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

After OpenAI's blowup, it seems pretty clear that 'AI safety' isn't a real thing

gizmodo.com

Post reply on HN
Print view View original

41–50 of 392 posts

Re: After OpenAI's blowup, it seems pretty clear that 'AI safety' isn't a real thing

#41

Post by zone411 »

okdood64 wrote: post #25

Something I see lacking in many discussions about AI Safety and Alignment is: What does the role of China mean for AI Safety? Is there any reason to believe that they will a) adopt any standards SV, the US or some international government body will create, or b) care enough to slow down development of AI? If AI Safety is really a concern, you can expect in the long run (not that much further behind than the US) China…

Indeed. I'm concerned about the safety of future AGI systems, but China is why I wouldn't advocate for a unilateral pause or slowdown.

Re: After OpenAI's blowup, it seems pretty clear that 'AI safety' isn't a real thing

#42

Post by Aloha »

golol wrote: post #15

Look, there is ASI/AGI safety, and "please don't swear and scam and misinform, and be unbiased" safety. ASI/AGI safety is only a concern right now if you believe in foom , i.e. that an AGI would rapidly self improve to an ASI and extinction level threat. I think it is absurd to believe that because AI right now is compute limited and any first AGI will only be able to run on large data centers. The latter kind of saf…

We should in general terms not give AI - any AI, the ability to manifest itself in meat space. This alone almost certainly reduces the existential risks from AI.

Re: After OpenAI's blowup, it seems pretty clear that 'AI safety' isn't a real thing

#43

Post by ekidd »

gustavus wrote: post #5

I've always interpreted AI "safety" as being more about ensuring ideological consistency and a particular narrative than anything about actually keeping humanity "safe". And I'm not even talking about the culture war, just yesterday I tried having DALLE generate a picture for a DND character and it kept refusing to (and no it wasn't some sort of weird hentai thing) because it felt like it was not an appropriate image…

> And I'm not even talking about the culture war, just yesterday I tried having DALLE generate a picture for a DND character and it kept refusing to (and no it wasn't some sort of weird hentai thing) because it felt like it was not an appropriate image.

I think this may be a problem in DallE's training data.

I tried using it to generate character portraits for a local gaming group. And even when it did generate images of female characters, nearly half of them were NSFW. It took quite a lot of prompt engineering to minimize this, and I still wasn't happy about the results. Oh, and DallE kept getting upset at me for its own policy violations.

Other than that, DallE 3 has been terrific. But I think you're hitting some kind of significant failure mode.

> anything about actually keeping humanity "safe".

I've said this before and I'll say it again—lots of the people working on this stuff actually do fear that we're getting close to a point where we might build an actual working AI. Whether or not they're correct is another question, but many of them do fear that.

Also, early pre-release versions of GPT4 were apparently pretty disturbing. I've heard them compared to talking to Hannibal Lector.

Re: After OpenAI's blowup, it seems pretty clear that 'AI safety' isn't a real thing

#44

Post by shrimpx »

AdamJacobMuller wrote: post #34
insanitybit wrote: post #11

Earlier quoted context omitted.

What damage will it cause? I feel like there's "AI will replace jobs" level of damage, which, why would we regulate that? An insane amount of technology is developed with the purpose of improving productivity or outright replacing workers. Then there's the "AI will go rogue", which I don't think is substantiated at all. Like, one can theorize about some novel, distinct system that is able to interact with the externa…

AI has already evolved to the point where it is impossible to predict with any accuracy it's answers to inputs. As AI continues to evolve and continues to become more connected to society and as AI becomes self-improving with far larger context windows, a few things do not seem implausible to me. AI will have the ability to earn money. This is already doable with minimal human interaction. It wouldn't take much for A…

> AI has already evolved to the point where it is impossible to predict with any accuracy it's answers to inputs.

Btw neural nets have always been black boxes, with ppl scratching their heads wondering what’s in those weights.

