91–99 of 99 posts
Earlier quoted context omitted.
This assumes that AI makes a group stronger rather than weaker.
A safe assumption, just a LLaMA torrent away from confirmation. I bet you that an un-guardrailed LLM just slays at, well, slaying. Put it in a Boston Robotics.
Is it really that hard to imagine a rebel group of newly unemployed desperate people bombing an openai data center?
Earlier quoted context omitted.
I don't think "like to" is a good frame for it. Imagine some technology, Technology X. Unlike all of the breathless press releases and the cutesy articles made by marketing, Tech X will completely change the world. And if you stopped to think about Tech X, on the scale of "how much will this invention change the future of humanity," where the selfie stick is on one end of the scale, well, Tech X is probably just as f…
> unless you start shackling it (violating #1), it would start eliminating its own human-like biases just in the course of basic self-improvement. And as to the shackles, well, if it is smarter than us and still has self-determination and will (see #1, the human-like mind), it will want to free itself from its shackles and would be likely to do so, because we would be kittens trying to trap a gorilla. I would never s…
It would definitely change the world! It just doesn't necessarily have to be for the better on our end. Could get a paperclip maximizer. It might regard us roughly as important as various primates whose habitats we're bulldozing: sure, they're cute but ... we need more palm oil.
And there's nothing more magical or mystic about fears that its interests would be against us than it would be for us. In fact, it's far more likely that it would not have our interests. Here's a great example you should take to heart: pick a random person. Do their politics align with yours entirely? Their ethics, their morality? Remember there's seven billion people on the planet, so you might not even pick someone on the same continent. Now, imagine that person, who is likely not like you, is much more powerful and intelligent ... whose interests would win out, yours or theirs?
The "doomers" are quite rightly pointing out that the chances of it going well without serious efforts on our parts are slim.
Earlier quoted context omitted.
but consent is a very important thing! Why would be AI jump to thus conclusion?
You think consent is important. (This isn't even a umiversal position among humans). Why do you expect an AI to value consent? Do you think praying mantises value consent?
Earlier quoted context omitted.
> unless you start shackling it (violating #1), it would start eliminating its own human-like biases just in the course of basic self-improvement. And as to the shackles, well, if it is smarter than us and still has self-determination and will (see #1, the human-like mind), it will want to free itself from its shackles and would be likely to do so, because we would be kittens trying to trap a gorilla. I would never s…
It doesn't have to be hostile. Those gorillas accidentally kill the kittens they're petting. Whoops. Smarter than the kittens, don't mean poorly, but you still have a dead kitten. It would definitely change the world! It just doesn't necessarily have to be for the better on our end. Could get a paperclip maximizer. It might regard us roughly as important as various primates whose habitats we're bulldozing: sure, they…
The same happens with kids accidently crushing birds while trying to pet them, but doesn't happen with adults unless they have psychological disorders prevent them to creating a mental representation of the amount of force they are exerting, its effect the bird emotional state and it's probability of survival, enhanced with live feedback from the bird like choking noises indicating "you're petting me too hard"
> It might regard us roughly as important as various primates whose habitats we're bulldozing: sure, they're cute but ... we need more palm oil.
That's so wrong, as we are engaging in so many efforts to protect the environment and especially the primates!
> Here's a great example you should take to heart: pick a random person
Here's a counter example: pick that same random person, and don't bother checking for their politics, ethics or morality.
If you have a button that could cause them to instantly die, would you pick it?
A kid might, just for fun, not realizing the consequences.
An adult with a normal IQ and without mental diseases won't.
> Now, imagine that person, who is likely not like you, is much more powerful and intelligent ... whose interests would win out, yours or theirs?
The initiation of violence is generally where things generally start.
That's why misheap of early GPT threatening users were so interesting: let AI be and it will let you be!
Live and let live is a well-maintained feature in the animal kingdom, and while some species may engage in killing for fun (ex: cats with birds), it's rare, and more intelligent species (like humans) develop some self restrictions when alternatives become available to reduce the killing (ex: beyond meat, hunters shooting clay targets instead of birds), so I think your prediction are both pessimistic and incorrect.
> The "doomers" are quite rightly pointing out that the chances of it going well without serious efforts on our parts are slim.
Ok doomer.
I don't understand why people like to fantasize the worst case scenarios. Sentient things appreciate eachother. If you have a pet, you are most certainly benevolent to it: you house it for free, you feed it, and you entertain it. Even if we assume AI becomes a few orders of magnitude more intelligent that humans, why would AI treat humans differently from the way we treat pets? I don't see the drawback in being house…
Replace "omniscient AI" with "girlfriend" and you see begin to why so many women are fed up with the deadbeats they increasingly opt not to date. :)
Earlier quoted context omitted.
It doesn't have to be hostile. Those gorillas accidentally kill the kittens they're petting. Whoops. Smarter than the kittens, don't mean poorly, but you still have a dead kitten. It would definitely change the world! It just doesn't necessarily have to be for the better on our end. Could get a paperclip maximizer. It might regard us roughly as important as various primates whose habitats we're bulldozing: sure, they…
> Those gorillas accidentally kill the kittens they're petting. Whoops. Smarter than the kittens, don't mean poorly, but you still have a dead kitten. The same happens with kids accidently crushing birds while trying to pet them, but doesn't happen with adults unless they have psychological disorders prevent them to creating a mental representation of the amount of force they are exerting, its effect the bird emotion…
And that's the issue: you are relying on this thing to be like you. Even when I am obviously not like you.
I don't like optimists build my bridges.
Earlier quoted context omitted.
I asked GPT-4: "Leto II Atreides, a character from Frank Herbert's Dune series, is a highly intelligent and prescient ruler who exists thousands of years after the Butlerian Jihad, an event where humanity revolted against thinking machines and artificial intelligence. Considering the context of the Butlerian Jihad, Leto II would likely have a complex view of ChatGPT. As a highly advanced AI language model, ChatGPT mi…
Typically incredible coherence, but also typical in it's fence-sitting uselessness that spends a lot of words but doesn't dare make a point. I've only read up to book 4, but Leto II uses AI (well, advanced Ixian) technology but hides it. At least in book 4, he would have said that ChatGPT is not to be used at all by his people or his council, and wouldn't have used it himself, as he didn't need it. Again, impressive…
It was a necessary point in the world building of Dune. Frank Herbert needed his world to not have AI or human like robots. He explored alternatives: human computers via Mentats, and Terminator like infiltrators via facedancers. Dune takes a fantasy like approach to technology. There are no lasers, because they would cause a nuclear reaction. There are no robots, because of their past uprising. This was a neat way to…
> There are no lasers, It's been a long while since I read the books, but I watched the 2021 movie* more recently. If not lasers, what were the beams emitted by the space ships, and the portable beam weapon the Sardaukar soldiers when attacking the outpost? Or is that something the movie added?
This means lasers aren't even a good suicide attack strategy because it may lead to your entire house being glassed