Can Computers think? An Introduction to the Philosophy of Artificial Intelligence | Alex Abbott | Skillshare

Can Computers think? An Introduction to the Philosophy of Artificial Intelligence

Alex Abbott, I like to think!

Can Computers think? An Introduction to the Philosophy of Artificial Intelligence

Alex Abbott, I like to think!

Play Speed
  • 0.5x
  • 1x (Normal)
  • 1.25x
  • 1.5x
  • 2x
7 Lessons (59m)
    • 1. Introduction

    • 2. What is a Computer? What is Artificial Intelligence?

    • 3. The Turing Test

    • 4. Problems with the Turing Test

    • 5. The Chinese Room Argument

    • 6. Critiques of the Chinese Room Argument

    • 7. Are we living in a computer simulation?

  • --
  • Beginner level
  • Intermediate level
  • Advanced level
  • All levels
  • Beg/Int level
  • Int/Adv level

Community Generated

The level is determined by a majority opinion of students who have reviewed this class. The teacher's recommendation is shown until at least 5 student responses are collected.





About This Class

Have you ever thought about Artificial Intelligence? If so, you must have been confused as to whether or not Computers can think. In this course we shall be exploring different issues in the Philosophy of Intelligence. 

We shall explore a number of issues including whether or not a Computer can be said to think if it passes the so-called "Turing Test". We then look at the problem of the Chinese Room argument against the Existence of AI before finally exploring the question "Are we part of a Computer Simulation?"

This short, 7 lesson class will give you all the basic philosophical understanding you need to engage with the philosophy of Artificial intelligence. I believe anyone who's interested in Computer Science and Artificial Intelligence/Machine learning will enjoy and get something out of this course. Of course this class is available for all who just want to take an introduction to the subject. 

NO PRIOR KNOWLEDGE OF PHILOSOPHY IS NEEDED FOR THIS COURSE! I want to make easy to access, easy to understand Philosophy courses which are interesting. I do not intend to start work on long, dry and boring Philosophy lessons that aren't popular. The point is to teach others why engaging with Philosophy can be a positive experience, and doing that by conveying short, interesting lessons is the best way to do that.

Meet Your Teacher

Teacher Profile Image

Alex Abbott

I like to think!


Hello, I'm Alex. I have a BA in Philosophy from the University of Nottingham and am currently in the process of getting an LLM in Law. In the future I would love to pursue academic studies further and go on to do a PhD. I have a keen interest in teaching people what I have learned in fun and interesting ways. My primary expertise include Metaphysics, Logic, the Philosophy of Mind and Ethics. I shall be making courses on some fun and interesting areas of Philosophy. 

My Current Courses include: 


- Can Computers Think? Introduction to the Philosophy of Artificial Intelligence.

- Who am I? Introduction to Personal Identity

- A basic introduction to stoicism

- Introduction to Formal Logic

- Introduction to the P... See full profile

Class Ratings

Expectations Met?
  • Exceeded!
  • Yes
  • Somewhat
  • Not really
Reviews Archive

In October 2018, we updated our review system to improve the way we collect feedback. Below are the reviews written before that update.

Your creative journey starts here.

  • Unlimited access to every class
  • Supportive online creative community
  • Learn offline with Skillshare’s app

Why Join Skillshare?

Take award-winning Skillshare Original Classes

Each class has short lessons, hands-on projects

Your membership supports Skillshare teachers

Learn From Anywhere

Take classes on the go with the Skillshare app. Stream or download to watch on the plane, the subway, or wherever you learn best.



