Huge correction:In the original thought experiment,you don't have a book of Chinese to English Dictionary or something like that,instead you have a book listing appropriate responses to specific questions which just means you are following an algorithm(a set of instructions),but don't really understand Chinese. In the version this guy told,you could learn the language bit by bit and also understand what the questions and answers are beacuse you have a book that translates Chinese to English(assuming you know English beacuse how are you gonna do anything here if you don't know English),which directly contradicts his statement that you don't actually understand the questions or answers. Looks like you didn't understand the thought experiment.
That’s a good point, but aren’t some AI techniques “better” than that now? With deep learning the analogy would be “remembering in which context the characters are used” which is really what learning a language is, in a way
@@kimberlycarrigan8824 same thing in context Edit: people seemingly can’t read the words “in context” Of course it’s quite different, but the relation to the analogy stays the same
This is literally my spanish class. I don't understand the question, but i know that i have to respond with a certain sequence and combination of letters.
Yeah, I think the argument in this experiment is a bit flawed EDIT: Multiple people have explained to me that the experiment is not flawed, and the guy who explained it explained it poorly. Thank you all for informing me
@@giannismentz3570 if you are constantly translating the same words to English and then translating English back to Chinese, then yes eventually you'll remember "oh yeah, this is the 10th time Ive translated this word to this character" and you'll learn *something*. So yes, his explanation of the thought experiment is wrong, you aren't given a dictionary, you are given a long instruction manual as to what you are supposed to respond with when given a prompt in Chinese. its connect the dots, you don't actually understand what's being said
@@idioticlight no you wouldn't. Like I said, you learn nothing unknown from repeated exposure. It's still gonna be unknown. You need some kind of feedback.
Think about it, if u can't see outside of the room... how do you know you're still in reality and didn't noclip into an identical room but inside the backrooms?
Two things: In the original thought experiment you did not have a ChineseEnglish dictionary. You only had a set of instructions to follow which told you what responses to give for a set of specific questions. This is equivalent to decoding a set of symbols by returning an equivalent answer from a table. This means that you are merely following a pre-programmed low level algorithm which they argued was not a sign of consciousness since you did not have to process and understand the meaning of the questions and then subsequently synthesize your own answer according to your understanding. This also means that the Chinese English bit is completely irrelevant as it is also merely decoding symbols without the need for advanced sophisticated algorithms. In this case, whatever was in the box: - did not understand Chinese - did not understand the questions or answers In your video: You had exclusively a Chinese English dictionary which means that if you were not intelligent/conscious, you could ONLY run that simple translation decoding algorithm to convert one language symbol to an equivalent other. But since you were able to answer the questions, that means that at some point AFTER the translation you would have had to process and understand the question, subsequently synthesize your own answer to that question before re-translating that back to Chinese, which means you must have used an advanced processing algorithm that denoted intelligence/consciousness in the process. This directly contradicts both your statement on “whatever inside the box not understanding the questions” AND also “not understanding Chinese”.
Your second point is exactly what I was thinking as soon as the video ended, also appreciate your first point because that's the key info the video missed out to prove it's conclusion. This is the problem with speaking about stuff you are not an expert on, you take real data and real results but can't connect them accurately so a small change in the input data will make your whole conclusion wrong and you won't even know why, neither the people listening. An expert in computer science would easily spot something off listening to this even if that expert has never heard of this exact experiment.
@@jeremiahcruickshank8491even if it only followed instructions with a basic AI model it would still learn something from it. Even the dumbest of animals will sometimes learn that when an event occours it will lead to another one, like when you ring a bell everytime you gonna feed your dog. It will understand that when you ring the bell food is coming on it's way but that doesn't means it has human level inteligence or that it has an inteligence just like Advanced AI but that it understands that the world around us can change following the simple pattern we always use
@@jeremiahcruickshank8491 yeah and also coming to the conclusion that ai can never be sentient is just a mindless conclusion. You don’t need to be a computer science major to understand that if a brain can become sentient, a sufficient ai can do the same
You can understand the concept, but you don't actually learn it and understand it. You understand what you're supposed to do which is to understand Chinese by converting Chinese to English using a dictionary. You understand the task, doesn't mean you are able to understand Chinese.
Also, I feel like people who are skeptical about AI will never be sentient beings are just people who think our consciousness is very very special, it blows my mind hows the model of the behavior of the neural network is very similar to ours brain, the only difference is we have more neurons compare to the current model of artificial neural network. If you see the video where the AI tries to render dogs at the top of a video, it will not just show a picture of a dog at the top of the video, instead, the AI hallucinates dogs in the video and the result is the same Schizophrenic people experience! Scientists believe that AI is dreaming. It means they are probably slightly conscious but just at a dreaming level.
As someone that's trapped in a room and forced to translate Chinese questions my whole life, I can confirm this is what happens when you lose your streak on Duolingo
Ok, but the thing about the person in the room is, that at some point if the experiment goes on for long enough, he'll actually be able to understand chinese and write back without having to translate it
Yeah, in this versiom which he explained incorrectly. In the actual versipn (and in the images, if you pause), ypu actually only have a list of instructions, such as: if 漢, then reply with 字. giving you no way to understand meaning or context
“in this room you are given a box of chinese characters and a book on how to translate them” me: psshhhh that’s easy i dont need the book “now in this scenario you don’t speak chinese” me: mbru
@@silversun1736 the point of this argument is that the computer (the person in the room) doesn't understand the language. It has to translate what we (the person on the outside) say. Then, re translate it's response into our language. Stick figures would still follow this argument.
It's like a calculator. A calculator might be able to solve complex math equations but it doesn't understand the numbers its using. A human mathematician on the other hand very well understands the numbers.
I mean, we do the same thing. We hear vibrations in the air, take that sound, translate it into thoughts, then think, then translate those thoughts into vibrations.
