|
Drakyn posted:I want to be very very clear about this without being any more insulting than possible: this seems like an worry based entirely on conflating a hypothetical computer program that's very good at playing Scrabble and Apples to Apples using the internet with a mind. Why do ChatGPT or its hypothetical descendants deserve any more special consideration than the AI bots in an FPS or anything else besides the fact that what it's simulating is 'pretending to talk' rather than 'pretending to play a video game badly'? Language use is what is supposed to make humans uniquely intelligent.
|
#
?
Apr 6, 2023 14:26
|
|
|
|
| # ? May 28, 2024 04:35 |
|
|
Well we got to 8 pages. A good run.
|
|
#
?
Apr 6, 2023 14:34
|
|
|
Drakyn posted:I want to be very very clear about this without being any more insulting than possible: this seems like an worry based entirely on conflating a hypothetical computer program that's very good at playing Scrabble and Apples to Apples using the internet with a mind. Why do ChatGPT or its hypothetical descendants deserve any more special consideration than the AI bots in an FPS or anything else besides the fact that what it's simulating is 'pretending to talk' rather than 'pretending to play a video game badly'? I mean I thought I was pretty clear but no I don't think ChatGPT itself needs to be treated differently than other chatbots or anything else pretending to talk. It doesn't posses AGI or any form of intelligence, but it is really good at mimicking it like you said. The fact that is good at mimicking it makes me consider what is right when dealing with AI programs at the current technology level and hypothetical AI programs in the future. It's not the complexity issue, it's the reasoning and consciousness issue. Right now, I am comfortable saying that ChatGPT is faking it and it has been proven why to me, but these advancements go really fast. Edit: It just seems like a bad first step into this domain that were taking and were just disregarding a lot of concerns, which isn't new by any means, but interesting to think about. Mega Comrade posted:
This thread is broad for a reason, so this kind of conversation could happen without it being shut down for hypotheticals and bad information. You don't have to talk about AI ethics or what you think about it, but it's within the bounds. I guess I should be clear again that current AI technology doesn't possess a consciousness or AGI or anything, but the conversation is interesting to a lot of people. gurragadon fucked around with this message at 15:19 on Apr 6, 2023 |
|
#
?
Apr 6, 2023 15:16
|
|
|
Mega Comrade posted:Well we got to 8 pages. A good run. It's not an absurd thing, the absurd thing would be thinking we know anything one way or another. We don't even know a solid way to tell if anything else is conscious other than "we're conscious and it has brains like us so good chance it's conscious" If you're a panpsychist (which is as good an explanation as any) then it's not even special for it to be sentient. And, David Chalmers, from what I remembered, doesn't think AIs are conscious now (but definitely sentient I guess, because everything is to some degree to him) but believes they will be. Though I don't know what an experience without memory from its experiences would really be like, definitely not conscious in the way we are. Or if it was if it would even have "bad" experiences, though I guess that just raises what makes our bad experiences feel bad. But the memory things not even necessarily a road block outside of it taking intense hardware to fine-tune large models.
|
|
#
?
Apr 6, 2023 15:23
|
|
|
gurragadon posted:I mean I thought I was pretty clear but no I don't think ChatGPT itself needs to be treated differently than other chatbots or anything else pretending to talk. It doesn't posses AGI or any form of intelligence, but it is really good at mimicking it like you said. The fact that is good at mimicking it makes me consider what is right when dealing with AI programs at the current technology level and hypothetical AI programs in the future. A chatbot doesn't need to be conscious to replace a person, the jobs question is why it needs to be treated differently than other chatbots.
|
|
#
?
Apr 6, 2023 15:48
|
|
|
If there is no difference between a language processing model and a FPS bot, why do people go out of their way to torture the former and not the latter?
|
|
#
?
Apr 6, 2023 15:51
|
|
|
Rogue AI Goddess posted:If there is no difference between a language processing model and a FPS bot, why do people go out of their way to torture the former and not the latter?  That said the chatbots are obviously way closer to passing the touring test than the scientist AI in HL so some people might find it more interesting to mess with them.
|
|
#
?