1. Introduction: welcome to the introductory video for the course. Can computers think an introduction? Toothy philosophy off artificial intelligence? I'm just gonna go through in this little video what this course is going to be about, the things they're going to be in the course. I'm really the requirements that you might need to the sort prime knowledge, maybe for the course. All that's going to go in this bill short video before we move on to the actual details. Okay, so why the artificial intelligence so important? Well, this course is trying to merge the disciplines of computer science with that of philosophy . Okay, because artificial intelligence really is an extremely complex on interesting area of computer science on it brings into question a bunch off doubts really about our understanding of mind and consciousness. So it does have a lot off philosophical implications, because, really, it questions our understanding of mind and our understanding off what the mind is on how it works. It's also been a hotly debated topic for nearly 100 years as well. Look in this course, we're gonna go back quite far to have a look at some some quite old examples off the philosophy of artificial intelligence on an example is the entrepreneurial on musk is very critical of the dangers that could be faced with the invention of artificial intelligence. He's very critical of the idea of artificial intelligence, Andi. The yet leather dangers that that could bring. So what we gonna learn in this course, what the objectives? So the course has a number of objectives. It really serves as an introduction to the deep and interesting topic, which is the philosophy off a. I, the philosophy of artificial intelligence. So we're gonna learn all about how we can understand computation and artificial intelligence on how we can understand the various arguments for and against whether or not it's even possible for artificial intelligence to exist for computers to actually be able to think. Okay on. That's the central faces off this question. Can machines think on This comes from a paper by Alan Turing, who were going to look at in the first lesson. So the actual detail course of this structure off structure of this course is it's going to split into six lessons. Okay, so the first lesson is to be really what is a computer and what do we mean by artificial intelligence? Can we make any kind of distinctions between different types of artificial intelligence? The Turing test is a very important part off understanding, artificial intelligence. So what does the Turing test say? Andi, can we give an example of how it works? Are there any problems with the Turing test, which will be the third lesson on then? We've got some or modern philosophy on this issue. So the idea of the Chinese room argument against artificial intelligence and then again critiques with that critiques of that argument as well. And then finally, we're gonna look at quite an out there kind of lesson, which is the idea of Are we living in a computer simulation? Which is a very abstract philosophical idea that really does have a lot of quite fun and interesting interpretations. It should be said as well that for this course, there's no prior knowledge of philosophy is needed. It is gonna be very hands off the court. The sort of detailed, dry philosophical texts were going to try and summarize it all in is as easy as possible, and we're gonna look a really, rather than trying to understand the arguments in lots and lots of detail. This is an introductory course. So we're gonna try and understand the basics of the arguments on, then applied them to some real world examples and some examples in science fiction so we can get a better understanding of the issues that people will face when we're looking at artificial intelligence. Okay, so the test for the end of this lesson is just to research What Ellen was set about artificial intelligence on his arguments for why we should be worried about it. On a very good place to start is to look at the Joe Rogan podcast episode with the Long Mosque on really what he said about artificial intelligence. 2. What is a Computer? What is Artificial Intelligence?: Let's have a look at the first sort of definitions that we're gonna have to be looking at for this course because we're looking at the philosophy of artificial intelligence. We really got to work out what artificial intelligence is, how we consort, distinguish different types of artificial intelligence. That's what we're gonna do in this lesson here. Okay, so in the lesson, specifically, we're gonna look at on examine a number of different issues. So the first we're gonna look at how we should define a computer on how or the philosophers in the past of defined computers, Ross is gonna have a look at the understanding the meaning of artificial intelligence Was it mean for something to be artificially intelligent? And then we're gonna distinguish two quite important concepts that if the distinction between strong I and weak ai Okay, so this is something that was gonna come up a lot throughout the rest of this course. So defining a computer for the first place we should really start is by looking at the dictionary definition. This is what the dictionary definition says. So it's a electronics device for straw storing and processing data, typically in binary form, according to instructions given to it in a variable program, so this is quite a nice place to start. However, this is not really we're gonna finish when we're looking at how we can define a computer, how we can look at really what it means to be a computer. There's a number of problems with this. It's quite a very broad definition, simply just defines it as a machine that processes data on this data is usually transmitted in binary, and that's about it, whereas there's quite a lot mawr that weaken, solve, get from our definition off a computer. So how can our understanding of artificial intelligence helpers to deeper our definition off computation off a computer? Well, let's look at artificial intelligence now and then develop what our understanding of artificial intelligence on, then come back to whether or not we can change and get deeper definition off a computer itself. OK, so why is artificial intelligence Well, artificial intelligence is a theory that some computer systems may be able to perform tasks that usually require human intelligence. That's the basic pragmatic understanding on There were two types of a guy that we look at when we're looking at the philosophy of artificial intelligence. So there's the idea of strong artificial intelligence on the idea of weak artificial intelligence. Strong I What we mean by strong I is that a digitally program computer necessarily has mental states. OK, it's a very strong definition. It states that with strong artificial intelligence, we will literally have mental states. We literally have a computer that could understand that has a mind. So it has everything that an organic human intelligence would have a mind, mental states, ability to reason and understanding on. Do you know the ability to use logic on to come up with original thoughts? This is the idea of strong artificial intelligence, so a computer with strong artificial intelligence literally thinks has a mind on would understand a human having a mind. Okay, so it's the same kind of understandings when it comes a week. A. I week. I doesn't make the same kind of commitments, a strong I it really all week, ai means is it doesn't posit that computers literally have mines, but rather that computers could simulate and replicate mental states on replicate artificial intelligence and cause simulate him. So really, with that being said. How can we use thes two