This guy described it wrong, in the original you use computer software to figure it out, basically he came up with this idea after a computer sort of learned Chinese, you could put in a sentence and it would come up with an answer, but he asked if the computer really knew Chinese or if it was just following instructions
I think the original idea is that you input the chinese and get given a response to write out. This guy says its translating the chinese which is different, because you'd need knowledge about chinese to translate it, and can think up your response to convert. The original point is you srent thinking, youre just doing what you're told, like youre on autopilot and dont actually understand it. Like AI
You ever take a SHIT so large where you just got to get a new toilet afterwards? God damn what am I done?People don't realize that I am RIPPING ASS like no other god. Damn I am blowing this toilet out. This is what happens when you put too much butter on your popcorn. God damn. I am going to rip ass for 5 more minutes.i took a MASSIVE SHIT in 2020. it was a chicken sandwich and boy it was spicy, pretty good too...but i RIPPED ASS HARD AFTERWARDS. wish you would of told me back then to prepare for that, and to get two, 1 to eat for dinner and 1 to eat while blowing out the toilet 15 min later. 🥪GOD DAMN💨LORD😩HAVE💩MERCY
*An important distinction:* in the room you aren’t actually translating in the OG thought experiment, you just match the symbols you don’t understand to a response in symbols you don’t understand and give it to them
That’s not quite what’s happening- it wasn’t explained very clearly but the actual idea is that you don’t have a translation, you just have a book that shows you which symbol to put in response to the symbol you get. You don’t know what the symbol you get means, or what you send back means. Just how to reply to what
You say the person in the room doesn't understand the question or the answer, but from what I understand from the scenario, the person would understand both the question and the answer. They can't understand the question initially because they don't speak Chinese, but after using their translation book they can understand the question. Furthermore, they can now come up with a truly unique and original response and just translate into Chinese to send out. So from this I think the analogy is flawed. If I'm wrong, please comment. I think the overall message was correct since for an AI it literally has a "rule book" which tells it which output to send based on a given input.
Yep, I made the same comment. Either his articulation of the thought experiment is completely off, or it's stupid. Not just flawed, but stupid, for the reasons you stated. There's no intelligence or understanding in the translation forth and back, but who exactly is generating the answer if not an intelligent being with genuine understanding?
Apparently the video author messed up and couldn't explain it properly. In the experiment you get a "translation book" with lots of Chinese questions and answers. You get a question in Chinese, you find it in the book and write back the answer. The person outside the booth would get a sensible answer that sounds properly Chinese and is written in proper Chinese, they would reasonably assume you're Chinese or understand Chinese, but all you do is find the question in a book and write the answer assigned to it. So, you don't understand Chinese, it just looks like you do. Now, my question would be - who wrote the book? Is it a person that understands Chinese? Alright, well, then the person outside the booth isn't even talking to you, really, they're talking with the book author and they think that author clearly knows Chinese. Now, consider that a narrow AI like GPT-whichever-number studies Chinese on a large dataset. Now, you sit in the booth and instead of a book you get a scanner and a display. You scan the question you get and you get an answer displayed, generated by GPT. You write that answer, still in Chinese, on a piece of paper and give it back to the questioner. If the questioner still thinks this is a sensible answer and is convinced that you understand Chinese... Clearly, you don't, but does the GPT? This is basically turning into a Turing test now.
The problem with that is the human brain could also be represented by wires and switches, meaning that we are also ran by a set of instructions, yet we are sentient.
Another problem: to translate Chinese and read instructions, you need your own language. It is implied that the translator has their own language. It is therefore implied that the translator is not limited by their intellectual ability or awareness, but the compatibility of their language with the Chinese person. Switch the places of the Chinese person and the translator in a reverse of the situation. They are not that different, they just don't understand each other
He just narrates already well-known ideas. It's too smart for youtube shorts newbies and not smart enough for people who already know about these. That being said he is growing so I wouldn't say underrated.
@@terryriley6410 You got a point. but nobody will go search for blogs related to ai and read them for hours to find something interesting that you can understand as well. shorts makes it understand way easier and less of a chore. that being said, 1 minute is really not enough for learning a theory. so for deeper understanding reading is necessary.
Exactly @@heroispro8918 This man is the lumped summary of in depth research that is not only accurately informative but also well comprehensive by those who don’t have the will to look it up themselves
@@heroispro8918 Lot of people learn about AIs. If learning non-superficially about something feels like a chore for someone, they aren't interested in the subject. Shorts are a great way to fool yourself that you learn about something or that you even care to. It's usually a massive waste of time. These shorts surf the trends of AI without offering anything new and even what they offer are superficial interpretations or explanations of deeper concepts. It's cool but it's not "underrated" cool imo.
@@cullenhitchanson7753 I'm pretty sure this man doesn't do in-depth research. His work is the summary of summaries. If you want to see good summaries of what is happening in the world of AI I suggest the youtube channel "Two minutes papers".
I mean, the argument could easily be made that that is all WE do- follow a set of instructions- biological imperatives deep in our DNA where our personalities and preferences are already encoded from birth, and we are just responding to whatever outside stimulus hits us. Most people have very, VERY little understanding of how even the things they depend on every day works. Most humans I know respond all the time to questions and subjects they have no understanding of. 😂 But they would insist they are sentient…
Yes but there’s no logical way to replicate the phenomenon that we call “understanding” to truely understand something so far is a unique property of consciousness
@@saibaMasuku true- which also means there is no way to prove something is NOT sentient- including ourselves. “I think, therefor I am” has not been improved on since Plato. Separated Twin studies and the studies on feral children have proven, rather than refuted, the idea that our personalities and preferences are a product of most nature (DNA, epigenetic) and some operant conditioning (language essentially). If we can’t prove we ARE sentient, we can’t prove any AI that claims it IS, is not. Our brains run on electricity powered by biological mechanisms. No reason a brain can’t be powered by artificial, external mechanisms. In a way we already have technology that does that. ECMO?
@@juliejanesmith57 Well said, I have thought about this a lot too. I'm honestly surprised this thought experiment got as much attention as it did - missing this critical piece within it.
My dad works for a company that develops AI. He has often said “AI has no understanding of reason, and will forever be confidently wrong.” The way AI works is that it takes data from the internet or specific “training” that the model has gone through. It has essentially compiled a bunch of correct responses to a specific thing or statement, and creates its own unique response made up of those responses (which is why ai generated images of people look kinda wonky sometimes). It’s unlikely that AI could become sentient due to this reason, as it doesn’t have any true understanding of reason
As long as AI is being fed a constant stream of live information and data from reality, it will continue to generate and converge on an increasingly accurate view of the world and therefore more accurate and congruent synthesised responses to questions relating that world view.
Think is, you cant just look up the Chinese character for a word because both Mandarin and Cantonese have different grammatical structure than English. The person receiving your answer would know you don’t speak Chinese very quickly. Many languages have such differences in grammatical structure including ASL.
The Chinese and Japanese language share many characters as the Japanese language kind of evolved(??) from Chinese. The Japanese characters that look like these are more complicated, traditional Chinese, this is part of one of the Japanese alphabets. Meanwhile, the characters shown in the video are simplified Chinese, only used by the Chinese. So, the characters shown are indeed Chinese characters, even if they were traditional.
@@yaoyaozhu1978 Some extra info on the bit you seemed unsure about. Japanese and Chinese are not related languages. The japanese just copied the writing system from Chinese, using those characters to stand for their own syllables and words. The full Chinese characters for kanji and simplified forms for hiragana and katakana.
The problem now is that when we create those instructions, we don't know what those instructions are. And there are a lot of instructions generated in machine learning.
