undefined | Better HN

0 pointsnext_xibalba26d ago0 comments

> This is LITERALLY the plot line of Terminator. Literally.

No, it's not. Skynet was a recursively self improving ASI. You are conflating an autokill bot and, apparently, an ASI that can embody and replicate itself.

> If giving the ukranians nuclear warheads could help them default Russia, then isn't that good?

Surely, you can recognize how an autokill bot and a thermonucelar weapon are different, right? These are categorically different concepts. What's more, Russia is a nuclear armed opponent with, reportedly, dead man's hand systems that would launch their entire nuclear arsenal even if their command structure is destroyed in a nuclear first strike.

I'll just repeat the basic point here: autokill bots are coming. Whether any of us like it or not. Just like nuclear weapons. If I could wave a magic wand and eliminate all weapons of mass destruction in the world, I would. But that's not reality. So, walk me through how you think this plays out if we don't develop them, but Russia, China, etc. do?

I can't think of a more clear cut case of moral, justified deployment of autokill bots than to aid Ukraine in expelling the Russian invaders.

0 comments

array_key_first26d ago

> No, it's not. Skynet was a recursively self improving ASI. You are conflating an autokill bot and, apparently, an ASI that can embody and replicate itself.

It never said it was any of that. The point of terminator is that decisioning around war was taken out of the hands of humans, and then nobody could control it.

You people really don't get it do you? Skynet doesn't need to be evil, or conscious, or self improving. It can be good, very good. But when WE don't control it, we don't know the consequences of what we created. Nobody saw AI psychosis coming but we created it, by making the models good. By making the models listen to you and agree with you.

For fucks sake, you could make an automated system that just signs postcards and, if you give it enough access, it could wipe out the human race. Not because it's evil, it might not even have an understanding of evil, but because we don't control it, and it will meet it's own goals without concerns for us because it's not human.

> autokill bots are coming. Whether any of us like it or not.

Inevitability is not an argument, and I won't humor it. It's cognitively lazy and dishonest. With this reasoning you can justify ANYTHING. Rape, murder, nuclear warfare, killing and eating children. This reasoning is bad and stupid and nobody should do it anymore.

next_xibalbaOP26d ago

> you could make an automated system that just signs postcards and, if you give it enough access, it could wipe out the human race.

I mean this sincerely. You really ought to stop reading Bostrom and Yudkowsky. It is very hard to take this kind of hysteria seriously.

> Inevitability is not an argument, and I won't humor it.

It is and I don't care what you will or won't humor. Just answer me this: how will you convince all the other countries of the world to not build terminators? The leading example of "it is inevitable" is of course China. They are already testing and deploying semi-autonomous robots throughout their national security apparatus. If you're answer is: "Just because they do it doesn't mean anyone else should" then you're not to be taken seriously on this topic.

> killing and eating children

I'd really like to know what convoluted scenario you could conjure in which one would argue that killing and eating children is inevitable.

array_key_first25d ago

Saying something is hysteria is also not an argument. Again, it's just intellectually lazy. Just because you refuse to take problems seriously doesn't mean they cease to exist, it just means you lack critical thinking.

And, as for eating and killing children, it's easy: starvation. If you're hungry enough you'll eat children. All it takes is a supply chain disruption, much more likely than nuclear war even.

So why not eat the children now? It's gonna happen anyway.

It's true that I am jumping the gun here. We don't need an apocalypse for AI to suck ass. It sucks ass right now and is causing massive problems. We should probably focus on that.

1 more reply

j / k navigate · click thread line to collapse