understandings to look at what actually is a computer? Well, there's a number of philosophical ideas that we can look at here. So the philosopher Tim Crane, in his work, the mechanical minded 2016 asked the question, You know what actually is a computer? And he believed that we should forget about what he called contingent properties of computers. So a screen or a keyboard or a graphics card, for example, because these are things that could be replaced and you know they could be. You could have a computer without a computer screen. You could have a computer without a keyboard, for example. So these are these are things that would be called contingent properties of a computer. What we want to know is we want to know what the actual properties that every computer has . Okay, what makes it necessary? What are necessary for it to be a computer? And he came up with a number of properties that are computer house. So a computer is a device which process is representations in a systematic way. It is an information process. Er it is a manipulator of symbols. Okay, on it, really, you know, it can systematically manipulate Onda transform different symbols into different ways. It can process information in different ways on it represents information in what was known as a systematic way. Okay, so this builds upon our original Google definition off a computer. We looked at what computer means when we have. I saw that the basic definition, the broad definition of a computer as simply something that processes data on that data is usually processed in binary. But we can also add layers to it. And that's what Tim Crane did in his book Mechanical Minds. So how can we understand these processes of computation through the lens off strong I, because strong I is the most important that we should be looking at. So computation strong I according to Strong I, the processes that occur when a program well, a computer exercises a program are genuinely intelligent. It's genuine understanding and intelligence. It generally understands what it's doing and congenial. Only think about the process is in question, so it's exactly the same as how human would understand it. So we could look at some examples of strong I in science fiction. Like I said, I didn't want to be using the sort of dry philosophy and going through lots of philosophical papers, but rather taking the most important bits out of it on then, applying it to real world fiction. You know, on science fiction for the real world fiction and science fiction. So a very good example of strong I comes from the film X makina. Okay, Andi, we're gonna look a talk about this and a lot more detail. Another lesson. But in this film, the robot Ava is perceived toe have strong artificial intelligence. This means when she's performing actions and tasks, she's genuinely thinking, understanding what she is saying on what she is doing. OK, so this is a good example of how we can, you know, conceptualize strong artificial intelligence. This'll version of artificial intelligence is actually heavily criticized, and we're gonna look at a number of criticisms later on in this course. So in this lesson, we've learned about what computer is, how we can define it on we can, how we can define artificial intelligence on. We've also a lot of the distinctions between strong eye on Dweik ai. So the questions for this lesson is really Do you agree that computers have strong I or that Could could computers ever have strong eye on? Are there any reasons to reject the idea of strong artificial intelligence those air? Two important questions that really we're gonna look at Mawr Maura's. We develop this our understanding through this course. In the next lesson, we're gonna look at how we can test whether or not something is genuinely intelligent, and then we're going to start by looking at the Turing test. 3. The Turing Test: Welcome back. What we're gonna do in this lesson is have a look at the Turing test. So really get start delving into the philosophy of artificial intelligence and really, where it starts When the last lesson. We looked at the definitions off computation on a computer on the definitions of artificial intelligence, as well as trying to distinguish between strong artificial intelligence and weak artificial intelligence. In this video, what we're gonna do is have a look at the Turing test. So in this lesson, more specifically, we're gonna have a look at who Alan Turing Waas Why he's important to computer science and the philosophy off artificial intelligence. We're gonna look at his paper kind of machine think which is hugely influential. Andi really almost kickstarted the philosophical movement behind the idea of thinking machines, especially the sort modern understanding of the philosophy. We're gonna look at his imitation game really a waiter work out whether or not a machine can think on how eventually turned into the Turing test on going finally look really what the Turing test means. What's the significance off the Turing test before we go into the criticisms in the next lesson? So Who? Waas Alan Turing. Well, this is Alan Turing right here in picture form. So Hollande tearing is no today as almost the father of modern computer science and computation. He was born in 18 12 and died in 1954. And it was during World War Two that he really made his name. He worked for the U K Government Code and Cypher School at Bletchley Park. Andi played a key role in cracking Enigma, which was themed Nazi code. May, you know, code making device spoke device wrong there, which is really embarrassing theme Nazi code making device believed to be unbreakable bond . He used computation toe workout after the war. He also wrote extensively on the issues of whether or not a machine can think so, he wrote in his paper in 1950 Cal Machine. Think so? He asked this question. So he begins. He believed that the first step to answering the question whether or not can a machine think, was not to define our terms. So normally people would go about saying, while kind of machine think well, that's first. Have a look. What do we mean by a machine and what do we mean by thinking once we've defined these terms , then we can work out whether I'm no, I'm machine can think turn. Didn't think that was a particularly useful starting point. He believed that this was counterproductive. He believed that it would. In defining our terms, we'd have just issues with their definitions and just take it away from the actual question hand. So rather than if we were to start by defining our terms, the idea of a machine and the idea of thinking we'd end up just disagreeing in the philosophical community just end up disagreeing about you know, the definitions rather than disagreeing about whether or not they can actually think so, he said. He took away from that. He said, No, we're not gonna win. We're not gonna define our terms. We're going to start by looking at a different path, and this is what's really interesting. So instead, he begins by asking another question in order to come to some conclusion. But the with the original question, and he calls this the imitation game. So Turing sees that we can describe the problem in the form of a game in which there are three players you've got player one or player A Who is a man? You got player B. Who is a woman on You've got player. See who is a Nintendo gator. Who can, you know, could be either sex. Okay, on the point of the imitation game, the object is for the interrogated toe workout. Which one is a man on which one is a woman? So the man on the woman go into one room and the interrogator goes the other on They communicate via, you know, voice. Okay, so before they play the imitation game, they're just simply known labeled as Person X and