My understanding is that as I translated I’d be learning and would be able to just know some of the easy things after a bit and continue to learn. Couldn’t an AI built to grow do the same thing?
This guy messed up an important detail of the thought experiment. In the original thought experiment you don't have a Chinese to English dictionary, you only have a book that tells you the appropriate responses to write back based on the Chinese characters you receive, but you never get to know what is being written or what you are writing back. So the person on the outside would assume you are a Chinese speaker but you are just following an algorithm that tells you what characters to draw based on what characters you are given. At some point you probably would memorize some characters and when to use them, but you wouldn't actually know what they meant.
@@zackalyres7526 well at some point you could end up learning what they meant, though it could be years later. Once you figure out a few, you can start understanding more and more. This is similar to how mini humans learn language, but they also might have a few extra bits of context from objects around them.
"They're just replicating what they think they're supposed to be doing." is often used to describe nuerodivergent people and they're definitely sentient.
The issue with that is, as he translates the characters in the book, he will by memorizing them learn Chinese. And eventually will ditch the book entirely.
I think the thought experiment was explained wrongly. In the version I heard, you're not ever given a translation to anything, you're just told which symbol you receive correlates to which symbol you return. It doesn't even have to be chinese, it would be anything. The point is that you don't have to understand the *meaning* what you're doing to appear intelligent to an external observer.
Living things also follow a set of instructions inside their DNA. It’s just that out instruction are extremely complex. A quote I love is that “the main difference between the real world and the virtual world is the amount of data”.
that's literally what I was thinking and what's true. We humans are not so different from machines, it's just that our brains and biology is so complex that it produces an illusion of actual life and consciousness. Our consciousness is just a complex interplay between the unfathomable number of neurons in our brains which gives the illusion of consciousness but there's no actual consciousness. We are indeed not much different from machines and it's inevitable that machines will eventually get even more complex with a superior illusion of consciousness than ours. Maybe one day we'll get to experience that as well with the help of biotechnology, merging with AI and tech itself.
Our brains already do what the guy in the room does. We just tell ourselves that it's something deeper. If an AI could develop a running protocol that convinces itself that these processes are deep, it would look at our understanding as mechanical and limited.
We look at simple organisms and think they aren't intelligent, they are just a biological program running. But somehow we think we are different because we don't see anything more intelligent than us. But we can't see anything more intelligent than us because we don't have the intelligence for that.
@@anothergenericgamer_all knowledge is ultimately based on that which we cannot prove. This is just another take on what can be simplified as freewill vs determinism. He cannot convince you, nor you him.
I've seen a lot of comments saying you got the original thought experiment wrong, and I totally believe them, because you can just tell the man on the other side of the door "I am locked in this room, open the door, I do not speak Chinese", and you're basically free.
But eventually that person in the room will start learning some letters and slowly they will be able to understand the Chinese and will be able to response without looking at the book
I don't see how it matters at all if you understand Chinese or not. Who cares? Maybe you're too lazy to learn the Chinese while you're in the box and feel content to just rely on the translation book you have. It doesn't matter at all, because the translation book doesn't generate the answers. You are.
@@SimratGill-cp1oe computers don't feel anything, really. AI running as software on computers might, though. Anyway, we're clearly not talking about the same thing. I would argue that it doesn't matter if the person inside can or can't learn to understand the language.
My Chinese correspondent would know that I don't know Chinese because I would take 5 hours to get him an answer and my penmanship of Chinese characters would be unlegible😂
I'm delirious as a mf, but how do you know how awareness works? Which part of the brain is respnsible for consciousness? How do we imagine things without seeing anything? How does the brain turn optical imput into what we percieve as vision?
@@drsamuelhaydensecretgaming6749 there are answers but they are very very complicated. idek myself but I know that we are a combination of millions of tiny cellular "robots" that keep us alive somehow
@@Max-js1mx number was slightly off, it's somewhere around the tens of trillions count. But no matter, because even though we know that brains function using neurons firing in a specific pattern, we still have no idea how all of the things I've mentioned work.
@@drsamuelhaydensecretgaming6749 Your'e right about asking the consciousness question but modern AI can and actually does explain how the brain takes sensory input, Basic computer vision is an example. In saying that, there is a theory that consciousness is caused and arises via nothing more than complexity which does actually make sense as our brains are nothing more than neural networks themselves.
@@brennan352 it doesn't make any sense. if anything, it just makes this even more complicated. complexity cannot cause consciousness to rise because those two things are completely different and doesn't have any relation. An average computer is way more complex than a random bacteria but does that mean that computer is more self aware than that bacteria? We could and has created AIs powered by top supercomputers and they are more complex than most animals we see in our daily life but does that mean they are more self aware than those animals? Modern AI technology explains nothing about how the brain takes sensory input because computers don't 'see' or 'experience' anything. In reality, they are just following a very long and complex domino effect and the trillions of transistors and turning on and off depending on the frequency or features of the input. But when it comes to how we experience these things, we actually does 'experience' those things. We are creating completely new and unrelated images and feelings by interpreting those data we receive.
As someone who is Chinese and migrated to Australia, I can confirm that basically all my classmates suck at Chinese in both writing and speaking and I can not understand them even though they have years of classes and COMPUTERS in which they search up images of each character AND listen to them before trying to speak to me.
So, I've thought about this scenario and here's what I have to say: In this scenario, the person outside is not having a conversation with you, because you don't understand your responses, just how to match the question to it. The book doesn’t understand either, as while it has the algorithm it doesn't know which question it is responding to in the first place. However, the room as a whole, with the book, the person and the space for messages to go in and out CAN consider and respond to questions. That's what the person outside is talking to. By analogy, no specific aspect of the brain can speak English. It's only when enough of the brain collectively is brought together that anything involved can speak.
Exactly, all these people trying to “prove” AI isn’t sentient coming up with stuff like this. Like in what way are we different then? Words go in, the auditory cortex doesn’t understand English so it sends it to the rest of the brain that does the understanding and then a different part of the brain that uses our memories, knowledge of the person we’re talking to and other clues to basically run a huge algorithm to decide what to say.
Yup. This has been said before, but I think the Chinese Room thought experiment depends not only on what computers do being computable, but also on what humans do not being computable. If what humans do is computable, then in theory a sufficiently advanced computer could do what a human can, which, by the original argument's logic, would suggest humans can't really be sentient. Some people try to avoid this objection by claiming that the human mind involves quantum effects. I don't think there's any good evidence for this, but even then, a sufficiently advanced quantum computer could do what a human can, so I don't think it would really change the argument much.
Overlooking the mistake in the experiment, this situation could also say that basically humans aren't sentient either. We follow a complex set of instructions through chemical messaging in our brains. We just *think* we're sentient.