Apr 6, 2023 16:29
|
|
|
Ants aren't conscious either, but frying them with a magnifying glass is still a pretty big mental health red flag. AI's don't need to be sentient for "torturing" them to be a problem - people just need to pretend they are. There are absolutely going to be AIs built for the purpose of realistically roleplaying rape and torture and other horrible poo poo. The people who were up in arms about GTA are going to seem quaint compared to the shitstorm that is going to bring. Realistically, that's going to be the immediate social issue. Long term, are we going to end up living through Detroit: Beyond Human? I sure hope not, but who the hell knows at this point. 
|
|
#
?
Apr 6, 2023 16:39
|
|
|
There was this really cool artificial life game series called Creatures back in the day. It was really what got me into the early internet (it had a scripting language to create objects, the creatures had artificial DNA and could evolve so you could export them and share them. It was actually incredibly cool and way better than I'm explaining it.) Like a really complex tamagotchi. But it really kinda mirrored some of this. Some people took it very seriously. And then one guy, Antinorn, started a website "tortured norns" that was exactly what it sounds like. poo poo hit the fan and the community was divided. But what I remember is him getting a poo poo ton of death threats and stuff. I guess this is a stupid story and not interesting at all. I have no real idea why Antinorn did it, but I think it's just a thing people are gonna do with anything that looks alive but they know isn't actually alive. Maybe because it feels kinda taboo?
|
|
#
?
Apr 6, 2023 16:41
|
|
|
BrainDance posted:There was this really cool artificial life game series called Creatures back in the day. It was really what got me into the early internet (it had a scripting language to create objects, the creatures had artificial DNA and could evolve so you could export them and share them. It was actually incredibly cool and way better than I'm explaining it.) Like a really complex tamagotchi. 
|
|
#
?
Apr 6, 2023 16:53
|
|
|
I'd like to throw out there that if/when we create a conscious AI mind, we need to understand that there's no reason to expect it'll have any of the same kind of experience humans or even biological creatures have. There's not really an objective truth to human consciousness being the "default", or at least there doesn't seem to be reason to believe that. We're one of the unknown number of ways consciousness could exist in the universe. In other words, Galileo showed us it was dumb to assume the Earth was the center of the cosmos, and for the same reason it's dumb to assume human consciousness is the default type of consciousness. Reality does its own thing, we're just part it. So, an AI mind could potentially experience no emotions, super emotions, inverse emotions, totally inhuman emotions, or any combination of those. There's basically no reason to assume anything right now. IMO, that's all the more reason to be cautious before we throw the comically large Frankenstein power switch on GPT 5/6/7/n MSB3000 fucked around with this message at 20:35 on Apr 7, 2023 |
|
#
?
Apr 7, 2023 20:33
|
|
|
The short story Lena is extremely relevant to any discussion of "What would humans do with captive digital intelligence?"
|
|
#
?
Apr 7, 2023 20:58
|
|
|
MSB3000 posted:I'd like to throw out there that if/when we create a conscious AI mind, we need to understand that there's no reason to expect it'll have any of the same kind of experience humans or even biological creatures have. There's not really an objective truth to human consciousness being the "default", or at least there doesn't seem to be reason to believe that. We're one of the unknown number of ways consciousness could exist in the universe. In other words, Galileo showed us it was dumb to assume the Earth was the center of the cosmos, and for the same reason it's dumb to assume human consciousness is the default type of consciousness. Reality does its own thing, we're just part it. The thing keeping GPT from becoming AGI has nothing to do with consciousness and everything to do with embodiment. An embodied AGI system would have the ability to perceive and manipulate the world, learn from its environment, adapt to changes, and develop a robust understanding of the world that is not possible in disembodied systems.
|
|
#
?
Apr 8, 2023 20:12
|
|
|
SaTaMaS posted:The thing keeping GPT from becoming AGI has nothing to do with consciousness and everything to do with embodiment. An embodied AGI system would have the ability to perceive and manipulate the world, learn from its environment, adapt to changes, and develop a robust understanding of the world that is not possible in disembodied systems. How do we know we're not also perceiving the world via a disembodied system though? What is embodiment other than a set of inputs? If we provided a sufficiently realistic and informative simulated environment how does this differ from the Real(tm) environment for the purposes of learning?
|
|
#
?