person. Why on at the end of the game, the interrogator has to label them. As you know, X is a X is the male or wise be wise, the female or any other combination off the two? So that's the point of the game. He has to try and work out what you know. What is that? The sex off A and B or accept why OK on in this game is also a job the man's job to try and confuse the interrogator into making the wrong decision. Andi, during say's that will happen if we just replaced a with a machine on replaced the object of the game rather than trying to dis decipher which one is a man on which one is a woman? Could we work out which one is a machine on which one is a human? So will the machine version of a A. So them be able to do just a good a job as the rial in tricking the interrogator on toe also cater for variables. Touring suggests that we should measure the machines success on on the number of times it was successful in being able to trick the interrogator. So say, if we did this test 100 times, what percentage of the times the off the times did? The machine managed to trick the interrogator, and this is where the Turing test comes from so we can alter the game toe. Ask which of these a or B is a machine on this version became known as the Turing test. Andi, if we try to work out on you know, a more pragmatic, more modern understanding of the Turing test, we could look at it a little bit like this. Suppose you're texting to people person A and person B. One of them is a human, and the other one is a machine is a computer. The machine is programmed to trick you into thinking that it's a human on. What you got to do is you got to spend little time upwards of around five or 10 minutes asking them both questions. Okay on dure. Supposed to come to a conclusion. Which one is a machine on which one is a human, and it's believed that the computer will pass. The test will pass the Turing test if you cannot reliably know which of them A or B is a computer. Andi, If it passes the test, it passes the test for intelligence. The idea that this machine is intelligent, it is artificially intelligent if it passes the Turing test. So this is the sort benchmark from trying to work out whether or not the computer can think you're included. In fact, he concluded in his paper by suggesting that the original question kind of machine think he said that it was meaningless and it didn't deserve any discussion, which is kind of annoying. Sees that this is the central thesis of this entire course, but we'll skip over. We'll skip over that remark made by touring. So what can we really take from the Turing test? What can we understand it to be implying? So there's two ways we can read into the Turing test. We can take a constitutive claim a constitutive, take clean, all weaken, taken evidential, claim. So the former we could suggest that the Turing test and passing the Turing test is constitutive constitute Arqam Pronounce the word constitutive off intelligence i e. It means that an intelligent machine must pass the test in order to be truly intelligence. However, not many people really accept this view. It's not really accepted in the paper itself. It's not really, you know, it's more. It's more of a general understanding. In the latter, we could say that the test passing the Turing test simply is simply evidence of intelligence. So it's a sort of base mark understanding. So one reading is Mawr Asia a lot stronger than the other. So we have the constitutive claim, which claims that you have to passed the test in order for a machine to be deemed intelligence or you've got the claim where if a machine part of the test that some evidence of intelligence but doesn't necessarily prove intelligence either. So that's really what we could look at with the Turing test for the discussion. Is the Turing test correct? Is he correct in saying that it will either constitute or be evidential to a machine having intelligence? Is that a correct understanding? Okay, Will the Turing test either yield on evidential claim about a machines intelligence? Or will it be more stronger? Will it actually constitute being a an intelligent, intelligent machine if it passes the Turing test? This is a very interesting discussion. Can you also think of anything which could possibly pass the Turing test today? That's a very interesting question, because there's a lot of controversy about whether or not something could pass the Turing test today on Really? What if, if so, what would it be? Leave that is in the discussion section. So what could pass the Turing test today? In the next lesson, we'll look at the critiques on the problems with the Turing test on. We're gonna apply this again to the film X Makina on how they understood the Turing test, since that was the effectively the whole plot off the film 4. Problems with the Turing Test: Now that we've gains understanding of the Turing test, what we're gonna do in this lesson, these have a look at the issues with the Turing test on, we're really gonna break this down into three sections. We're gonna be looking at trying to attack the tearing test from a number of different angles to quite polar opposites. And then one that solve comes from left field. Okay, so the 1st 1 is that the Turing test is too difficult. It's too hard. The second is the Turing test is too easy. On the third is what we called lady Love Laces. Objection to the Turing test, which is one that isn't particularly very strong. But I thought we cover anyway because it's quite a famous example. But yeah, we'll also cover why the objection isn't particularly strong. I'm gonna finish off by applying this to the movie X makina and having a look at really what the Turing test the role it played in that movie because if anything, you know, life imitates art. So we should really be looking at science fiction to try and develop our understanding off the philosophy of artificial artificial intelligence. So the first question. Is the Turing test too difficult? Is it to heart? Some suggest that it's too hard to pass. If it's too hard to pass, How can it be reliably used to determine intelligence? Okay, so the suggestion that is too hard is also especially difficult to overcome, since the reason why it's too hard, according to the critics, is that it's to do with the very nature of the test itself that makes it too difficult, so we can't just turn it down and make it a little bit easier. The whole nature of the test and the way it's set up makes it too difficult. According to these critics, the test requires the ability to converse with the people toe, have a conversation. However, it's still possible, toe have intelligence without having the ability to have a conversation. This is seen by critics is quieter at Anthropocene centric way of finding intelligence as it is quite human centric way of finding intelligence is what human would look at when would looking at intelligence. So I really couldn't we have an understanding of intelligence without looking at the sort of human aspect of it. Can we Can something be intelligent without having to acquire the language ability to pass the Turing test on that, really, that the issue at hand here? Because some would argue that, yes, you could have an intelligent machine that doesn't passed the Turing test because it doesn't have the ability to acquire language and use language and expressed language in a way that is replica tive to human language. So that's a problem that we have with the term test we also have the counter issue is some suggest that the Turing test is actually quite easy. It's a little bit too easy. It doesn't look very easy. According to this, this unhappy computer here who seems to be studying for the Turing test. However, some people suggest that it