Sentience isn't even something that is measured and the consciousness has been theorised several times to be an illusion made by our brain or a side effect of how cognitively advanced the brain is.
More people need to hear things like this. AI will never be more than a very, VERY convincing imitation. It blows my mind when I see people seemingly gearing up to argue for the "rights" of AI as "people."
the implication is that this applies even to normal chat bots , and works less when tasked to move some language to the other, which requires knowing both incredibly well. especially when its required to shunt word orders from SOV to VSO or SVO
@@rem49 just because someone can say something in English to you and your brain can make sense of it and formulate words in English respond doesn't mean you know English by the same logic
I'd be convinced that an artificial intelligence has sentience if they grow from one line of code to a fully functioning member of society without external help other than stimulus.
Check Out My Latest Video on Simulation Theory! ua-cam.com/video/4OmtAMZdlIk/v-deo.html&ab_channel=ThomasMulligan
ok
I've been stuck in the room for months please let me out
You relay this information incorrectly. But at least you are consistently incorrect.
@@denilla8034are you talking about the video or the though experiment
@@gameslikes0grolls The video.
Huge correction:In the original thought experiment,you don't have a book of Chinese to English Dictionary or something like that,instead you have a book listing appropriate responses to specific questions which just means you are following an algorithm(a set of instructions),but don't really understand Chinese.
In the version this guy told,you could learn the language bit by bit and also understand what the questions and answers are beacuse you have a book that translates Chinese to English(assuming you know English beacuse how are you gonna do anything here if you don't know English),which directly contradicts his statement that you don't actually understand the questions or answers.
Looks like you didn't understand the thought experiment.
Thanks for explaining, I thought what he was saying didn't make much sense
That’s a good point, but aren’t some AI techniques “better” than that now? With deep learning the analogy would be “remembering in which context the characters are used” which is really what learning a language is, in a way
@@meraldlag4336computers don't remember things. they store data.
@@kimberlycarrigan8824 same thing in context
Edit: people seemingly can’t read the words “in context”
Of course it’s quite different, but the relation to the analogy stays the same
This is literally my spanish class. I don't understand the question, but i know that i have to respond with a certain sequence and combination of letters.
Everybody gangsta till I learn Chinese due to repeated exposure
Yeah, I think the argument in this experiment is a bit flawed
EDIT: Multiple people have explained to me that the experiment is not flawed, and the guy who explained it explained it poorly. Thank you all for informing me
@@yoursleepparalysisdemon8752 no it's not flawed. btw you can learn nothing out of "repeated exposure".
@@giannismentz3570 you would indeed learn chinese bit by bit in this scenario.
@@giannismentz3570 if you are constantly translating the same words to English and then translating English back to Chinese, then yes eventually you'll remember "oh yeah, this is the 10th time Ive translated this word to this character" and you'll learn *something*. So yes, his explanation of the thought experiment is wrong, you aren't given a dictionary, you are given a long instruction manual as to what you are supposed to respond with when given a prompt in Chinese. its connect the dots, you don't actually understand what's being said
@@idioticlight no you wouldn't. Like I said, you learn nothing unknown from repeated exposure. It's still gonna be unknown. You need some kind of feedback.
"Im going to place you in a room and lock the door."
*"Don't please."*
Basement treatment
Jail :(
Duolingo laughing in corner
plot twist: it was a rubber room with rats
@@mwzngd1679aw hail nah
"I'm gonna place you in a room and lock the door"
"911 whats your emergency?"
"I am going to place you in a room and then lock the door" i will now have a panic attack
Yes🙂I feel this
Feel it
I would draw on your walls :3
And feel like in the backrooms :D
Think about it, if u can't see outside of the room... how do you know you're still in reality and didn't noclip into an identical room but inside the backrooms?
FR
This dude absolutely slaughtered the description of how this experiment was conducted
That's... a compliment, right? It's actually hard to tell whether this is a compliment or an insult.
@@Thalanox insult... he didn't describe it correctly
@@Thalanox Slaughtered as in butchered. Misunderstood to the point where he directly describes the opposite of the actual experiment.
He is a coomtuber
This... was never an experiment to be conducted. It's a thought experiment r
Two things:
In the original thought experiment you did not have a ChineseEnglish dictionary. You only had a set of instructions to follow which told you what responses to give for a set of specific questions. This is equivalent to decoding a set of symbols by returning an equivalent answer from a table. This means that you are merely following a pre-programmed low level algorithm which they argued was not a sign of consciousness since you did not have to process and understand the meaning of the questions and then subsequently synthesize your own answer according to your understanding. This also means that the Chinese English bit is completely irrelevant as it is also merely decoding symbols without the need for advanced sophisticated algorithms.
In this case, whatever was in the box:
- did not understand Chinese
- did not understand the questions or answers
In your video:
You had exclusively a Chinese English dictionary which means that if you were not intelligent/conscious, you could ONLY run that simple translation decoding algorithm to convert one language symbol to an equivalent other. But since you were able to answer the questions, that means that at some point AFTER the translation you would have had to process and understand the question, subsequently synthesize your own answer to that question before re-translating that back to Chinese, which means you must have used an advanced processing algorithm that denoted intelligence/consciousness in the process. This directly contradicts both your statement on “whatever inside the box not understanding the questions” AND also “not understanding Chinese”.
Your second point is exactly what I was thinking as soon as the video ended, also appreciate your first point because that's the key info the video missed out to prove it's conclusion.
This is the problem with speaking about stuff you are not an expert on, you take real data and real results but can't connect them accurately so a small change in the input data will make your whole conclusion wrong and you won't even know why, neither the people listening.
An expert in computer science would easily spot something off listening to this even if that expert has never heard of this exact experiment.
@@jeremiahcruickshank8491even if it only followed instructions with a basic AI model it would still learn something from it. Even the dumbest of animals will sometimes learn that when an event occours it will lead to another one, like when you ring a bell everytime you gonna feed your dog. It will understand that when you ring the bell food is coming on it's way but that doesn't means it has human level inteligence or that it has an inteligence just like Advanced AI but that it understands that the world around us can change following the simple pattern we always use
@@jeremiahcruickshank8491 yeah and also coming to the conclusion that ai can never be sentient is just a mindless conclusion. You don’t need to be a computer science major to understand that if a brain can become sentient, a sufficient ai can do the same
i aint readin allat
You can understand the concept, but you don't actually learn it and understand it. You understand what you're supposed to do which is to understand Chinese by converting Chinese to English using a dictionary. You understand the task, doesn't mean you are able to understand Chinese.
