Live data from Hacker News

Meta disbanded its Responsible AI team

theverge.com

391–400 of 409 posts

Re: Meta disbanded its Responsible AI team

#391
post #389

Earlier quoted context omitted.

Evidence isn't the only valid form of argument. Consider a mathematical proof or a thought experiment, for example.

Show me a mathematical proof then.

Given that homo sapiens, the most intelligent species on this planet, has generally made life miserable for all of the other species, I'd like to turn that challenge around: How about a proof that superhuman AI won't harm us?

Suppose a nuclear reactor is being installed in your city. Your friend has reviewed the design and has some serious concerns. Your friend thinks the reactor has a significant chance of melting down. You go to the director of the project. The director says: "Oh that's nothing to worry about. I talked to that guy. He didn't have a mathematical proof that the reactor would melt down." Are you reassured?

Re: Meta disbanded its Responsible AI team

#392
post #389

Earlier quoted context omitted.

Show me a mathematical proof then.

Given that homo sapiens, the most intelligent species on this planet, has generally made life miserable for all of the other species, I'd like to turn that challenge around: How about a proof that superhuman AI won't harm us? Suppose a nuclear reactor is being installed in your city. Your friend has reviewed the design and has some serious concerns. Your friend thinks the reactor has a significant chance of melting d…

No, that's not how this work. You made the claim, so the burden of proof is on you. You're just making speculative statements based on zero scientific evidence and a childish misunderstanding of the basic technology.

Re: Meta disbanded its Responsible AI team

#393
post #392

Earlier quoted context omitted.

Given that homo sapiens, the most intelligent species on this planet, has generally made life miserable for all of the other species, I'd like to turn that challenge around: How about a proof that superhuman AI won't harm us? Suppose a nuclear reactor is being installed in your city. Your friend has reviewed the design and has some serious concerns. Your friend thinks the reactor has a significant chance of melting d…

No, that's not how this work. You made the claim, so the burden of proof is on you. You're just making speculative statements based on zero scientific evidence and a childish misunderstanding of the basic technology.

Did you read this post? https://www.lesswrong.com/posts/uMQ3cqWDPHhjtiesc/agi-ruin-a...

It might require context you don't have. Perhaps try one of these intros: https://www.lesswrong.com/posts/T98kdFL5bxBWSiE3N/best-intro...

Re: Meta disbanded its Responsible AI team

#394

Earlier quoted context omitted.

The issue isn't whether "his point" as you put it is correct. If I said people should safety test the space shuttle to make sure the phlogiston isn't going to overheat, I may be correct in my belief that people should "safety test" the space shuttle but I'm still a crank because phlogeston isn't a real thing.

See my comment here: https://news.ycombinator.com/item?id=38336374 The reason AI alignment is challenging is because we're trying to make accurate predictions about unusual scenarios that we have essentially zero data about. No one can credibly claim expertise on what would constitute evidence of a worrisome anomaly. Jeremy Howard can't credibly say that a sudden drop in the loss function is certainly nothing to worr…

That's not what the conversation was about- you're just doing the thing Howard said where you squint and imagine he was saying something other than he did.

Re: Meta disbanded its Responsible AI team

#395

Earlier quoted context omitted.

I'm a bit baffled as to who or what he is exactly. With no traces of secondary education, employment, or distinguished accomplishments between when I assume he graduated high school in about 1997 and when he started his own institute in 2000 at 21(according to his birthdate, with Wikipedia saying 23, despite contradicting his birthdate). I'll listen to AI concerns from tech giants like Wozniak or Hinton (neither of w…

>neither of which use alarmist terms like "existential threat" "Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war." https://www.safe.ai/statement-on-ai-risk Not sure you're making a meaningful distinction here. - - - Of course we all have our own heuristics for deciding who's worth paying attention to. Credentials are one heuri…

>For example, you could argue that investing in founders like Bill Gates, Mark Zuckerberg, and Steve Wozniak would be a bad idea because none of them had completed a 4-year degree.

You make a great point quoting Hinton's organization. I need to give you that one. I suppose I do need to start following their posted charters rather than answers during interviews. (not being sarcastic here, it seems I do)

The difference between him and Woz or Zuck isn't just limited to them actually attending college, but also the fact that the conditions under which they left departed early can not only be looked up easily, but can be found in numerous films, books, and other popular media while there's no trace of even temporary employment flipping burgers or something relevant to his interest in writing fiction, which seems to be the only other pursuit besides warning us of the dangers of neural networks at a time when the hypetrain promoting the idea they were rapidly changing the world, despite not producing anything of value for over a decade. I'll admit the guy is easier to read and more eloquent and entertaining than those whose input I think has much more value. I also admit that I've only watched two interviews with him and both of them consisted of the same rhetorical devices I used at 15 to convince people I'm smarter than them before realizing how cringey I appeared to those smart enough to see through it, but much more eloquent. I'll give one example of the most frequent one, which are slippery slopes that assume the very conclusions that he never actually justified. Like positing one wrong step towards AGI could only jeopardize all of humanity. However, he doesn't say that directly, but instead uses another cheap rhetorical device whereby it's incumbent on him to ensure the naive public realizes this very real and avoidable danger that he sees so clearly. Fortunately for him, Lex's role is to felate his guests and not ask him why that danger is valid and a world whereby a resource-constrained humanity realizes that the window of opportunity to achieve AGI has passed as we plunge into another collapse of civilization and plunge back into another dystopian dark age and realize we were just as vulnerable as those in Rome or the Bronze Age, except we were offered utopia and declined out of cowardice.

