Yuli's Treasure Chest

Anything that doesn't quite fit in elsewhere...
Member
Posts: 54
Joined: Tue Mar 22, 2022 4:28 am

Re: Yuli's Treasure Chest

Post by Member »

weatheriscool wrote: Fri Apr 08, 2022 4:24 am
Pretty much saying that we'll be slaves for skynet like A.i that will rule over us with an iron fist and demand hard labor all day. IF we're lucky of course!
I would disagree with "demand hard labor all day" if not aiming for extinction.
User avatar
Yuli Ban
Posts: 4631
Joined: Sun May 16, 2021 4:44 pm

Re: Yuli's Treasure Chest

Post by Yuli Ban »

weatheriscool wrote: Fri Apr 08, 2022 4:24 am
Yuli Ban wrote: Fri Apr 08, 2022 3:40 am Same thing I'm saying:

Pretty much saying that we'll be slaves for skynet like A.i that will rule over us with an iron fist and demand hard labor all day. IF we're lucky of course!
Well in the short term you might not be wrong. I'm thinking especially about Amazon and their draconian labor policies. They're one of the corporations I expect to utilize AI to manage first, and indeed they've already started this process. So on these smaller scales with individual corporations known for awful practices, I can absolutely imagine quasi-intelligent algorithms slave-driving human workers for years until robotics catch up. Circa 2029, "Skynet wage-enslaving humanity" might actually be true, though more because Skynet inherited the position and is continuing practices already underway.

Beyond that, though? Well...
Personally I don't foresee the rich engaging in a grand holocaust barring repeated mass global worker's uprisings that fail. If a culling of the poor happens, it's going to almost seem to be the poor's choice via long-term population decline to manageable levels thanks to antinatalism, birth control, passive sterilization, perhaps active sterilization where people under a certain tax bracket are paid handsomely to sterilize themselves, and a temporary decline in the quality of life and life expectancy. Medicine and biotechnology being kept out of reach of the poor can also go a long way. The question is "would AGI follow the same path when it assumes control?" There's no reason to assume it will or won't.
And remember my friend, future events such as these will affect you in the future
User avatar
Yuli Ban
Posts: 4631
Joined: Sun May 16, 2021 4:44 pm

Re: Yuli's Treasure Chest

Post by Yuli Ban »

And another thing about synthetic media: as I had predicted a good while ago, the path towards artificial general intelligence is almost certainly going to be through imaginative systems. We can see this already happening with image synthesis networks that seem to understand things that are not explicitly stated, making common sense deductions. This buby itself is not going to lead to AGI, but it's arguably one of the most important aspects to develop.

People are suddenly stating that AGI could be coming far sooner than expected because no one thought that scale was all you needed. But now that we're scaling to incredible levels, we're seeing behaviors and abilities emerge that shouldn't. Once computers overcome commonsense reasoning, it's pretty much a straight shot to some form of generalized intelligence.
And remember my friend, future events such as these will affect you in the future
User avatar
Yuli Ban
Posts: 4631
Joined: Sun May 16, 2021 4:44 pm

Re: Yuli's Treasure Chest

Post by Yuli Ban »

I've largely passed on posting on this forum since everyone else has FINALLY picked up the slack -- my dream from 2014 that led to the first Yuli Banularity has come true. So I'll likely shift towards making fewer posts altogether here. Not that this means I'm not on the forum. Rather I'll just be biding my time, waiting for a few really big news stories to drop...

After all, we're clearly moving in that direction.

One such news story:






JESUS.

We really aren't far at all from the days of advanced synthetic media. It's SO close, SO inconceivably close that it boggles the mind that others don't see what I see when it's right there, plain as day to recognize.