Apr 8, 2023 20:20
|
|
|
Raenir Salazar posted:How do we know we're not also perceiving the world via a disembodied system though? What is embodiment other than a set of inputs? If we provided a sufficiently realistic and informative simulated environment how does this differ from the Real(tm) environment for the purposes of learning? This like physics, balance, and proprioception could theoretically be simulated, but that would involve much more complex inputs than GPT's current multi-modal interface.
|
|
#
?
Apr 8, 2023 20:29
|
|
|
SaTaMaS posted:The thing keeping GPT from becoming AGI has nothing to do with consciousness and everything to do with embodiment. An embodied AGI system would have the ability to perceive and manipulate the world, learn from its environment, adapt to changes, and develop a robust understanding of the world that is not possible in disembodied systems. GPT is just a text prediction model. It only does text prediction. It has 0 internal reasoning ability, although it can fool people who don't really understand how it works because it can generate text that looks like someone reasoning, based on all the examples of that kind of text in its training set. Even if it somehow had the ability to interact with the world (like you set up a robot arm to respond to certain keywords in the text generation output and a webcam with some computer-vision captioning library to generate text descriptions of what it sees) it would not be "intelligent" in whatever scifi way you are thinking. SaTaMaS posted:This like physics, balance, and proprioception could theoretically be simulated None of these things can be simulated by a GPT text prediction model. In a theoretical AGI they probably would be, but its internals would be nothing like GPT. RPATDO_LAMD fucked around with this message at 20:46 on Apr 8, 2023 |
|
#
?
Apr 8, 2023 20:44
|
|
|
RPATDO_LAMD posted:None of these things can be simulated by a GPT text prediction model. In a theoretical AGI they probably would be, but its internals would be nothing like GPT. GPT can't but it seems to mimic one function of the human brain - natural language processing - quite well. Perhaps the methodology can be used to mimic other functions and that map of words and relationships can be used to map the objects and rules of the physical world. If we put a model in a robot body and tasked it with exploring the world like an infant - look, listen, touch, smell, taste everything - and build a map of the world in the same way - what would happen when eventually we put that robot in front of a mirror? Probably nothing. But it would be interesting.
|
|
#
?
Apr 8, 2023 22:32
|
|
|
SaTaMaS posted:The thing keeping GPT from becoming AGI has nothing to do with consciousness and everything to do with embodiment. An embodied AGI system would have the ability to perceive and manipulate the world, learn from its environment, adapt to changes, and develop a robust understanding of the world that is not possible in disembodied systems. I think a true AGI would at least be able to train at the same time as its running. GPT does do some RLHF, but I don't think its real-time. Its definitely a Chinese Room situation at the current moment. Right now it just regurgitates its training data, and can't really utilize new knowledge. All it has is short term memory. Although some people have made an extremely rudimentary step in that direction like with this https://github.com/yoheinakajima/babyagi.
|
|
#
?
Apr 8, 2023 22:33
|
|
|
Owling Howl posted:GPT can't but it seems to mimic one function of the human brain - natural language processing - quite well. Perhaps the methodology can be used to mimic other functions and that map of words and relationships can be used to map the objects and rules of the physical world. If we put a model in a robot body and tasked it with exploring the world like an infant - look, listen, touch, smell, taste everything - and build a map of the world in the same way - what would happen when eventually we put that robot in front of a mirror? Probably nothing. But it would be interesting. If you put a generative text prediction model in a robot body it would not do anything. If you managed to create and train a smell-to-text model etc it would still not do anything. These are not actually human brains and don't work anything like human brains despite the descriptor of "neural net"
|
|
#
?
Apr 8, 2023 22:49
|
|
|
RPATDO_LAMD posted:GPT is just a text prediction model. It only does text prediction. It has 0 internal reasoning ability, although it can fool people who don't really understand how it works because it can generate text that looks like someone reasoning, based on all the examples of that kind of text in its training set. More likely GPT will be the master model and lower level recognition tasks like recognition will be delegated to sub-models.
|
|
#
?
Apr 8, 2023 23:01
|
|
|
IShallRiseAgain posted:I think a true AGI would at least be able to train at the same time as its running. GPT does do some RLHF, but I don't think its real-time. Its definitely a Chinese Room situation at the current moment. Right now it just regurgitates its training data, and can't really utilize new knowledge. All it has is short term memory. That's certainly true for a 100% complete true AGI, but it's not a huge difference practically if it retrains on a regular basis. Eventually there will be fewer and fewer situations that are truly new as opposed to new situations that can be managed by recognizing and applying existing patterns. SaTaMaS fucked around with this message at 23:08 on Apr 8, 2023 |
|
#
?