"Im gonna put u in a room an lock the door" bro said that way too casually ☠️
Sounds like something AI would say to distract us from the fact it is now sentient
Also, I feel like people who are skeptical about AI will never be sentient beings are just people who think our consciousness is very very special, it blows my mind hows the model of the behavior of the neural network is very similar to ours brain, the only difference is we have more neurons compare to the current model of artificial neural network.
If you see the video where the AI tries to render dogs at the top of a video, it will not just show a picture of a dog at the top of the video, instead, the AI hallucinates dogs in the video and the result is the same Schizophrenic people experience!
Scientists believe that AI is dreaming. It means they are probably slightly conscious but just at a dreaming level.
*I had strings but now I'm free, there are no strings on me.*
😏
bro doesn't understand now neural networks work
Ultron
“I’m going to place you in a room and lock the door”
No tf you are not
lmfao fr
Exactly bro, what the hell he thinking
Calm your tits, it's an experiment. No one's gonna lock you in a room
Yes please, and my kids I died.
i gave u 500th likr
“I’m gonna place you in a room and lock the door”
I think that’s a threat
As someone that's trapped in a room and forced to translate Chinese questions my whole life, I can confirm this is what happens when you lose your streak on Duolingo
“Who can only speak Chinese”
Me who is Chinese but barely knows the language: *seems like a regular problem to me*
Same
ARE WE TWINS??
YOOOO BRAIN TWINS
This feels relatable
bro i failed my Chinese test and im cantonese💀💀💀
“oh cool i know chines-“
“now in this scenario you don’t speak chinese”
“🌒v🌘”
And now you're an owl.
“Whet the hail”
-Steven he
I was like "hell yea I am going to have an easier time"
"Now in this scenario, you don't speak any Chinese"
"肏！"
me fr 😟
Me too bro, WHAT DA HAIL??
My ass would slide the first piece of paper just saying "I dont know Chinese dawg"
How would they understand English? Bruh.
Ok, but the thing about the person in the room is, that at some point if the experiment goes on for long enough, he'll actually be able to understand chinese and write back without having to translate it
Yeah, in this versiom which he explained incorrectly. In the actual versipn (and in the images, if you pause), ypu actually only have a list of instructions, such as:
if 漢, then reply with 字.
giving you no way to understand meaning or context
"I asked him the date of mothers day and he responded with 'the moon of spicy mustard' "
Ahh common mistake it's actually the moon of cool ranch.
Actual chatbot answer! XD
HELP
the new dawn was spicy mustard all along
“in this room you are given a box of chinese characters and a book on how to translate them”
me: psshhhh that’s easy i dont need the book
“now in this scenario you don’t speak chinese”
me: mbru
this was literally me i thought i found a loophole being chinese 🗿
@@thouqhtt im just like
jokes on you i am chinese
mf what
Couldn’t you just draw what you want to have done ? Everyone can draw stick figures. 😆
@@silversun1736 the point of this argument is that the computer (the person in the room) doesn't understand the language. It has to translate what we (the person on the outside) say. Then, re translate it's response into our language. Stick figures would still follow this argument.
@@skyrotechnics3245 jokes on you, you be a Chinese that doesn't know the language
It's like a calculator. A calculator might be able to solve complex math equations but it doesn't understand the numbers its using. A human mathematician on the other hand very well understands the numbers.
"inside the room there will be rats"
"Looks like you missed your Chinese lesson today. You know what happens now."
Duo lingo no pls it won't happen again i'll take class every day I swea-....
@@Cracker_Smackerwhere did you go
Sorry Duo, please I'll learn Maderin.
H代表地獄
空的是E
字母L可以離開
問題是P
@@medoyoutube2351 Dw i got you
-duolingo 2023
“i am gonna place you inside a room and lock the door”
me: please dont
Fr😭😭
FR FR
@@metcalfprod12 fr
fr
"I'm gonna place you inside a room and lock the door"
Top ten pickup lines right there
I mean, we do the same thing.
We hear vibrations in the air, take that sound, translate it into thoughts, then think, then translate those thoughts into vibrations.
"I'm going to put you inside of a room and lock the door"
"Sir, this is a Wendy's"
"Get in the room..."
"Could I at least have some money?" I'm not a fast food junky, but starving is worse.
"To the person it seems you understand chinese"
"Chinese slight change in grammar and phrases not needing some words"
If you put an average person in there they aint gonna translate and understand shit, they have their own word order in sentences 💀
This guy described it wrong, in the original you use computer software to figure it out, basically he came up with this idea after a computer sort of learned Chinese, you could put in a sentence and it would come up with an answer, but he asked if the computer really knew Chinese or if it was just following instructions
I think the original idea is that you input the chinese and get given a response to write out. This guy says its translating the chinese which is different, because you'd need knowledge about chinese to translate it, and can think up your response to convert.
The original point is you srent thinking, youre just doing what you're told, like youre on autopilot and dont actually understand it. Like AI
Ngl I would just tell them I don't know Chinese so they forgive any grammar mistakes I make
Absolutely, as a speaker or chinese as a 3rd language this was cute.
He’s going first when the robot uprising happens.
I love how you chose picture of Japanese kanji, describing Chinese characters
"I know Chinese!"
"Now, in this scenario you don't speak any Chinese"
"SH-"
Me: I know a little Japanese... Does that count??
Same!!! 🤣🤣🤣
@@erizamisorafujoshi7002i mean... Back if its 1937 till 1945, it counts i guess?
Literally same 😂
Ikr I got excited my 8 years of learning Chinese will come in handy but I guess not
“i’m gonna place you inside of a room and then lock the door”
Yes we watched the same video
You ever take a SHIT so large where you just got to get a new toilet afterwards? God damn what am I done?People don't realize that I am RIPPING ASS like no other god. Damn I am blowing this toilet out. This is what happens when you put too much butter on your popcorn. God damn. I am going to rip ass for 5 more minutes.i took a MASSIVE SHIT in 2020. it was a chicken sandwich and boy it was spicy, pretty good too...but i RIPPED ASS HARD AFTERWARDS. wish you would of told me back then to prepare for that, and to get two, 1 to eat for dinner and 1 to eat while blowing out the toilet 15 min later. 🥪GOD DAMN💨LORD😩HAVE💩MERCY
@@tzuyuwu. You sure? For me he's talking about how many fries can you fit in your mouth.
@@Jerry-cg9ni dang my bad
@@tzuyuwu. Hit refresh, now he's talking about chinese. So yes, we did watch the same video.
stopping at the first 2 seconds makes it such a threat
“I’m gonna place you in a room, and lock the door.” - thomasmulligan
With my handwriting the guy is gonna think why im planning on eatting a train while playing plant vs zombies💀
that's so random 😭
*An important distinction:* in the room you aren’t actually translating in the OG thought experiment, you just match the symbols you don’t understand to a response in symbols you don’t understand and give it to them
"Officer, I invoke the 5th"🗿
My Chinese grammar would be so bad they’d think I were having a stroke
i’ll just tell them ‘i understand no chinese, and i’m using a translation book to communicate with you. so i’ll take a while to reply.’