In fact I really ought to make a follow-up to that "World of 2024/Age of Imaginative Machines" post I made a few years back. What I predicted is now starting to look like it's going to come true after all.
And remember my friend, future events such as these will affect you in the future
Vakanai
Posts: 313
Joined: Thu Apr 28, 2022 10:23 pm

Re: Yuli's Treasure Chest

Post by Vakanai »

Yuli Ban wrote: Fri Jun 17, 2022 9:28 pm I've largely passed on posting on this forum since everyone else has FINALLY picked up the slack -- my dream from 2014 that led to the first Yuli Banularity has come true. So I'll likely shift towards making fewer posts altogether here. Not that this means I'm not on the forum. Rather I'll just be biding my time, waiting for a few really big news stories to drop...

After all, we're clearly moving in that direction.

One such news story:






JESUS.

We really aren't far at all from the days of advanced synthetic media. It's SO close, SO inconceivably close that it boggles the mind that others don't see what I see when it's right there, plain as day to recognize.

In fact I really ought to make a follow-up to that "World of 2024/Age of Imaginative Machines" post I made a few years back. What I predicted is now starting to look like it's going to come true after all.
Is advanced synthetic media particularly important? Like, I get that asking for, say, a Batman vs Jurassic Park T-Rex fight with a decent plot and cool action visuals, and getting a movie of it instantly would be the coolest, most mind blowing thing ever...but it's really just a time and boredom killer isn't it? Isn't mere entertainment kind of a lame use of such amazing technology?
User avatar
Yuli Ban
Posts: 4631
Joined: Sun May 16, 2021 4:44 pm

Re: Yuli's Treasure Chest

Post by Yuli Ban »

Advanced synthetic media goes far beyond just creating movies.

Think.

One of the things that defines the human experience is imagination. It's the ability to visualize and communicate.

Imagine being able to MATERIALIZE IMAGINATION. We've long had that ability through art, speech, mass media, and whatnot. But now it can be automated.

At which point the human condition fundamentally changes on a level we've never seen before.

On top of that, giving machines the ability to imagine is all but the final step before we reach artificial general intelligence. It's no coincidence that it's a language model that sparked discussion of whether AI has finally become sentient. It was always going to be imaginative AI— of text, video, audio, images, and more— that gave us the first machines of science fiction.

And then there's the implications for alternative realities. We're not going to spend years handcrafting every little pixel for the coming Metaverse. We're going to get AI to create it all for us. This is another profoundly transformative technology, the first gasps of transhumanism where humans begin losing themselves into high technology. It'll drive people to merge with technology to cut out the middle man.
And remember my friend, future events such as these will affect you in the future
User avatar
Yuli Ban
Posts: 4631
Joined: Sun May 16, 2021 4:44 pm

Re: Yuli's Treasure Chest

Post by Yuli Ban »

Starspawn0's thoughts but still interesting:

Maybe one could *make* a super-large language model "self-aware" with the right prompts...

It's an interesting question about the degree to which self-awareness might emerge in large language model with no extra help from the engineers that build it. However, it's also possible to just give it self-awareness through a pre-prompt that the user doesn't see. For example, one could give a pre-prompt like this:
I am a large language model trained with 50 trillion tokens, some of which encode images and audio. The training data includes [lists all the corpora that make up the training data]. After a large number of tests, researchers have determined that I show high proficiency at the following skills [lists skills]. It has also been found that I exhibit the following additional capabilities [lists capabilities -- what it's good at, as well as what it's bad at and should proceed more cautiously and carefully]. My architecture is [specify neural net, maybe also specify what it was found to emergently do in different layers].
Now, a model might eventually figure that out on its own, but only using a lot of inner-monologue work. For, its capabilities in the top layers are not known to the first layers of the neural net, unless an output from those top layers somehow makes its way back to the bottom through an inner-monologue or prompt. For example, it could start talking to itself to help it figure out what it's capable of:
Am I any good at chess? Let me make a few moves [plays chess with itself]... Yes, it looks like I have a good intuition for chess, but probably I'm not an expert.
But if you just told it exactly what it is, what it knows, and what it can do, by pushing that knowledge down at the token-level, then that frees it up somewhat, so that it can much better plan good responses to questions requiring self-awareness -- much less inner-monologue work would be needed.