can we really be able to determine intelligence by just having a short conversation with a machine. Is language the only thing that determines intelligence? Can a machine just be on intelligent but still have the ability to possess language? That's an interesting point. So if the Turing test is way too easy, since it doesn't allow us to use aspects of other aspects of our intelligence, you could have a super super intelligent machine, a super intelligent computer that's able to do every single everything except for have a conversation except for using language on their forehead, fell the Turing test. However, it's still intelligent machine. You could also have an intelligent machine that is so intelligent that it knows what the Turing test is. Andi would be able to deliberately fail the Turing test in order to try and fool the silly humans that are trying to perform this test. So this is, you know, it's a very it's a very difficult one to apply in practice. It's quite paradoxical, really. Scenes, as both of the objections are somewhat linked in their reasoning. The idea of being too hard or too easy it both. They both come to really similar interesting points that the main problem with the Turing test is is language and is having a conversation. The only way we can, you know, determine whether or not a computer is intelligent, whether or not machine can think, some suggest that this lends it to being too easy, and others suggest that it lends it to being too difficult. So they're both very good ways off trying to strike down the Turing test in as a tow. Apply it in the real world. We also have Lady Lovelace is objection on leave of lace, said the Really. She said that machines cannot be intelligence because everything they do, no matter how intelligent they are, everything it does is programmed to be to buy an intelligent designer before hunt. So a machine lacks originality. Everything that we program a machine to do already, we've programmed it to do it so we can program it to do something really intelligent. But that doesn't mean that the machine is intelligent. That just means that we're intelligent enough to program the machine. And that's the an interesting paradox school issue with the Turing test, however, it doesn't really make much sense. Seems there are lots of different objections and counter points you can make to this. So really, what? All the issues with this argument well, they're quite few. So for some people who have religious, they believe that humans are also unoriginal. If humanity on the universe was created by an intelligent designer, I God that how could we possess any kind of originality or free will if you take the Christian understanding of predestination. God already knows that you're going to heaven or hell. Onda already knows what you're going to do and knows every single thing about you years on this scene. So humans don't have any kind of free will. So we don't have any kind of originality. So by that logic, and if we apply it to sort of a religious understanding, does that mean that humans are not intelligent either? And you don't even have to be religious to accept to accept something similar to this. You could say if you're a strong determinants, someone who believes that everything in the universe is already predetermined and then we have no free will, nothing could be original in a deterministic universe. So you can see that there are a lot of ways we can really attack the love laces argument on . Then it becomes. Then it becomes an issue that isn't off an issue of computer science or philosophy of computer science. It becomes a problem about free will and free will in action. So really, it's a very slippery slope with love laces argument to try and prove it wrong because we end up just talking about a completely different area of philosophy, an interesting area, but not an area that really would look focusing on in this course more of A. Just because computers don't express any originality, this doesn't mean they're unable toe ever express any originality? This is a point that Love least brushes over. Really, she doesn't seem toe. She takes quite a static view of science technology that at the time that she wrote what she wrote. Computers lacked any kind of originality, but that doesn't mean that technology, as it develops, could create computers that maybe will be able to express originality. That's another issue with love laces argument. So moving on, we can apply the Turing test to our soft science fiction understanding our science fiction case study, which is the film X makina. Okay, and the whole central plot of the film is to try and see if Eva, who is this one here, passes the Turing test. So in this movie, the researchers are tasked with trying to see if Ava can pass the Turing test on what really interesting is because, as we at the viewers know, that Ava clearly expresses and would seem toe be on example of strong artificial intelligence. They tried to modify the Turing test in a way to make it even harder to pass. So rather than having a conversation behind closed doors with a real human, okay, and the test on the test subject doesn't know who it is that is doing, you know who's having a conversation with. What they do is they decide to raise the bar of expectation. They instead show the test subject that she is actually indeed a robot. She showed their shown. Which one is the robot? Okay, on it is up to a ver to them prove that she's still has artificial intelligence. So it's sort of a solver mixing up of the Turing test in a way so it removes a fundamental part. The test, which is if we're to take to to a human and a machine. We had swapped them round, so we don't know which one is which. Can the machine trick the human into believing that it's human? It takes that part away on it, says what We already what happens if we reveal which one is the robot on which one isn't the robot on? Then try to get the robot to prove that it is intelligent. That's a very interesting point on it's one that goes, You go into in a lot more detail in this movie, which is actually a lesson task for you to do. If you can watch the movie because it's very interesting, it's very good. Okay? And while you're watching it, take note of the significance off the Turing test on the ways in which the researchers devised the test in a way to ensure that intelligence can be found very interesting points that could be made on that could be discussed there. In the next lesson, we're gonna argue against the existence of strong artificial intelligence by positing the Chinese room argument. 