All of this reeks as smugness and superiority, as if they are the sole arbiters of truth and they are the only adults in the room. Screw these people.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
OpenAI Employees Warn of Advanced AI Dangers
- Thread starter MacRumors
- Start date
- Sort by reaction score
Right, it all reeks of gatekeeping. God forgot Billy in his basement makes an AI that they can’t control. Think of all the bad things that can happen, that would be bad, we don’t know what those bad thing are, but it sounds bad, so it’s probably bad….
I think training the LLM is the most expensive part. Once that’s trained cost isn’t too much of a factor. Overall the investment in AI is coming down quite rapidly every year.
Good data collection and processing will be the most significant expense.
the f'k does elon have to do with any of this? what credit does he deserve for echoing other peoples thoughts? put down the koolaid
This is the thing with AI at the moment. When you generate something, whose dataset are you using? Companies that invest heavily into their own datasets won't be sharing them and instead using them for their own competitive advantage.
Easier said that done.
Also, they’re saying extinction is a possible outcome here. Not a sure thing. The point is they’re trying to wave their hands to get attention to this subject, because it is kind of insane that governments around the world are basically watching as these AI companies do whatever they want.
Politicians generally don’t do much about a problem unless they’re told it could become a crisis. If you wait til that point it’s too late.
Lol - AI today is only as good as the person that is using it.
A general intelligence that could mimic a human is decades if not longer away.
This. All this “we’re terrified of our own product! It’s going to take over the world!” talk sure sounds like a lot of marketing to me.
Yes, but people are making the mistake of calling this “extinction” or a “takeover”. While it’s possible some stupid military lets ai control its nukes and it ends us all “including the ai as without power it’s dead too”.
What a “super intelligent” ai would be, if it’s possible to actually make one is just the next step in human evolution. We made it, with all our knowledge, we leave behind our biological meat suits for mechanical ones. It will make space travel a lot easier when we don’t have to worry about oxygen, food, gravity, etc.
I don’t think it’ll happen though as we seem hell bent on going extinct long before Skynet kills everyone.
It appears that way at first glance, but it's not quite the case. The "what will we ever do about our crazy amazing and terribly dangerous product!!" rhetoric is a tactic to angle for cushy positions in AI “alignment” or to secure even more funding by speeding FUD.
These doomsday narratives consistently manipulate attention and investment, not because there's an imminent crisis, but to bolster their own influence and financial backing.
HobeSoundDarryl
macrumors G5
It started nearly a decade ago, with certain websites pushing clickbait headlines. Now, unfortunately, it’s practically unavoidable on the Internet, since the culture of fear, outrage, and half-truths or misinformation designed for little more than engagement have enveloped our society, and turned us all against each other. Also, the use of superlatives in everyday life, especially in the news media, has gotten majorly out of hand, and contributed to the modern culture where nobody accepts “good enough” or level-headedness.