Earlier quoted context omitted.
Yes, I see a huge problem. Preventing damage is an order of magnitude more difficult than causing it.
For humans, maybe. Not an AI superintelligence.
Citation needed
Statement on AI Risk
361–370 of 964 posts
Re: Statement on AI Risk
#362I'm not sure they can keep claiming this without becoming concrete about it. Nuclear weapons are not nebulous, vague threats of diffuse nature. They literally burn the living flesh right off the face of the earth and they do it dramatically. There is very little to argue about except "how" are we going to contain it, not "why". In this case I truly don't know "why". What fundamental risks are there? Dramatic, loud, l…
To use Eliezer's analogy, this is like arguing about which move Stockfish would play to beat you in chess. If we're arguing about whether you can beat Stockfish, I will not be able to tell you the exact moves it will play but I am entirely justified in predicting that you will lose. Obviously we can imagine concrete ways a superintelligence might kill us all (engineer a virus, hack nuclear weapons, misinformation cam…
I understand but agentic/learning general intelligence has not been shown to exist, except ourselves. I’d say this is like worrying about deadly quantum laser weapons that will consume the planet when we are still in the AK47 phase.
Edit: it could still be true though. I guess I like some more handholding and pre-chewing before giving governments and large corporations more ropes.
Re: Statement on AI Risk
#363Oh gee, listen to the AI experts of Benguo, Hinton, Altman, and Sutskever…
or random HN commenters who mostly learned about LLM 6 months ago…
Congrats guys, you’re the new climate change deniers
Re: Statement on AI Risk
#364Pathetic attempt at regulatory capture. Let’s all be smart enough to not fall for this crap, eh?
I’m smart enough to not fall for all sorts of things as I sit here and watch Congress pass bullshit on C-SPAN anyways. It can’t be stopped, any attempts to influence are just teaching the system how to get around objections. Until power is actually removed the money will continue to flow.
Re: Statement on AI Risk
#365Earlier quoted context omitted.
It doesn't matter who wrote it, it got picked up, had a good argument and affected market opinion. The execs now need to respond to it. Humans also don't grasp that things can improve exponentially until they stop improving exponentially. This belief that AGI is just over the hill is sugar-water for extracting more hours from developers. The nuclear bomb was also supposed to change everything. But in the end nothing…
It's too early to say definitively but it's possible that the atomic bomb dramatically reduced the number of people killed in war by making great power conflicts too damaging to undertake: https://kagi.com/proxy/battle_deaths_chart.png?c=qmSKsRSwhgA... The USA and USSR would almost certainly have fought a conventional WWIII without the bomb. Can you imagine the casualty rates for that...
I'd actually guess those casualties would be quite less than WW2. As tech advanced, more sophisticated targeting systems also advanced. No need to waste shells and missiles on civilian buildings, plus food and healthcare tech would continue to advance.
Meanwhile, a single nuclear bomb hitting a major city could cause more casualties' than all American deaths in ww2 (400k).
Re: Statement on AI Risk
#366This is a great start but the only way you really get ahead of this is to get these people on board also:
- AI _hardware_ executives and engineers
- high level national military strategists and civilian leaders
Ultimately you can't prevent _everyone_ from potentially writing and deploying software, models or instructions that are dangerous such as "take control". Especially in an explicitly non-civil competition such as between countries.
You have to avoid manufacturing AI hardware beyond a certain level of performance, say after 2-4 orders of magnitude faster than humans. That will hold off this force of nature until desktop compute fabrication becomes mainstream. So it buys you a generation or two at least.
But within a few centuries max we have to anticipate that unaugmented humans will be largely irrelevant as far as decision-making and the history of the solar system and intelligent life.
Re: Statement on AI Risk
#367Earlier quoted context omitted.
> It’s hard to take these safety concerns seriously I don't get this mindset at all. How can it not be obvious to you that AI is an uniquely powerful and thus uniquely dangerous technology? It's like saying nuclear missiles can't possibly be dangerous and nuclear arms reduction and non-proliferation treaties were a scam, because the US, China and the Soviet Union had positioned themselves to capture the majority of t…
It's not clear at all that we have an avenue to super intelligence. I think the most likely outcome is that we hit a local maximum with our current architectures and end up with helpful assistants similar in capability to George Lucas's C3PO. The scary doomsday scenarios aren't possible without an AI that's capable of both strategic thinking and long term planning. Those two things also happen to be the biggest limit…
We don't need an avenue to super-intelligence. We just need a system that is better at manipulating human beliefs and behaviour than our existing media, PR, and ad industries.
The problem is not science fiction god-mode digital quetta-smart hypercomputing.
This is about political, social, and economic influence, and who controls it.
Re: Statement on AI Risk
#368Earlier quoted context omitted.
For humans, maybe. Not an AI superintelligence.
Citation needed
The entire GP's argument has no citations either and that is the framework we are working under - that superintelligence can do anything you tell it to do. Ask him for his citation, then the rest follows.
Re: Statement on AI Risk
#369Earlier quoted context omitted.
It's not clear at all that we have an avenue to super intelligence. I think the most likely outcome is that we hit a local maximum with our current architectures and end up with helpful assistants similar in capability to George Lucas's C3PO. The scary doomsday scenarios aren't possible without an AI that's capable of both strategic thinking and long term planning. Those two things also happen to be the biggest limit…
>It's not clear at all that we have an avenue to super intelligence. All problems in reality are probability problems. If we don't have a path to superintelligence, then the worst problems just don't manifest themselves. If we do have a path to super intelligence then the doomsday scenarios are nearly a certainty. It's not really any different than saying "A supervolcano is unlikely to go off tomorrow, but if a super…
Why are doomsday scenarios are certainty then. What's the model to get to that that isn't just some sort of scary story that waves away or into existence a lot of things we don't know if they can exist.
Re: Statement on AI Risk
#370In the limit, AI is potentially very dangerous. All intelligence is. I am a lot more worried about human intelligence. Re: alignment. I'm not concerned about alignment between the machines and the owner of the machines. I'm concerned about alignment between the owner of the machines and me. I'm happy I see comments like "Pathetic attempt at regulatory capture."
I used to be in this camp, but we can just look around to see some limits on the capacity of human intelligence to do harm.
It's hard for humans to keep secrets and permanemtly maintain extreme technological advantages over other humans, and it's hard for lone humans to do large scale actions without collaborators, and it's harder for psychopaths to collaborate than it is for non-psychopaths, because morality evolved as a set of collaboration protocols.
This changes as more people get access to a "kill everyone" button they can push without experience or long-term planning, sure. But that moment is still far away.
AGI that is capable of killing everyone may be less far away, and we have absolutely no basis on which to predict what it will and won't do, as we do with humans.