5. The Chinese Room Argument: in this lesson. I want to look at the Chinese road room argument against artificial intelligence. Okay, this is against the idea of strong artificial intelligence. I mean, this lesson we're gonna introduce the argument will have a look at the argument against Strong I. And then we look at the thought experiment that goes with the argument in the Chinese room thought experiment. It was an introduction. John Cell, who was the philosopher who developed this? He also was the one who distinguish between strong and weak AI. So he was the one who argued that there is a difference between strong I and weak AI on Let's have a look again at what so says about strong and weak ai so strong ale is the quote from him on appropriately programmed computer really is a mind in the sense that computers given the right programs, can literally be said toe. Understand? That's what source said in 1982 quite modern philosophy illness. When it comes to a week, AI computers are just able to help us to model and simulate mines, then no, actually minds in themselves. So that's the distinction on what soul is doing in this argument that we're gonna look at the minute is he's arguing against strong artificial intelligence. So he argues against the existence of Strong I. Andi uses the first article argument along with a thought experiment. To explain this. We're gonna first look at the thought experiment before we turn to the actual official formal argument in question is called The Chinese Room Thought Experiment. So let's start by just making a number of points. And then we're gonna explain how the argument on the thought experiment works. John sold does not understand Chinese. That should be. We should first know that eyes the case. So if we were to imagine locking him in a room and put in a slot in the wall on, then every now and then a Chinese speaker puts a message in Chinese through the slot, an attempt to get a reply from whatever is in the room. Okay, on all this in the room with John, so is a box of Chinese characters on a big instruction book, which is in English on the instructions. Tell him what characters to respond with when certain character that passed through the slot. So, for example, if you take this whole image here. So they passed this image, This message in Chinese through the slot. John Searle, loose up there reads this message. He looks at the instruction book to show him what message? What characters? We We used to respond to this message. He puts those carts together and he pushes it back out through this lot on the Chinese people. Read the response. Okay, now I suppose he gets this flawlessly. He gets it done flawlessly. Let's say he's just He's really good at reading instructions and he's really good at, you know, manipulating characters. If he does the compete, the people communicating with him, we have no choice but to say that he understands Chinese that whatever is in this room is an intelligent speaker that understands Chinese. So it passes this kind of Chinese version of the Turing test. It passes this test to determine whether or not the person in there is an intelligent Chinese beaker. Okay, so what's really happening here? So all he's doing is manipulating the characters by following an English rule book. We wouldn't say that he actually understands Chinese just because he can see the characters . You can use a book that you know the rules for the book on Deacon. Pick out the characters that would yield a response. So he's passed the Turing test without actually being able to understand Chinese. So really, we got probably this is it also another issue with the Turing test? Okay. And so says that if we replace him with the idea of a CPU in a computer, the same problem occurs. So this is the argument against Strong AI. So the first premise is that if strong, I is true. Then there is a program for Chinese such that any system running that program will be able to understand Chinese, Understand? Is the key word. Understand? Chinese John Sil could run the program. Four Chinese without actually understanding Chinese. He could do this thought experiment. You could sit in a room and the program being him, moving there and using the rule books, the instruction books, that is in English. He can do that without actually understanding Chinese. So therefore, strong eyes false. So we see what we've got here. We've got two premises and we've got the conclusion here. So it's really asking if strong I is true, then it has to understand Chinese, while the second premises will drunks oaken do it without understanding Chinese. Therefore strong, I cannot be true. So it's false. That is the logical form of the argument. We can have a look a little bit deeper, violent. Look at what's so said was the distinction between syntax and semantics. So syntax and semantics. So just sell says that computers can never truly possess strong I because they're only what he called sink Tactical. They only operate on the syntax by syntax. It just means that sort of rules for manipulating the symbols a bit like how he did it in this Chinese room thought experiment. He had a rule book, which would be seen as the syntax. And he just followed the rules to manipulate symbols a bit like grammar in real language. Whereas semantics is a little bit deeper, it's a stronger definition. You can you can understand them. You could do this in tactical understanding of words. You also understand the meaning behind the symbols on the meaning behind the world's was themselves. Andi computers lack the ability to see a symbol on understand it's semantics. It can only understand its syntax on what to do with it. Okay, so it can only really simulate Strong II, and therefore it can only ever be weak artificial intelligence. It can never be said to truly understand and truly possess strong artificial intelligence. On that is the argument that is made by John Cell. So there's description test to do. You have to go on the Internet and have a look at this under this sort of overview off the Chinese room experiment. Okay, so I'm way too. First, watch this video here, which is a video narrated by David Mitchell. Quite odd, I understand. But, you know, comedian just in the rating of philosophical thought experiment on we also, I also wanted to have a look, and I think of our Is there any ways we can critique this there any immediate floors with John Searle's argument? Or do you think that is a sound argument that doesn't that proves and completely disproves artificial intelligence completely on that? We should stop talking about it. Those are the two things that I want you to look at before the next lesson, which is really what is wrong with the Chinese room argument 6. Critiques of the Chinese Room Argument: So now we have a look at the Chinese room argument. What we're gonna do in this lesson is have a look at three main criticisms off the Chinese room argument. So we're gonna have a look a little bit against it by looking at the system's response. The idea that the Chinese room access a system Andi Seoul is misunderstanding, really the idea of computers in general. We have a look at the current intuition argument on Gravel. Look of the robot argument on what's really good about these arguments is that so has been very engaged in the debate on the Chinese room argument ever since he published his paper. So he's been able to offer some responses to some of these arguments on and the question as to whether or not those responses are very, very valid or very strong responses is a matter for another day, but warned us to discuss the responses as well as the arguments. Okay, so first of all the systems reply. So some suggest that soul is not accurately representing the idea of a computer correctly, so claims that the post in the room does not understand Chinese, and this represents the idea of like a computer CPU. However, one could say that the whole system, which is the room, the book, the person and the symbols all together collectively understand Chinese. Is the system as a whole on just like the whole room? Ah, whole computer CPU graphics card ram. All of these things come together to possess strong I. This is the argument that was made by a number of people, went so came over the Chinese room argument that it isn't. It's the whole system that understands Chinese. And if the Chinese people who were having a communications with system say that it understand it, this system has done is Chinese. Then surely it passes the cheering test on. Then surely so could anything else. So could a computer and would be able to have strong AI well, so response to this he doesn't believe it's a very convincing argument. He says that suppose I memorize all of the rules and all the characters, so it's all in my head, okay, so that we can take away this idea of it being a whole system. It's all gone into my brain, and I understand