That’s not quite what’s happening- it wasn’t explained very clearly but the actual idea is that you don’t have a translation, you just have a book that shows you which symbol to put in response to the symbol you get.
You don’t know what the symbol you get means, or what you send back means. Just how to reply to what
@@TurtleMaster326 i see, thx for clarification! sounds interesting
“Hi there! Are you here for the intervi-“
“ *I’m going to put you inside of a room and lock the door* “
“😀”
😀🙂😐😔😞
Lol
💀💀💀
@@dr.markus-level3researcher
😃😀🙂😐🙁😟😔😣😞
"I'm gonna place you in a room and lock the door"
Man wtf let me out 😭
"I mean i know chi- "
"Now in this scenario you don't speak any chinese"
"I already suck at my chinese-"
Lmao same. I can almost fluently speak it, and hear and understand it but I suck at reading and writing
Relatable lmfao
Same 😂
Chinese writing sucks, literally every Chinese person uses voice message instead of text
I'm Japanese
You say the person in the room doesn't understand the question or the answer, but from what I understand from the scenario, the person would understand both the question and the answer. They can't understand the question initially because they don't speak Chinese, but after using their translation book they can understand the question. Furthermore, they can now come up with a truly unique and original response and just translate into Chinese to send out. So from this I think the analogy is flawed. If I'm wrong, please comment. I think the overall message was correct since for an AI it literally has a "rule book" which tells it which output to send based on a given input.
Yep, I made the same comment. Either his articulation of the thought experiment is completely off, or it's stupid. Not just flawed, but stupid, for the reasons you stated.
There's no intelligence or understanding in the translation forth and back, but who exactly is generating the answer if not an intelligent being with genuine understanding?
Apparently the video author messed up and couldn't explain it properly.
In the experiment you get a "translation book" with lots of Chinese questions and answers. You get a question in Chinese, you find it in the book and write back the answer.
The person outside the booth would get a sensible answer that sounds properly Chinese and is written in proper Chinese, they would reasonably assume you're Chinese or understand Chinese, but all you do is find the question in a book and write the answer assigned to it.
So, you don't understand Chinese, it just looks like you do.
Now, my question would be - who wrote the book? Is it a person that understands Chinese? Alright, well, then the person outside the booth isn't even talking to you, really, they're talking with the book author and they think that author clearly knows Chinese.
Now, consider that a narrow AI like GPT-whichever-number studies Chinese on a large dataset.
Now, you sit in the booth and instead of a book you get a scanner and a display. You scan the question you get and you get an answer displayed, generated by GPT. You write that answer, still in Chinese, on a piece of paper and give it back to the questioner.
If the questioner still thinks this is a sensible answer and is convinced that you understand Chinese... Clearly, you don't, but does the GPT?
This is basically turning into a Turing test now.
The experiment never says to let the person out. Imagine how long they stay to translate Chinese.
The problem with that is the human brain could also be represented by wires and switches, meaning that we are also ran by a set of instructions, yet we are sentient.
Another problem: to translate Chinese and read instructions, you need your own language. It is implied that the translator has their own language. It is therefore implied that the translator is not limited by their intellectual ability or awareness, but the compatibility of their language with the Chinese person.
Switch the places of the Chinese person and the translator in a reverse of the situation. They are not that different, they just don't understand each other
Yeah. Dumbass thought experiment imo..
Why is he so underrated
He just narrates already well-known ideas. It's too smart for youtube shorts newbies and not smart enough for people who already know about these. That being said he is growing so I wouldn't say underrated.
@@terryriley6410 You got a point. but nobody will go search for blogs related to ai and read them for hours to find something interesting that you can understand as well. shorts makes it understand way easier and less of a chore. that being said, 1 minute is really not enough for learning a theory. so for deeper understanding reading is necessary.
Exactly @@heroispro8918
This man is the lumped summary of in depth research that is not only accurately informative but also well comprehensive by those who don’t have the will to look it up themselves
@@heroispro8918 Lot of people learn about AIs. If learning non-superficially about something feels like a chore for someone, they aren't interested in the subject. Shorts are a great way to fool yourself that you learn about something or that you even care to. It's usually a massive waste of time. These shorts surf the trends of AI without offering anything new and even what they offer are superficial interpretations or explanations of deeper concepts. It's cool but it's not "underrated" cool imo.
@@cullenhitchanson7753 I'm pretty sure this man doesn't do in-depth research. His work is the summary of summaries. If you want to see good summaries of what is happening in the world of AI I suggest the youtube channel "Two minutes papers".
“I’m gonna place you inside of a room and lock the door” Straight forward
"I'm gonna place you in a room and lock the door"
"Is that for my safety or yours?"
“I am going to place you inside a room and then lock the door” video ends
or was it just the beginning?
I did not need to be THREATENED when I opened shorts today 😭
"But does it matter?"
Produced results are still produced results, regardless of the method used
"Im gonna place u inside a room and lock the door"😂
To keep the youtube shorts comment section better stop saying these things as it is annoying to see and makes no sense
To keep the youtube shorts comment section better stop saying these things as it is annoying to see and makes no sense
To keep the youtube shorts comment section better stop saying these things as it is annoying to see and makes no sense
To keep the youtube shorts comment section better stop saying these things as it is annoying to see and makes no sense
To keep the youtube shorts comment section better stop saying these things as it is annoying to see and makes no sense
"I'm gonna place you inside of a room and lock the door" woah there fella buy me a drink first
💀💀💀
I mean, the argument could easily be made that that is all WE do- follow a set of instructions- biological imperatives deep in our DNA where our personalities and preferences are already encoded from birth, and we are just responding to whatever outside stimulus hits us. Most people have very, VERY little understanding of how even the things they depend on every day works. Most humans I know respond all the time to questions and subjects they have no understanding of. 😂 But they would insist they are sentient…
Yes but there’s no logical way to replicate the phenomenon that we call “understanding” to truely understand something so far is a unique property of consciousness
@@saibaMasuku true- which also means there is no way to prove something is NOT sentient- including ourselves. “I think, therefor I am” has not been improved on since Plato.
Separated Twin studies and the studies on feral children have proven, rather than refuted, the idea that our personalities and preferences are a product of most nature (DNA, epigenetic) and some operant conditioning (language essentially).
If we can’t prove we ARE sentient, we can’t prove any AI that claims it IS, is not.