Re: Meta disbanded its Responsible AI team

#396
post #372
post #359

Earlier quoted context omitted.

Hence training an LLM to keep secrets (your apparent practical definition of safety) is as much of an existential risk as not doing it at all? Your position makes no sense, and your arguments are all based on basically false data.

That you’re phrasing it as “training an LLM to keep secrets” suggests a misapprehension: A downloadable LLM that knows secrets fundamentally cannot be expected to keep them. The fixable condition is much weaker: capabilities, if an LLM has the capability to do dangerous things. The problem for secrets in general (which is a separate issue to LLMs, I interpreted you asking me about the EU Chat debate as an attempted g…

But LLMs only have the capability to statistically associate strings of words. That's all they are. There is no other capability possible there.

And you admit that they cannot be expected to keep secrets. So what is the point of trying to have a "security" team hammer secret keeping into them? It doesn't make sense.

I bring up chat control since I've noticed most "AI Safety" advocates are also vehemently opposed to government censorship of other communication technology. Which is fundamentally incoherent.

Re: Meta disbanded its Responsible AI team

#397

Earlier quoted context omitted.

AI alignment is sort of like computer security except computer security is a real pressing problem and AI alignment is based on science fiction.

Suppose you were working on an early mission-critical computer system. Your coworker is thinking about a potential security issue. You say: "Yeah I read about that in a science fiction story. It's not something we need to worry about." Would that be a valid argument for you to make? It seems to me that you should engage with the substance of your coworker's argument. Reading about something in science fiction doesn't…

In this analogy it's not your coworker. It's some layman (despite self-declared expertise) standing outside and claiming he's spotted a major security issue based on guesses about how such systems will work

From what I have observed the reaction of most people working in the AI to "What do you think of Yudkowsky?" is "Who?". He's not being ignored out of pride or spite, he just has no qualifications or real involvement in the field

Re: Meta disbanded its Responsible AI team

#398
post #34

Because Meta is releasing their models to the public, I consider them the most ethical company doing AI at scale. Keeping AI models closed under the guise of “ethics”, is I think the most unethical stance as it makes people more dependent on the arbitrary decisions, goals, and priorities of big companies, instead being allowed to define “alignment” for themselves.

Exactly. I can't speak about meta specifically, but from my exposure "responsible ai" are generally policy doomers with a heavy pro-control pro-limits perspective, or even worse-- psycho cultists that believe the only safe objective for AI work is the development of an electronic god to impose their own moral will on the world. Either of those options are incompatible with actually ethical behavior, like assuring tha…

Yes, moral Puritans just in a new modern form.

Re: Meta disbanded its Responsible AI team

#399
post #389

Earlier quoted context omitted.

Show me a mathematical proof then.

Given that homo sapiens, the most intelligent species on this planet, has generally made life miserable for all of the other species, I'd like to turn that challenge around: How about a proof that superhuman AI won't harm us? Suppose a nuclear reactor is being installed in your city. Your friend has reviewed the design and has some serious concerns. Your friend thinks the reactor has a significant chance of melting d…

Thats unreasonable as we all know its impossible to prove a negative, especially about a as yet hypothetical in a new feild of research.

As for a nuclear reactor in my city, yeah if my friend doesnt have qualifications to make him capable of evaluating the designs of such a technical and esoteric field and someone who is qualified assured me it was fine I would trust them. If we dont trust experts in their fields about their field then we are no better intellectually than the antivaxers and flatearthers

Re: Meta disbanded its Responsible AI team

#400
post #361

Earlier quoted context omitted.

You keep throwing examples of weapons of mass destruction, meant to evoke emotions. For better or worse, nuclear weapons have been democratized. Some developing countries still don't have access, but the fact that multiple world powers have nuclear weapons is why we still haven't experienced WW3. We've enjoyed probably the longest period of peace and prosperity, and it's all due to nuclear weapons. Speaking of, Cold…

Nuclear weapons have absolutely not been democratised, and still stand as a technology that has largely been restricted and not proliferated. Only 9 countries in the 190 or so out there currently nuclear weapons, and 3 countries (South Africa, Ukraine, and Kazakhstan) decided that maintaining stockpiles was more trouble than it was worth. Huge effort has been made to keep nuclear weapons out of the hands of non-state…

To be fair many countries dont want or need them or rely on other allied countries that do have them so they dont have to feild the expense. Other have them but dont officially admit to it. Some cant afford them. But a basic gun type nuclear bomb design like little boy is not technically difficult and could be made by a highschool ap physics class given access to the fissionable material.
Post reply on HN