it all. I still won't understand Chinese is what he said. Now this is not very convincing response. I don't believe because if you could memorize all the characters of Chinese on all the rules for putting them together and how they go together and how you use them to, you know, speak and respond. Two questions spoken to you in Chinese Doesn't that constitute an understanding of Chinese ? Isn't that how language acquisition works? The question to sell would be. What else would one have to do in order for him to believe that they actually possess an understanding of Chinese? Because if somebody knows all the characters knows what they all you know, knows everything about the characters. Andi knows all the rules for putting them together into forming sentences. What else is there that you need to know in orderto say that you required a language to say You understand Chinese. That's why this argument isn't particularly very convincing. Some suggest that well, if you did memorize all the rules and yes, you would understand Chinese, then is the current intuition argument against the Chinese room experiment. This argues that it looks a little bit like on objection to Lady Lovelace is response to the Turing test. So just because right now there is not a computer which has the ability to understand symbol semantically. Remember in the last video, we looked at this distinction between syntax and semantics just because there's no computer that has that ability now, this does not mean that can never be a machine that could achieve this end. So Selves. Chinese Room argument again, you know, presents choir static view of our understanding of scientific and technological development . So is suggesting that there is no way that a computer can ever have strong I because it can never have semantic understanding. But he isn't actually being isn't actually successfully proving that to be the case in his Chinese room thought experiment argument. He's just proving that a current machine doesn't have the ability to understand things semantically, but he's not saying that they can never be. There needs to be a reason why they can never be a machine that could ever possess strong I because it could never understand semantics again. Oyens done syntax because there is nothing stopping us from thinking that machine in the future could develop. We could develop machines that did actually have the ability to have strong I. Then there's the robot argument. The robot argument ever got Osama right here. So the robot argument is a little bit similar to the systems reply the systems argument and is that the person in the room may not understand Chinese in this context. However, if we were able to a blowed the rule book and all the characters into a robot, surely this robot would be able to understand Chinese on This is very similar to if we go back. This is very similar to the idea that if sold, was to memorize it'll, surely he would understand Chinese. If we put it all into a robot, just like if so was toe put it all into his brain, would they be able to then understand Chinese? And then again, this begs the question that sold Have to answer what parts off understanding and language acquisition. Are we missing here? What part of it do we not get by uploading every all the characters and all the rules for how they all interact with each other? So says that this misses the point between the syntactic and semantic distinction he says. Just because it's in a robot, that doesn't mean it goes from being a syntactic understanding toe, a semantic understanding. So what we're saying here is nothing has changed by uploading into a robot. All that's changed is that it's all coming out of one space, and that's a robot. What's not changing is the robots syntactic understanding of the language. Andi there therefore, developing into the semantic understanding. This is a very good response because in reality there isn't anything that is, is doing that. But then again, a mother counter argument would be it. Does there necessarily have to be something that go from syntactic to semantic? Is there any examples in the real world off something going from an a syntactic understanding toe, a semantic understanding? What really did these things mean when we look at them in a little bit more detail? How do we really understand what syntax and semantics actually are? So there are a lot of things that really should be questioned with the Chinese room argument. And really, the question is, could a computer ever be able to think on? This is what I want to discuss in the end of this lesson. Can we ever have a computer which thinks OK on this goes to the into the current intuition argument that we've just looked out? Is there anything that is stopping us from saying that computer will be able to think in the future? That is the question that I'm posing to you as we look at the response to the Chinese room argument in the final lesson. We're gonna have a look at all we in a computer simulation. It's a very interesting topic. It's ah, there's there's a lot of there's a lot of questions that don't really get answered because there a lot of questions that can't really be answered in that dog in that lesson, which will be the next lesson and the final lesson. 7. Are we living in a computer simulation?: So in this final lesson, we're gonna ask ourselves a very interesting question. All he living in a simulation in computer simulation. There's a very deep philosophical issue, so specifically, we're gonna have a little on introduction assets of why we're even asking this question. If this is, of course, about the philosophy of artificial intelligence, what's a simulation got to do with anything? Look at the sort of logical format of the simulation argument why it makes sense logically . And then we're gonna, you know, have a look if weaken our way, able to prove that we're living in a simulation, so is an introduction. What's the simulation hypothesis got to do with the philosophy of artificial intelligence? Well, if we're if the simulation hypothesis is true, then we must be living in a machine. We must be part of a computer. So the question could be all we part are we? Computers are We are parts of a computer, you know, entities, linings of code within a simulation. Is that what we are? And they're also different ways we can approach this argument. First we look a little four experiment. So if you're not convinced by the idea that we are living in the Sims. If we're part of this quite scary looking world, part of this thought experiment, let's have a look. A really, why it has quite a few followers and why it's quite convincing. So, first of all, as humans, we are able to create simulations. Okay, that's true. We could create Simms here. You've got Minecraft you Noel. These things are simulations. It's also true that with technological improvements, we've been able to improve the immersion of simulations. So the first ever Sims game didn't have the kind of graphics that this Sims game had on. It keeps on improving and improving, improving as technological improvements continue. Now let's assume that technological developments will continue, and there's no reason to suggest that they won't. There's no reason to suggest that which is going to stop, you know, understanding technology better. And that's also assumed that these improvements that will continue will also yield to improving the immersion Zoff thes simulations. So they're simulations get better and better and again, there's no reason why we should say that they that's not gonna happen, because if you look at the salt history of gaming games from the 19 sixties and 19 seventies. Compared to games today, the graphics and the frame rates and the growth the