that passable bullshit generated with a single keystroke grants them way more power

Re: White House Secures Commitments from Leading AI Companies to Manage AI Risks

#95

Post by mrguyorama »

johngladtj wrote: post #78
mrguyorama wrote: post #76

Earlier quoted context omitted.

Consider this: There are people much much smarter than you RIGHT NOW that you do not listen to or refuse to agree with. There are clear statistics RIGHT NOW about things that you choose to disbelieve. If you don't think this is true, then it is even more the reality. The idea that a very smart "AI" can magically convince anyone of anything is completely unsubstantiated and made entirely up, by people basically invent…

Given that we already have AI convincing people to kill themselves or that they are aware, I'd say the chance that a more advanced ai is capable of convincing humans to do certain actions that the human may not fully understand the totality of the consequences is pretty high.

And plenty of people are convinced of god by an "image" on toast. The existence of very gullible people does not imply that ANYONE is always convincible by a sufficiently "magical super-intelligent" "AI". People bring up "oh but think of it like your mental superiority to ants" and well, go ahead and convince a colony of ants to nuke russia, I'll wait.

There are already millions of people that are stupidly easy to convince of anything, so why aren't these magical end of the world scenarios already happening? Where's the evil maniac convincing Kelly the soccer mom who is into MLMs to end the world? Where is the supervillain raising an army of a million gullible idiots to invade Canada and have their own kingdom?

AGI "Superintelligence" even as a concept is just completely unsupported. It's people looking at an extremely cropped graph and saying "Line goes up, must go up forever" and wouldn't you know it that group has massive overlap with those MLM people from before.

Re: White House Secures Commitments from Leading AI Companies to Manage AI Risks

#96

Post by kalkin »

LatteLazy wrote: post #92
kalkin wrote: post #66

Earlier quoted context omitted.

People have literally tried to hack together an agent on top of GTP-4 and given it an explicit goal of "kill all humans" for the lulz (ChaosGPT). I don't think that particular case is going anywhere, but assuming that nothing roughly GPT-shaped can acquire harmful goals seems naive.

Respectfully, you're mixing things again: you have described 2 separate unrelated things: * A system like chat gpt that can do things a human cannot (find a way to kill all humans that's achievable by a small group). * A system different to chatgpt that has its own motives and initiative. Both need to be achieved.

I was responding to:

> No rate improvement in chat GPT will make it "decide to kill all the humans" etc.

All I'm trying to say is that GPT-style architectures not having "agency" shouldn't be very reassuring. How intelligent they can get is a separate question and one on which I am agnostic.

Re: White House Secures Commitments from Leading AI Companies to Manage AI Risks

#97

Post by LatteLazy »

lesuorac wrote: post #93
LatteLazy wrote: post #18

Earlier quoted context omitted.

If (and it's a huge if that we are no where near) we created a super intelligent, general ai, how would you keep it from accessing things it shouldn't? It would be smarter than the world's best hacking team and work 24/7. That's sort of the deus ex machina of AI...

Of course if you create some super intelligent, general ai, and put it into a robotic humanoid its going to have a lot of agency. However, Chat-GPT is only able to interact with people because it was hooked up to the internet not because it figured out how to access the internet. It also didn't create a worm and infect the entire world, the second you leave the webpage you no longer have Chat-GPT. The way to keep the…

Yeah, chat gpt is in no way intelligent. People are using it (incorrectly) as a substitute for an actual general ai...

Re: White House Secures Commitments from Leading AI Companies to Manage AI Risks

#98

Post by LatteLazy »

dylan604 wrote: post #88
LatteLazy wrote: post #41

Earlier quoted context omitted.

You can't fit a manual override to these systems any more than you can a person.

what? if a computer control is installed to a physical anything, you can always add cutoffs to the control at the physical point so that even if the computer is telling it to do something the physical thing no longer receives those controls/signals. designing it without those safety precautions would be 100% asinine. it'd be like allowing openai to design everything with no controls. even Musk allows the drivers to t…

Modern planes are purely computer controlled. I imagine the same is true of Teslas. And nuclear power plants and missile systems etc.

Re: White House Secures Commitments from Leading AI Companies to Manage AI Risks

#99

Post by dylan604 »

LatteLazy wrote: post #98
dylan604 wrote: post #88

Earlier quoted context omitted.

what? if a computer control is installed to a physical anything, you can always add cutoffs to the control at the physical point so that even if the computer is telling it to do something the physical thing no longer receives those controls/signals. designing it without those safety precautions would be 100% asinine. it'd be like allowing openai to design everything with no controls. even Musk allows the drivers to t…

Modern planes are purely computer controlled. I imagine the same is true of Teslas. And nuclear power plants and missile systems etc.

Huh? if this were true, the plane would never have been safely landed in the Hudson by a human pilot. yes, the majority of flights are flown by the computer autopilot because of the fuel efficiency, but the pilot can take control at any point.

Re: White House Secures Commitments from Leading AI Companies to Manage AI Risks

#100

Post by adamsmith143 »

hgsgm wrote: post #52
adamsmith143 wrote: post #24

Earlier quoted context omitted.

>I have been asked by many people to explain how AI could kill us all. How about some clarity for common people. There are countless resources to find this stuff out. To be clear there may not be a level at which common people understand why e.g. Fission Chain Reactions are dangerous but they can grok why a giant bomb that spreads radiation is bad. For Starters: https://www.lesswrong.com/posts/uMQ3cqWDPHhjtiesc/agi-r…

That's the same as saying "through God all things are possible". Yes if you skip over the hard step of creating an omniscient, omnipotent being, then you have the problems of dealing with an omniscient, oomnipotent being.

So the fact that thousands of researchers with billions of dollars behind them to build this omnipotent being means nothing to you? Because they all certainly think it's achievable.
Post reply on HN