re: If You Call it Artificial Intelligence You Are the Enemy

*Woosh*

(Your head)

quote:

---

*Woosh*

(Your head)

---

What went over my head? The fact that I was agreeing with you?

quote:

---

Super Intelligence is already defined as a hypothetical intellect that greatly surpasses the cognitive performance and capabilities of the best human brains across virtually all domains of interest, including scientific creativity, general wisdom, and social skills.

---

Right. And what I'm saying is that that definition describes a hypothetical thing we have never, and may never, reach. At the same time, calling it artificial simply is not accurate, no matter how much people might not want to hear that.

So he rebranded it. Big frickin deal. The idea here is to try and get the American people to sorta rally behind it. And they should. The technology is incredible. Does it have its risks? Sure, but we can manage those and the benefits of the technology is off the charts. It absolutely is something the people should support as a national endeavor, winning the race to SI supremacy, and considering the term "artificial" doesn't accurately describe it and the term Super intelligence, as it was previously known, was a hypothetical state we may never reach anyways, I support the rebrand.

So yeah, I call it SI because I'm both a fan of accuracy and I am an American Patriot. I agree with what Trump is doing here. And honestly, those who don't like it can also lick my taint.

I don't think they have created an SI god. ... Yet.

quote:

---

The semantics are fun to discuss

---

Funny enough, this is the thing I'm having the most trouble with with mine. Getting the semantics right. It's sometimes hard to tell if your bots are making inferences or if your inputs are flattening out.

A good read on a subject many of us have been sorta tracking for years now, but what does that have to do with this thread? I'm not trying to be overly critical here, I'm just curious if you have an angle on that that pertains to the rebranding of AI into SI.

quote:

---

A good read on a subject many of us have been sorta tracking for years now, but what does that have to do with this thread? I'm not trying to be overly critical here, I'm just curious if you have an angle on that that pertains to the rebranding of AI into SI.

---

Re-read the quote.

quote:

---

Jews threatened to frick with his money

---

quote:

---

Austrianpaynter

---

For fricks sake. Go jump off a frickin bridge or something, you frickin retard. Spare us your stupidity.

quote:

---

Reading through your comments, you seem to have a significant grasp on the subject at hand. With that said, what is your opinion about the use of Ai or whatever level you mentioned, as it pertains to the military’s use of it, specifically things such as threat evaluations and kinetic weapons targeting? Is there a point where it is trusted even if there are extremely small margins for error?

---

I honestly don't know. Things change so quickly anymore.

I'm assuming you are talking about to a point where humans wouldn't be needed for the final decision.

You'll need a target identification thing, and that would likely be a convolutional network. That's used for processing images. So there is a degree to which it could have false positives. That's kind of like a driving AI. Where it's just processing images and then making classification decisions. Is that a tank or person is a classification issue.

You can improve the amount of false positives with multiple systems. IR, Radar, multiple frames of positive confirmation with different angles etc.

At that level, which I'm guessing we are probably already at. The targets would just be presented to a human. I suppose you could have it just fire away when it detects the target and then it's a matter of classification quality.

But then if you mean more strategical strikes rather than something in battle already, they'd probably use LLMs to process the data beyond the classification. Still use the CNN's, but then a LLM would take the data and process it over. Maybe one target is better than another, maybe you want to strike in a certain order, etc.

So yeah I don't know. There is a lot there and high stakes consequences.

And in terms of a LLM going rogue, I'm not sure. I would think they would be pretty focused in their training. I don't generally worry about rogue AI for that reason, most AI is going to be finalized models designed for specific tasks. They don't have the capability. So yeah you can probably trust them in that capacity.

And that's also at the same time what scares me. I don't fear AI, I fear what people might do with AI - purposely or not. And the AI will never change because it's on a finalized model and incapable of it.

That said, humans make mistakes too and I think there are laws and rules when it comes to trusting AI to make those decisions.

quote:

---

The hard problem is how do you get the benefits of super intelligence without the inherent catastrophic risk.

---

Perhaps a workable solution might be to use a simpler, easily controllable Ai to manage the SI (if that isn't already being done.)

quote:

---

Perhaps a workable solution might be to use a simpler, easily controllable Ai to manage the SI (if that isn't already being done.)

---

It's definitely being done. And not talking about nefarious activity here or anything, but using multiple SI bots is how you keep them honest, on task, and accurate.