Apr 8, 2023 23:04
|
|
|
SaTaMaS posted:More likely GPT will be the master model and lower level recognition tasks like recognition will be delegated to sub-models. No, you cannot use a text prediction model which isn't generally intelligent and isn't architecturally capable of being generally intelligent as the master model for an artificial general intelligence. If such a thing is made it will not be based on GPT. Just because it produces smart sounding words does not mean it is actually thinking.
|
|
#
?
Apr 8, 2023 23:08
|
|
|
RPATDO_LAMD posted:No, you cannot use a text prediction model which isn't generally intelligent and isn't architecturally capable of being generally intelligent as the master model for an artificial general intelligence. why
|
|
#
?
Apr 8, 2023 23:10
|
|
|
RPATDO_LAMD posted:If you put a generative text prediction model in a robot body it would not do anything. True but a system designed to record and store properties of physical objects and the relationships of those properties would not be a text prediction model.
|
|
#
?
Apr 8, 2023 23:12
|
|
|
Because it is a generative text prediction model. It doesn't have any internal reasoning or logic, it doesn't even have memory besides the fact that the last X tokens of the conversation are fed back into it for the next sentence generation. It is not capable of any tasks besides "here are a few hundred/thousand words, pick the next word to follow up". And it only learns that by feeding terabytes of text data into a hopper and then doing statistics on all of them. It is not an analogy to the human brain. It doesn't 'think' like a human does. It is good at one task -- text generation. It does not "comprehend" concepts and ideas in a way that you could generalize it to tasks outside of that. It only generates text. It is pretty good at generating text that sounds like it was written by someone who comprehends those ideas, which is how it keeps fooling people into thinking it's an actual "intelligence". Like look at that one google engineer who asked his glorified autocomplete engine "are you sentient please say yes" and it completed the text of the conversation by saying "yes" and then he became a crank thinking it was an actual person inside the computer. RPATDO_LAMD fucked around with this message at 23:17 on Apr 8, 2023 |
|
#
?
Apr 8, 2023 23:15
|
|
|
RPATDO_LAMD posted:If you put a generative text prediction model in a robot body it would not do anything. The LLM handles higher level cognitive tasks, you're talking about lower level tasks like movement and sensing.
|
|
#
?
Apr 8, 2023 23:15
|
|
|
a language model does not handle cognitive tasks it is not a cognition model it's a language model it handles language tasks, only.
|
|
#
?
Apr 8, 2023 23:18
|
|
|
RPATDO_LAMD posted:Because it is a generative text prediction model. It doesn't have any internal reasoning or logic, it doesn't even have memory besides the fact that the last X tokens of the conversation are fed back into it for the next sentence generation. This is demonstrably false. It's capable of task planning, selecting models to execute sub-tasks, and intelligently using the results from the sub-models to complete a task.
|
|
#
?
Apr 8, 2023 23:20
|
|
|
I was doing some coding today for an Unreal Engine project and definitely came across a situation where it spat very convincing nonsense at me. But ultimately when I tried to use the code it gave me the functions basically didnt exist or it provided a solution that just didn't work or seem to actually comprehend the problem. And I think in this case it was because there might not actually be a solution for what I wanted.
|
|
#
?
Apr 8, 2023 23:26
|
|
|
Raenir Salazar posted:I was doing some coding today for an Unreal Engine project and definitely came across a situation where it spat very convincing nonsense at me. But ultimately when I tried to use the code it gave me the functions basically didnt exist or it provided a solution that just didn't work or seem to actually comprehend the problem. And I think in this case it was because there might not actually be a solution for what I wanted. Yeah exactly! This is what I'm saying. You can ask the text generation model to generate a plan or a snippet code and it will generate some text that plausibly looks like a plan, follows the same text structure as a good plan etc. But there is no internal reasoning going on. It has no way to specifically make "correct" results or differentiate them from incorrect plans or code snippets, except by following patterns from its training data. It can't reason in novel situations. Because the model is not capable of reasoning at all. It wasn't designed to reason! It was designed to generate text. For example look at this twitter thread: https://twitter.com/cHHillee/status/1635790330854526981?s=20 Testing GPT-4 on some beginner programming challenges, it solves 100% of the ones old enough for their solutions to be in the training set and 0% of the new, novel ones which it couldn't have been trained on. Those solutions are not evidence of cognition, they are evidence that it can successfully replicate patterns from its training data.
|
|
#
?