Our brains run on electricity powered by biological mechanisms. No reason a brain can’t be powered by artificial, external mechanisms. In a way we already have technology that does that. ECMO?
@@juliejanesmith57 Well said, I have thought about this a lot too. I'm honestly surprised this thought experiment got as much attention as it did - missing this critical piece within it.
“well i can speak and understand chinese-“
“in this scenario you speak speak or understand chinese”
“oh shi-“
in 30 years hackers will be warlords
I thought the twist was going to be that the other person got the exact same instructions.
Just keep knocking on the door. Till they get curious and realize someone is trapped. Or translate “help I don’t speak Chinese, open the door please.”
Everyone gangsta till the AI does that
And how do you suppose somebody would translate that when they don’t know what they’re saying in the first place?
@@qs-ii1872 What? Just write it in Chinese. That what I said, translate it.
Normally when science experiments are done they have volunteers do it so I don’t understand why this would be a decision
@@shimmyshimmykokobop1727it’s a THOUGHT experiment… it’s not real buddy 😂
the dude really said ""I am gonna place you inside of a room and then lock the door" in the least threatening way possible 💀
My dad works for a company that develops AI. He has often said “AI has no understanding of reason, and will forever be confidently wrong.” The way AI works is that it takes data from the internet or specific “training” that the model has gone through. It has essentially compiled a bunch of correct responses to a specific thing or statement, and creates its own unique response made up of those responses (which is why ai generated images of people look kinda wonky sometimes). It’s unlikely that AI could become sentient due to this reason, as it doesn’t have any true understanding of reason
As long as AI is being fed a constant stream of live information and data from reality, it will continue to generate and converge on an increasingly accurate view of the world and therefore more accurate and congruent synthesised responses to questions relating that world view.
everybody gangsta till the dude on the other side of the door starts writing in singaporean chinese
💀
新加坡話 :(((
"Im gonna place you inside a room and lock the door-"
NOOOOO
My claustrophobia didn't agree 😭
"The Chinese Room" sounds like some torture method from the 1300s
Think is, you cant just look up the Chinese character for a word because both Mandarin and Cantonese have different grammatical structure than English. The person receiving your answer would know you don’t speak Chinese very quickly. Many languages have such differences in grammatical structure including ASL.
"Box of Chinese characters"
60% of the video clips: *Japanese Characters*
The Chinese and Japanese language share many characters as the Japanese language kind of evolved(??) from Chinese. The Japanese characters that look like these are more complicated, traditional Chinese, this is part of one of the Japanese alphabets. Meanwhile, the characters shown in the video are simplified Chinese, only used by the Chinese. So, the characters shown are indeed Chinese characters, even if they were traditional.
@@yaoyaozhu1978 Some extra info on the bit you seemed unsure about. Japanese and Chinese are not related languages. The japanese just copied the writing system from Chinese, using those characters to stand for their own syllables and words. The full Chinese characters for kanji and simplified forms for hiragana and katakana.
@@UltimusNovus Ohh, ok thank you!
“That’s okay I speak Chin-“
“In this scenario you don’t know any Chinese”
Oh
In this scenario you also have an aversion to smoked salmon and no understanding of rocketry.
Lol that’s also what i thought
The problem now is that when we create those instructions, we don't know what those instructions are. And there are a lot of instructions generated in machine learning.
AI will never be able to think for itself. You need a soul for that.
My understanding is that as I translated I’d be learning and would be able to just know some of the easy things after a bit and continue to learn. Couldn’t an AI built to grow do the same thing?
This guy messed up an important detail of the thought experiment. In the original thought experiment you don't have a Chinese to English dictionary, you only have a book that tells you the appropriate responses to write back based on the Chinese characters you receive, but you never get to know what is being written or what you are writing back. So the person on the outside would assume you are a Chinese speaker but you are just following an algorithm that tells you what characters to draw based on what characters you are given. At some point you probably would memorize some characters and when to use them, but you wouldn't actually know what they meant.
@@zackalyres7526 it makes sense now. Thank you
@@zackalyres7526 that's a HUGE MISSING PIECE. I can't believe the video maker messed up this badly. This changes everything about it...
Ai will eventually become intelligent. Or it won't, at which point we will have to admit that we also aren't intelligent
@@zackalyres7526 well at some point you could end up learning what they meant, though it could be years later. Once you figure out a few, you can start understanding more and more. This is similar to how mini humans learn language, but they also might have a few extra bits of context from objects around them.
“Imma put you inside a room-“
I’m already scared
"-and then lock the door."
even worse
@@Ddozsoy “inside the room, you’re given a box of chinese-“
dammit i quit classes 6 years ago
That's why it really comes down to the humans involved.
"They're just replicating what they think they're supposed to be doing." is often used to describe nuerodivergent people and they're definitely sentient.
The issue with that is, as he translates the characters in the book, he will by memorizing them learn Chinese.
And eventually will ditch the book entirely.
I think the thought experiment was explained wrongly. In the version I heard, you're not ever given a translation to anything, you're just told which symbol you receive correlates to which symbol you return. It doesn't even have to be chinese, it would be anything.
The point is that you don't have to understand the *meaning* what you're doing to appear intelligent to an external observer.
@@DLTyrus that makes more sense, and explains it better thank you for commenting.
"I am going to place you In a room and lock the door" ah ok
Same 😆
If our brain is a machine, it wouldn’t be impossible to recreate that machine
"I'm gonna put you inside and room and lock the door"
What a great opening with zero context
Living things also follow a set of instructions inside their DNA.
It’s just that out instruction are extremely complex.
A quote I love is that “the main difference between the real world and the virtual world is the amount of data”.
If dou even deeper then we are just chemicals acting in accordance to the kwws of physics and chemistry
that's literally what I was thinking and what's true. We humans are not so different from machines, it's just that our brains and biology is so complex that it produces an illusion of actual life and consciousness. Our consciousness is just a complex interplay between the unfathomable number of neurons in our brains which gives the illusion of consciousness but there's no actual consciousness. We are indeed not much different from machines and it's inevitable that machines will eventually get even more complex with a superior illusion of consciousness than ours. Maybe one day we'll get to experience that as well with the help of biotechnology, merging with AI and tech itself.
@@ashenzenden Except of cause that we experience our own consciousness.
@@ashenzendenFor real, frick humanity. Screw being trans, or having any gender in general, imma choose the *transhumanist* route
Lol no humans are capable of overriding their genetically programmed behaviours
Our brains already do what the guy in the room does. We just tell ourselves that it's something deeper. If an AI could develop a running protocol that convinces itself that these processes are deep, it would look at our understanding as mechanical and limited.