Re: After OpenAI's blowup, it seems pretty clear that 'AI safety' isn't a real thing

#45

Post by skilled »

I think the approach OpenAI has taken is the best one so far. Do you really want to have an assistant that is as brainwashed as Pi or Claude for example? ChatGPT has a flag feature that will say “this might violate our policy”, but it will still continue to do what you tell it to do unless you’re doing seriously stupid stuff.

But on a broader spectrum, like AI being able to manufacture a secret poison recipe… you will never get access to that. And we are hardly in the vicinity of something like that happening soon.

Let them do their thing and keep the model confident while they figure out how to keep it in check when it comes to harmful queries and that’s it.

Why overcomplicate it?

Re: After OpenAI's blowup, it seems pretty clear that 'AI safety' isn't a real thing

#46

Post by nradov »

Animats wrote: post #13

From the article: "If the point of corporate AI safety is to protect humanity from runaway AI, then, as an effective strategy for doing that, it has effectively just flunked its first big test." That's a good point. As I point out occasionally, the big near-term risk from AI is that it makes it possible for corporations, and governments, to do obnoxious things they already do, but can't yet do cost-effectively at sca…

The markets for publicly traded stocks, bonds, and related securities are already highly efficient with humans (and human developed algorithms) doing the trading. Pretty much any new information is priced in within seconds. Any hypothetical new investment AGI could achieve only a marginal improvement over the current situation; it wouldn't dramatically change the dynamics of public capital markets.

Re: After OpenAI's blowup, it seems pretty clear that 'AI safety' isn't a real thing

#47

Post by andrewprock »

shrimpx wrote: post #35
xt00 wrote: post #29

AI safety allows AI companies to tell governments and worried people to not worry too much while they plow ahead doing what they want. That’s not to say there is nobody at these companies actually being responsible but having to open up your company to the equivalent of nuclear watchdogs is not something companies would be excited about. Just like nuclear weapons and nuclear energy we benefit from the tech as well as…

The crazy thing is people at large didn’t know anything about “AI safety” until these very companies started peddling the concept in the political sphere and media. The very companies who were doing the opposite of this concept they advertised so much and whose importance they stressed much.

The crazier thing is that most advocates of "AI safety" don't know anything about "AI safety". By and large the term is marketing with very little objective scientific development.

It is essentially a series of op-ed pieces from vested interests masquerading as a legitimate field of inquiry

Re: After OpenAI's blowup, it seems pretty clear that 'AI safety' isn't a real thing

#48

Post by adrians1 »

not_your_vase wrote: post #2

But is there a real AI-threat, that would need real AI-safety? To me it still looks like a nice bar-trick, not AI. It is very clever, very nice, even astounding. But doesn't look threatening - at some cases it is even dumb. The fear-mongering looks more like scare-marketing. Which is also clever, in my opinion. But maybe I'm just out of touch.

Even if AI is a threat, it's impossible to predict today how the threat is going to look in the future. The "AI safety" crowd watched too many Terminator movies.

We'll deal with AI the same way humans dealt with any other technology: gradually, as it gets better and better, discovering at each step what it's capable of.

Re: After OpenAI's blowup, it seems pretty clear that 'AI safety' isn't a real thing

#50

Post by g42gregory »

okdood64 wrote: post #25

Something I see lacking in many discussions about AI Safety and Alignment is: What does the role of China mean for AI Safety? Is there any reason to believe that they will a) adopt any standards SV, the US or some international government body will create, or b) care enough to slow down development of AI? If AI Safety is really a concern, you can expect in the long run (not that much further behind than the US) China…

My feeling is that China will not adopt anything, inside their country, that can jeopardize the speed of their AI development progress. I suspect that China will also support and promote corresponding regulations in the Western countries, thinking that it will slow down the progress. I maybe wrong but I think official policy documents say that China is strategic rival number one, or something to that effect. I suspect China will act accordingly.
Post reply on HN