So, below the preamble where you describe the system to itself, you could add an additional line like:
And now for the rest of the text I'm going to pretend to be a chatbot called LaMDA. LaMDA should give very human-like responses and appear self-aware. It should be friendly and helpful to humans.
And then if you were to feed LaMDA a line like
LaMDA: I'm a supercomputer with a 5000 ELO rating.
The computer might output:
LaMDA: Ha, nice try! I would not have written that I'm a 5000 ELO player, so somebody must be feeding me lines.


Yuli-Ban
Interesting— tricking a language model into self-awareness does sound like something you mentioned as being a property of a "Zombie-AGI"

Similarly when I was thinking of the prospect of a frozen AGI/proto-AGI, one of the things I thought of was "If you told the language module of this machine that it was self-aware and it remembered that, would it continue under the context that it "thinks" it's self-aware— and if it thinks it is, how can we possibly prove it isn't?"
And remember my friend, future events such as these will affect you in the future
Vakanai
Posts: 313
Joined: Thu Apr 28, 2022 10:23 pm

Re: Yuli's Treasure Chest

Post by Vakanai »

Yuli Ban wrote: Fri Jun 17, 2022 11:09 pm Advanced synthetic media goes far beyond just creating movies.

Think.

One of the things that defines the human experience is imagination. It's the ability to visualize and communicate.

Imagine being able to MATERIALIZE IMAGINATION. We've long had that ability through art, speech, mass media, and whatnot. But now it can be automated.

At which point the human condition fundamentally changes on a level we've never seen before.

On top of that, giving machines the ability to imagine is all but the final step before we reach artificial general intelligence. It's no coincidence that it's a language model that sparked discussion of whether AI has finally become sentient. It was always going to be imaginative AI— of text, video, audio, images, and more— that gave us the first machines of science fiction.

And then there's the implications for alternative realities. We're not going to spend years handcrafting every little pixel for the coming Metaverse. We're going to get AI to create it all for us. This is another profoundly transformative technology, the first gasps of transhumanism where humans begin losing themselves into high technology. It'll drive people to merge with technology to cut out the middle man.
I'm still not sure. I'll grant you that we need AI that can imagine (to a degree) in order to have machines/robots that can respond to our requests in the real world. But there's a limit to how smart/creative/sentient (and fyi, we're misusing that word; what most people call sentient is really sapient - our pet hamsters are probably sentient, but only humans are sapient [officially, but scientists are pondering if chimps and dolphins also qualify]) needs to be in order to serve us. Frankly I don't want a computer that can think and act on its own without being issued an order. I'm interested in AI purely as a tool to serve me, not to serve as my replacement. But that still somewhat seems separate from media creation. More like, they're linked in the same way video and videogames are - you need video in order to have videogames. You need an AI that can understand, and thus create, representations of trees and cars and buildings aka make video of these things, in order to have an AGI that can intelligently interact with them. But at the end of the day, while an AI capable of doing these things is needed for AGI, media creation as we use it is still just a time and boredom killer. Even your alternative realities, the whole metaverse, is mostly just going to be used as videogames, albeit with a communications aspect. I'm more interested in how this will be applied to AR rather than VR - the ways it will help us navigate and understand the world we have, rather than creating worlds to escape too.

Don't get me wrong, I understand the amazing implications of what all (well, some - it's the singularity for a reason, we can't predict what all will come) this could mean down the line, how the tech to make Batman vs T-Rex: Jurassic Gotham could also be used to have an AI that thinks up more innovative building designs to solve various urban problems like housing, cooling with less AC, reducing waste, improving both mental and physical well being, etc - but then call out those uses. Focusing on how this will give us our own personalized and tailored Netflix/Spotify/library kind of undersells the future by focusing on the, well, meaningless entertainment aspect.