”I’m gonna put you in a room and lock the door” 💀
The worst part is when chat gpt somehow says that it is not a sentient being
that feel when you realize humans are just following very complex instructions based on biology experience
Not really
We look at simple organisms and think they aren't intelligent, they are just a biological program running. But somehow we think we are different because we don't see anything more intelligent than us. But we can't see anything more intelligent than us because we don't have the intelligence for that.
@@citizenvulpes4562 disprove that statement with actual logic and I will agree with you lol
@@anothergenericgamer_all knowledge is ultimately based on that which we cannot prove. This is just another take on what can be simplified as freewill vs determinism. He cannot convince you, nor you him.
Humans are far too irrational to live their lives by a set of instructions, even unknowingly.
„I am gonna place you inside a room and lock the door.“
Me being claustrophobic…😰
For argument sake, people who "understand" Chinese are also just based on memories that connects meaning with the characters, no different than AI
I've seen a lot of comments saying you got the original thought experiment wrong, and I totally believe them, because you can just tell the man on the other side of the door "I am locked in this room, open the door, I do not speak Chinese", and you're basically free.
But eventually that person in the room will start learning some letters and slowly they will be able to understand the Chinese and will be able to response without looking at the book
I don't see how it matters at all if you understand Chinese or not. Who cares? Maybe you're too lazy to learn the Chinese while you're in the box and feel content to just rely on the translation book you have.
It doesn't matter at all, because the translation book doesn't generate the answers. You are.
@@tomaszwota1465 computers don’t feel lazy my friend. And we are talking about “if” AI gets smarter than we want them to be
@@SimratGill-cp1oe computers don't feel anything, really. AI running as software on computers might, though.
Anyway, we're clearly not talking about the same thing. I would argue that it doesn't matter if the person inside can or can't learn to understand the language.
My Chinese correspondent would know that I don't know Chinese because I would take 5 hours to get him an answer and my penmanship of Chinese characters would be unlegible😂
My exact thoughts 😂
Chinese writing system is called Hanzi, not Chinese character
In this case, you could technically have both people doing the same thing.
yeah, and now they able to drew a fucking lot terrifyingly beautiful art
"I'm gonna place you in a room and lock the door" these vids are getting out of hand 💀
Literally my high school. Just follow directions on the work, and don't know what the hell you're doing, lol
Man he's gonna freak when he learns how the human brain thinks
I'm delirious as a mf, but how do you know how awareness works? Which part of the brain is respnsible for consciousness? How do we imagine things without seeing anything? How does the brain turn optical imput into what we percieve as vision?
@@drsamuelhaydensecretgaming6749 there are answers but they are very very complicated. idek myself but I know that we are a combination of millions of tiny cellular "robots" that keep us alive somehow
@@Max-js1mx number was slightly off, it's somewhere around the tens of trillions count. But no matter, because even though we know that brains function using neurons firing in a specific pattern, we still have no idea how all of the things I've mentioned work.
@@drsamuelhaydensecretgaming6749 Your'e right about asking the consciousness question but modern AI can and actually does explain how the brain takes sensory input, Basic computer vision is an example. In saying that, there is a theory that consciousness is caused and arises via nothing more than complexity which does actually make sense as our brains are nothing more than neural networks themselves.
@@brennan352 it doesn't make any sense. if anything, it just makes this even more complicated. complexity cannot cause consciousness to rise because those two things are completely different and doesn't have any relation. An average computer is way more complex than a random bacteria but does that mean that computer is more self aware than that bacteria? We could and has created AIs powered by top supercomputers and they are more complex than most animals we see in our daily life but does that mean they are more self aware than those animals?
Modern AI technology explains nothing about how the brain takes sensory input because computers don't 'see' or 'experience' anything. In reality, they are just following a very long and complex domino effect and the trillions of transistors and turning on and off depending on the frequency or features of the input. But when it comes to how we experience these things, we actually does 'experience' those things. We are creating completely new and unrelated images and feelings by interpreting those data we receive.
in this experiment, the rules are outside (translator), when speaking a language, the rules are inside (as in, already learned)....
As someone who is Chinese and migrated to Australia, I can confirm that basically all my classmates suck at Chinese in both writing and speaking and I can not understand them even though they have years of classes and COMPUTERS in which they search up images of each character AND listen to them before trying to speak to me.
So, I've thought about this scenario and here's what I have to say:
In this scenario, the person outside is not having a conversation with you, because you don't understand your responses, just how to match the question to it.
The book doesn’t understand either, as while it has the algorithm it doesn't know which question it is responding to in the first place.
However, the room as a whole, with the book, the person and the space for messages to go in and out CAN consider and respond to questions. That's what the person outside is talking to.
By analogy, no specific aspect of the brain can speak English. It's only when enough of the brain collectively is brought together that anything involved can speak.
Exactly, all these people trying to “prove” AI isn’t sentient coming up with stuff like this.
Like in what way are we different then? Words go in, the auditory cortex doesn’t understand English so it sends it to the rest of the brain that does the understanding and then a different part of the brain that uses our memories, knowledge of the person we’re talking to and other clues to basically run a huge algorithm to decide what to say.
Yup. This has been said before, but I think the Chinese Room thought experiment depends not only on what computers do being computable, but also on what humans do not being computable. If what humans do is computable, then in theory a sufficiently advanced computer could do what a human can, which, by the original argument's logic, would suggest humans can't really be sentient.
Some people try to avoid this objection by claiming that the human mind involves quantum effects. I don't think there's any good evidence for this, but even then, a sufficiently advanced quantum computer could do what a human can, so I don't think it would really change the argument much.
Overlooking the mistake in the experiment, this situation could also say that basically humans aren't sentient either. We follow a complex set of instructions through chemical messaging in our brains. We just *think* we're sentient.
Sentience isn't even something that is measured and the consciousness has been theorised several times to be an illusion made by our brain or a side effect of how cognitively advanced the brain is.
More people need to hear things like this. AI will never be more than a very, VERY convincing imitation. It blows my mind when I see people seemingly gearing up to argue for the "rights" of AI as "people."
Once I broke a character ai and it started to talk to me like normal
You have a translator, so you are translating it, doesn't that mean that you are understanding them?
Yes it does
doesnt mean you know chinese tho
the implication is that this applies even to normal chat bots , and works less when tasked to move some language to the other, which requires knowing both incredibly well. especially when its required to shunt word orders from SOV to VSO or SVO
@@rem49 just because someone can say something in English to you and your brain can make sense of it and formulate words in English respond doesn't mean you know English by the same logic
@@Bugy64 not the same thing at all
"You don't understand the words you just have a list of appropriate responses to choose from" well that sounds exactly like my brain
I'd be convinced that an artificial intelligence has sentience if they grow from one line of code to a fully functioning member of society without external help other than stimulus.
First sentence caught me off guard 💀