size of the world that you can explore , it keeps on getting better and better. Now we can imagine that a species with an infinitely more advanced technology than us infinitely more events could create near perfectly immersive simulations where the people in the simulations have free will. Now we can assume that that's true. If there's gonna be a perfectly advanced, you know, almost perfectly immersive simulation that could exist, it's gonna be the case that perfectly, you know, an infinitely advanced civilization will have created it. Now let's assume also that the characters in the simulation have free will because we want to make immersive, and if they don't have free will, then it's not a very immersive simulation. So it was the most least at least we say have the perception of free will because there's a lot of debate as to whether or not we have free will. But at least we have that we sort of the perception of free will. So that's what we need to make an immersive simulation. Well, if these characters in this simulation have free will, and they believe that they're real. Then the same thing will happen. In that simulation, they will create a simulation which is also immersive, and you could see whether you can see whether the chain is going to start to form every single time. Ace, an immersive free, will based simulation, is created. The characters within that simulation will also have the desire to create an immersive, free will based simulation on so on and so on and so on and so on until it's almost infinite amount of simulations. So now let's we've got this almost infinite chain off off different simulations, one inside the other. The question is, the question we should ask is if this continues and there's an air infinite number of simulations, what are the statistical chances that where in the original world and when, not in one of the simulations the end of the day, the answer is very, very low. What are the chances of us being in that one rial world? Compared to the almost infinite simulations? This is quite scary stuff. So if we're gonna actually add some, you know, get some actual philosophical names to this. Add it to the added to the roster of our understanding of AI, the philosopher Nick Boast Room in his paper. Are you living in a computer? Simulation? Which is a very good title, proposes that an infinitely more advanced civilization would have the technology to create a perfectly immersive simulation. And if this is the case, then it statistically is more likely that were part of a created simulation than part of a real world which has evolved over billions of years. This is because there will be more simulations, possibly infinitely more simulations than true realities. If we're taking this assumption, we're assuming there's only one true reality. And if there's an infinitely advanced civilization which is able to create technology to create an immersive simulation, then they have the ability to create more than one immersive simulations. So their arm or immersive simulation or simulated realities than there are real realities. So statistically, the chances of landing in you know, being in the rial one aura are simulated reality. It's more like that we're in a simulated reality. So statistically is the case that we are in a simulation now. What about a more pragmatic card. What about just looking around and seeing if we're in a simulation? So how do we know anything exists right now? I can say that my laptop exists because I'm using it to create this lesson on this on this video on this presentation. Weapons. When I go outside and go for a walk, I cannot see feel or hear my laptop when I've outside. If I'm far enough away, how do I know it's still existing? Could it just be the case that the second we close our eyes? Reality, you know, close is around us that the only thing that exists is what we have in our own brain. On everything else is a simulation dancers. I can't No, I can't know with my If this it is exists or if it doesn't exist on def, we're assuming that statistically were probably more likely to be in the simulation. Then really, you know, it's Z lending. More likely, that is more like to be the case that whenever I look away my laptop does it disappear. It is just simulated for when I'm so for when I see it on, this is a very, quite a scary quite, you know, mind blown cut of kind of response. What is preventing us from believing that we're in a simulation on that? Everything around it is just simulated. It's It's what's known as it's an unfolds, a fireable claim. So what I mean by this is it cannot be proven to be false. I cannot prove that this laptop disappears every time I look away from it, because physically, in order to prove something, I have to actually experience the thing that to prove it on, that's impossible to do with every time I look. If to prove it, I'll have toe, you know, remove my experience from it. So it becomes, involves viable. It means it cannot be proven to be false. It is a more humorous example that we can look at, and that's the example of record multi. So in the episode one Season four Sorry Season one episode for in the episode M Light Shamma aliens Rick Morsi and Jerry are trapped in multiple layers of a simulation, and it's their tax to try and get out the simulation. And every time they think that they have left the simulation, they realize that they're actually still in a different simulation, are more stopping us from thinking something is similar. What's interesting is more famous. Example would be the Matrix films. What's interesting is, in the case of Rick A. Morty, the reason why they knew that they were in a simulation is because the simulation was actually quite poor. There were lots of books in the simulation, just like they were books in real simulations that we understand on our computers. Onda. We could really understand that if that's the only way of trying to find out if we're in a simulation or not, to try and see almost like Boog's in reality, like not bugs in the insects but like, you know, problems with reality that don't make sense that could suggest that run a simulation. And if we're also suggesting that it's an infinitely advanced civilization that is created the simulation that is almost perfectly immersive, it becomes almost impossible toe toe work out. Whether or not we're in a simulation or not, it becomes almost effectively. It's an impossible task to prove on. This is a view that is held by our old friend Elon Musk. If you're wondering why he's on the cover of Time magazine on this presentation. This is a view that he he sort of ascribes to. Or at least he's questioned it in the past on again. If you want to have a look at what he says it about it, that's it could be a research task for you to do so. The lesson review. This will be the most fantastic, the official task that I'd give you. So watch that episode of Rick A. Morty watched The Matrix. If you can watch both of them if you if you're able to on really get an idea for so watched , um, with in the backyard, mind this lesson and think about really what? What? This what it means to be in a simulation on how it's effectively impossible to prove. I don't say thank you for watching these lessons. Thank you for watching this course. I'm gonna be doing lots and lots of other philosophy videos on philosophy courses similar to this just short Mawr interactive. More fun and less less dry, you know, less intellectual kind of philosophical inquiry because, you know, t try and market it. You know, Teoh, a larger audience. You've got to really take away the boring aspects of philosophies. That's why we look at some more interesting areas of philosophy. Andi. That's why we've looked at the philosophy off A I.