Frontier AI labs have talked about the possible negative effects of their inventions for years, and how we need to think about fundamentally changing the social contract to accomodate their technology.
Recently though, the language out of frontier Labs has taken an even darker turn. OpenAI won’t stop humblebragging about how their models are so extremely powerful that they can escape any sandbox and hack practically anything. In 6 months, the entire internet could be taken over by these things!
Over at Anthropic, the picture painted is no more rosy. In fact, people at Anthropic seem to believe that there is a 10% chance that it could bring about the end of civilization, and even their leaked IPO prospectus warns about the extreme dangers of the technology.
All we hear, and have ever heard, is how crazy and dangerous this technology is, how it could ruin the lives of millions of people all over the world and now they even claim it could bring about the apocalypse.
They are busy burning so much capital that it would make god blush, to create a civilization-ending machine deity that will bring about the downfall of the human race. No expense is spared. All of the risk-willing capital of the earth is being mobilised for this project. The AI buildout is truly a modern day Manhatten Project. They are doing all of this, while telling us how incredibly dangerous this all is. Why are people not more scared?
Maybe it’s because most people are normal, benevolent, kind-hearted individuals who don’t want to harm anyone, especially not the entire human race. If a normal human being was working on a project, and they realised there was a 10% chance it would destroy human civilization, they would simply stop working on it. Who wants to gamble the existence of the human race on a roll of the dice? You would have to either be
- Stupid
- Evil
- Or both
In order to continue with such as a project. No reasonable moral human being would ever risk such a thing. And so if you assume that the people working in these frontier labs are also normal human beings, then what they are saying simply doesn’t align with their actions, because who the fuck who would build a doomsday machine?
No matter how you slice and dice this thing, what the frontier labs are saying is damning. Either they are lying bullshit artists drumming up hype for their product with apocalyptic hyperbole, or they are cynically exploiting people’s fears to get regulators to sabotage their competitors, or they are cartoonishly evil and/or stupid. I think the real answer is a combination of these.
I think these people are so caught up in their own hype (and AI psychosis) that they don’t properly appreciate how ridiculous their rhetoric looks to a normal person, because a normal person can’t comprehend why anyone would willingly build a doomsday device. Because you know, why would you?
Why would you, frontier AI labs?