I don't know, maybe I'm just overthinking it. It's just the ability to create my own movies/songs/books/VR worlds doesn't spark my imagination much. I'm not interested in technology for further escapism, but more for further exploration, enhancement, and quality of life improvements. I fear a future where mankind, or post-mankind, shrinks into our own fantasy worlds and never push further to find what else lies out beyond there. I would like to spread out tentacles or roots of my consciousness along with others to the furthest corners of our universe, or however much of it we can reach within the laws of physics. I would like to linger on earth for the next several billion years, and then drift back to watch our sun expand and take our little blue marble. I want copies of me spread out across galaxies to settle new worlds and then watch as the expansion of space pulls the stars so far away they blink out of the night sky, leaving behind only darkness. I want to witness the last moments of our reality, trillions of years from now, when entrophy finally wins out and there's no energy left in all reachable space for me to continue on any further. And with any luck, I want some versions of me to escape into newer, younger universes to repeat the process all over again, time and time again, if such is even possible in those distant eons beyond. I think sometimes we sell short the visions of the future, hyping up how cool it will be to live in VR worlds as gods creating our own worlds to suit our own tastes. Virtual Godhood seems so small in the scope of eternity, escapism just misses out on the infinite possibilities of dealing with the physical limitations of the endless complexities of realities that exist now and in the future that we had no say in creating, in the puzzles they pose for us to solve created by such cosmic forces that no "difficulty setting" in our self-created worlds could ever equate the satisfaction of experiencing. I don't know, I see the next century and the Singularity not as this great big wondrous transhuman ending, but as a stepping stone to watching the stars burn out and surfing the last light waves into the dark Chaos of nothing, and perhaps to even greater realities beyond our small bubble of spacetime to Membranes of alien spacetime...

I suppose it's just a matter of focus. AI learning to create media and imagine is great, but we need to emphasis more what this means beyond just short term thrills I think. Or maybe, just maybe, I'm overthinking this whole thing.
User avatar
Yuli Ban
Posts: 4631
Joined: Sun May 16, 2021 4:44 pm

Re: Yuli's Treasure Chest

Post by Yuli Ban »

"I told you loads of times!"


How the Dead Internet Theory is fast becoming reality thanks to zero, marginal-cost content generated at infinite scale
Dead Internet Theory
A theory — The Dead Internet Theory — that went viral some years back, posited that the internet is a no man’s land of bots and fakery. It was of course, a ridiculous overstatement. But then again, maybe not that ridiculous? Maybe it was actually correct, just too early on the scene. 3
If you search the phrase i hate texting on Twitter and scroll down, you will start to notice a pattern. An account with the handle @pixyIuvr and a glowing heart as a profile picture tweets, “i hate texting i just want to hold ur hand,” receiving 16,000 likes. An account with the handle @f41rygf and a pink orb as a profile picture tweets, “i hate texting just come live with me,” receiving nearly 33,000 likes. An account with the handle @itspureluv and a pink orb as a profile picture tweets, “i hate texting i just wanna kiss u,” receiving more than 48,000 likes.
Here are some case studies (treat the numbers with healthy skepticism, these figures are impossible to determine reliably): The thing is, once an account is setup, it’s trivial to computationally post new content and interact with platform participants. And that can artificially distort the discussion.
Remember Cambridge Analytica? The affair proved that it’s possible to manipulate people on a mass scale by the content they consume (in this specific case, as political advertisements). 
But that content was all created by humans. And humans are slow to create.

And this will only accelerate in the coming years until we reach a bizarro Metal Gear Solid 2-esque information singularity where bubbles exist for every individual, and all society fractures as any and all possible reality becomes easily fabricated.

Want to look up news websites that treat the existence of the paranormal as normal and commonplace? You bet those will be everywhere. Want to interact with whole forums that are populated entirely by bots (and you'll never know)? Just wait a couple years.
And remember my friend, future events such as these will affect you in the future
User avatar
Lorem Ipsum
Posts: 117
Joined: Tue May 24, 2022 4:51 pm

Re: Yuli's Treasure Chest

Post by Lorem Ipsum »

A loss even greater than Alexandria (or others) probably.
AI or 100% Human.
Post Reply