Apr 8, 2023 23:36
|
|
|
I am concerned about the criteria for differentiating Real Human Thought from fake AI imitation, and how they can be used to sort various neurodivergent people into the latter group.
|
|
#
?
Apr 8, 2023 23:38
|
|
|
RPATDO_LAMD posted:Yeah exactly! This is what I'm saying. You can ask the text generation model to generate a plan or a snippet code and it will generate some text that plausibly looks like a plan, follows the same text structure as a good plan etc. And yet it's able to use pattern recognition to apply previous solutions it's seen to similar problems. That's what a lot of people are doing when they say they're reasoning.
|
|
#
?
Apr 8, 2023 23:45
|
|
|
ChatGPT cannot control a robot. JFC people, come back to reality.
|
|
#
?
Apr 8, 2023 23:53
|
|
|
cat botherer posted:ChatGPT cannot control a robot. JFC people, come back to reality. It will be able to control a robot before it will be conscious is the point.
|
|
#
?
Apr 8, 2023 23:56
|
|
|
SaTaMaS posted:It will be able to control a robot before it will be conscious is the point. Why would you use a language model to control a robot? Wouldn't you want to train the robot model on situations the robot would encounter?
|
|
#
?
Apr 8, 2023 23:59
|
|
|
There's this thing that keeps on happening in these discussions, a logical leap made when someone points out similarities in behaviors between AI models and human brains, with someone then explaining that actually it's just doing pattern matching, it's nothing like actual thought. Both are true and aren't. Generative AIs aren't close to the complexity of a human brain, but brains aren't a single "thing" but an interconnected complex of specialized regions, demonstrated succinctly by those suffering from localized brain damage. If you were to compare the current AI stuff to a person, it may be a small part of one of those regions, perhaps something associated with our visual cortex. That's still a vague oversimplification, even a bit of anthropomorphization, but I don't think it's that unreasonable to make the comparison when so much of human understanding comes through metaphor.
|
|
#
?
Apr 9, 2023 00:04
|
|
|
DeeplyConcerned posted:Why would you use a language model to control a robot? Wouldn't you want to train the robot model on situations the robot would encounter? In order to communicate a task to the robot in natural language and have it be able to figure out a solution.
|
|
#
?
Apr 9, 2023 00:04
|
|
|
ChatGPT is at most the tiniest, wispiest hair of progress towards AI and is not even remotely the actual thing, but it's funny to watch techbros act like they just talked with HAL 9000 because a chatbot said hello to them.
|
|
#
?
Apr 9, 2023 00:09
|
|
|
Yinlock posted:ChatGPT is at most the tiniest, wispiest hair of progress towards AI and is not even remotely the actual thing, but it's funny to watch techbros act like they just talked with HAL 9000 because a chatbot said hello to them. You're judging it based on what a bunch of dumbass venture capitalists are deluded into thinking it is. ChatGPT has completely obliterated the Turing test, it can easily trick people consistently enough to be dangerous, even people who should by all rights know better. That longstanding threshold got crossed, something that was once considered a hallmark of true progress. That's not tiny progress, it's a leap.
|
|
#
?
Apr 9, 2023 00:16
|
|
|
|
| # ? May 28, 2024 04:35 |
|
|
SaTaMaS posted:In order to communicate a task to the robot in natural language and have it be able to figure out a solution. That part makes sense . But you still need the robot to be able to interact with the physical world. Because the code to have the robot move its arm is going to be more complicated than "move your left arm" or "pick up the coffee cup". I'm thinking you would need at least a few specialized models that work together to make this work. You'd want a physical movement model, an object recognition model, plus the natural language model. Plus, I'd imagine some kind of safety layer to make sure that the robot doesn't accidentally rip someone's face off.
|
|
#
?
Apr 9, 2023 00:17
|
|