Can computers think?

[Flow4421]

Hey, so today, in ITGS there was a huge debate, I was alone, against two people, arguying whether computers could think or not. So, I think computers cannot think, as thinking is considered as a "mental act" . I gave the argument that they are programmed to obey commands, made by human beings, and this is what can even give Artificial Intelligence for video games, anyways, what do you guys think?

Edited January 23, 2015 by Flow4421

[Thrashmaster]

No, they can't think. They can attempt to simulate thought, and are getting closer and closer to more accurate simulations of thought, but hopefully no one would argue they currently have consciousness. The only way they'd ever be able to truly 'think' is if they have a sense of randomness to their thoughts, rather than following an algorithm (even one with synthetic randomness). 

[Flow4421]

Yeah, i thought so!

[Starfall]

Depends how you define thinking! Computers still only obey programmed commands and they can't make logical conclusions like humans can, but on the other hand they do their thing very effectively, and that could be compared to less developed thinking, for example way the animals think.

[Vioh]

First of all, by taking the TOK approach , I would like to ask how you would define the act of “thinking”. Many years ago, people use to think that playing chess is one of the best examples to represent the act of thinking. But you see, nowadays, computers are generally so much better than human at playing chess. So I can assure you that in terms of logical thinking, artificial intelligence has already passed us, and will continue to become better.

 

I guess many people would distinguish AI with human intelligence by saying that humans can think more ‘dynamically’ compared to computers because we have ‘free will’ in our thinking (i.e. there’s some randomness in our thinking). However, according to recent research, this is probably not the case; because neuroscience & psychology have begun to show that our decision-making system & many of our behaviours are generated by our ‘unconscious brain’. Here are some video clips about this:

https://www.youtube.com/watch?v=-i3AiOS4nCE – a BBC research on decision making & free will

https://www.youtube.com/watch?v=vcJm-y7UnLY – Leonard Mlodinow on TED, talking about how ‘unconscious’ mind influences our behaviours

 

Another thing is that there are now many robots with creativity. The robot, called Emily Howell, can actually write music just like humans; and research has shown that people can’t even distinguish between Emily’s music & music written by an actual person. Robots also have the ability to improve their skills by themselves (i.e. robot learning). These robots can look at other people to ‘imitate’ their actions; or better, they can learn some of the actions by themselves without actually looking at somebody else, in just the same way as children start learning how to walk.

 

Now, I want to discuss whether it’s theoretically possible to create a robot that is similar to a human. To start, I think that rather than asking “Can robots think like humans?”, why don’t we reverse our question and ask “Do humans think in similar way that a robot thinks?”. These 2 questions differ a great deal, because by asking the first question, we are simply avoiding to analyze our own human thoughts. In other words, rather than asking whether robots have humans’ properties, we should focus on whether humans have computers’ properties, because in that way, we’ll have to actively analyze ourselves. By asking the later question, we’ll be able to realize that humans are not that special, thus avoiding ‘arguments by incredulity’ trap.

 

So do humans think in similar way that a robot thinks? According to many, or maybe even the majority of neuroscientists, consciousness & emotions are the emergent properties of extremely-complicated interactions between neurons in our brain. So looking at the problem from this perspective, then while computers are controlled by programming algorithms, our brains are controlled by physical and chemical processes within our brain. Both are deterministic in one way or another. Hence human brains are very similar to how a CPU works. So I would say that it’s theoretically possible to simulate human thoughts on a computer system.

 

Nevertheless, this is super super and super hard to do, and we have to give AI researchers more time & funding. Even though this might not happen in the near-future, many researchers are still very optimistic, & think that they are close at creating robots with consciousness & emotions (just like Jacie in Comic Potential by Alan Ayckbourn ). I think that robots of the future will look so different from the robots nowadays.

[Guest]

No, they cant. Check out ( the Chinese room ) or something like that, It is kinda of paradox or a theory which proposes that computer or artificial intelligences are not as smart as they look like.  I think there are some youtube video explains the Chinese room theory. All the best

Edited February 4, 2015 by Guest

[Vioh]

No, they cant. Check out ( the Chinese room ) or something like that, It is kinda of paradox or a theory which proposes that computer or artificial intelligences are not as smart as they look like.  I think there are some youtube video explains the Chinese room theory. All the best

 

To be honest, I think that the Chinese-room argument was constructed by philosophers who have extremely vague knowledge about computer science; hence, this argument is extremely flawed, & lacks understandings about the field of artificial intelligence and machine learning.

 

To summarize, people who advocate the argument argues like this: No matter how well a computer can respond in Chinese, it can never understand Chinese. Now, this argument can be countered by the “Summation-Room Argument” as follow:

• Summation-Room argument begins by asking: How do we know that computers don’t understand Chinese? (very TOK-like, I know ). Then by using the analogy of addition in maths, Hector Levesque was able to demonstrate that there is NO way for a programmer to program a computer to respond in Chinese without actually letting the computer understand Chinese. This is because such an algorithm would require more particles than the entire universe. Thus, a computer can ONLY respond well in Chinese if it understands Chinese itself. You can watch more in details here (https://www.youtube.com/watch?v=p9cpU_MRrEA from 21:50)

Another argument (also given by Hector Levesque) is that John Searle seemed to forget about the computer’s hard-drives (i.e. computer memory, which is very similar to human’s memory). Remember that for a computer to actually ‘imitate’ a person responding in Chinese, the computer must have 3 following parts:

• Part 1: A huge Chinese book (probably written by a person who understands Chinese) stored in the computer’s hard-drive
• Part 2: The computer program (i.e. a set of instructions that tells how the CPU process information in the book)
• Part 3: The CPU itself (which follows the instructions & gives out the output)

--> So remember that it is the WHOLE system (including the hard-drive, the CPU, & the algorithms) that imitates a person responding in Chinese. And since the Chinese book was written by a person who understands Chinese, it’s easy to see that the computer can actually understand Chinese. This is very similar to the way a non-speaker learns how to speak Chinese. The non-speaker needs to process the information in a huge Chinese book, by learning how a sentence in Chinese can be constructed. And the CPU here is nothing more than the brain of the non-speaker himself

 

Of course, you can argue that a native speaker doesn’t need to learn Chinese through the huge Chinese book, written by somebody else. However, this wouldn’t weaken the arguments above if you think about how a native speaker learns to speak the language. Let’s assume that a Person X doesn’t understand what the word – bull**** – really means; and X doesn’t have a dictionary, or anyone for him to ask. However, suppose that X overhears a conversation between Person Y and Person Z

• Y tells Z: Dude, I got 45 points as my IB final grades
• Z replies: You bull**** !!!

Then by putting the word into this context, Person X can guess that the word – bull**** – means that someone is lying, or over-exaggerating, or just making up stuff. Thus, a native speaker can learn a language through experience & through the techniques of putting words into contexts. And this is exactly what AI scientists working in the field of Deep Learning are trying to do. Deep Learning is concerned with the study of learning algorithms that can help the computers to collect raw data on its own, analyze them, and then learn a new concept by itself. This is actually called machine learning. Of course, it’s extremely challenging to do so, but it’s possible. In fact, Microsoft & Google have put quite a lot of resources on Deep Learning projects. Some computers nowadays can actually distinguish between different concepts (e.g. a dog or a cat, etc) based on its own collected data & analysis. Who says that a computer can’t think!!! (https://www.youtube.com/watch?v=t4kyRyKyOpo)

 

Another thing that you might not have known is that we actually use the results from the study of Deep Learning in our every-day life. Wolframalpha as well as Google Search!!! Take Wolframalpha, for example. You can communicate with it using normal everyday English. In other words, it can understand the English language!!! Not only that, Wolframalpha can also compute a new knowledge in real-time based on existing knowledge, and give out the result as the answer (thus it’s not only a search tool as many people think; it can understand English as well as being able to logically come up with something new). See more details here: https://www.youtube.com/watch?v=60P7717-XOQ)

 

Sorry for this super-long post. But I happened to have been debating with one of my friends about this, so I have a lot of thoughts

 

P.S. I want put here one of my favourite quotes from The Imitation Game (2014): “Of course machines can't think as people do. A machine is different from a person. Hence, they think differently. The interesting question is, just because something thinks differently from you, does that mean it's not thinking?  Well, we allow for humans to have such divergences from one another. You like strawberries, I hate ice-skating, You cry at sad films, I am allergic to pollen. What is the point of different tastes, different preferences if not to say that our brains work differently, that we think differently? And if we can say that about one another, then why can't we say the same thing for brains built of copper wire, and steel?”

 

Don’t you feel persuaded by this argument of Alan Turing???

[Guest]

 

No, they cant. Check out ( the Chinese room ) or something like that, It is kinda of paradox or a theory which proposes that computer or artificial intelligences are not as smart as they look like.  I think there are some youtube video explains the Chinese room theory. All the best

 

To be honest, I think that the Chinese-room argument was constructed by philosophers who have extremely vague knowledge about computer science; hence, this argument is extremely flawed, & lacks understandings about the field of artificial intelligence and machine learning.

 

To summarize, people who advocate the argument argues like this: No matter how well a computer can respond in Chinese, it can never understand Chinese. Now, this argument can be countered by the “Summation-Room Argument” as follow:

• Summation-Room argument begins by asking: How do we know that computers don’t understand Chinese? (very TOK-like, I know ). Then by using the analogy of addition in maths, Hector Levesque was able to demonstrate that there is NO way for a programmer to program a computer to respond in Chinese without actually letting the computer understand Chinese. This is because such an algorithm would require more particles than the entire universe. Thus, a computer can ONLY respond well in Chinese if it understands Chinese itself. You can watch more in details here (https://www.youtube.com/watch?v=p9cpU_MRrEA from 21:50)

Another argument (also given by Hector Levesque) is that John Searle seemed to forget about the computer’s hard-drives (i.e. computer memory, which is very similar to human’s memory). Remember that for a computer to actually ‘imitate’ a person responding in Chinese, the computer must have 3 following parts:

• Part 1: A huge Chinese book (probably written by a person who understands Chinese) stored in the computer’s hard-drive
• Part 2: The computer program (i.e. a set of instructions that tells how the CPU process information in the book)
• Part 3: The CPU itself (which follows the instructions & gives out the output)

--> So remember that it is the WHOLE system (including the hard-drive, the CPU, & the algorithms) that imitates a person responding in Chinese. And since the Chinese book was written by a person who understands Chinese, it’s easy to see that the computer can actually understand Chinese. This is very similar to the way a non-speaker learns how to speak Chinese. The non-speaker needs to process the information in a huge Chinese book, by learning how a sentence in Chinese can be constructed. And the CPU here is nothing more than the brain of the non-speaker himself

 

Of course, you can argue that a native speaker doesn’t need to learn Chinese through the huge Chinese book, written by somebody else. However, this wouldn’t weaken the arguments above if you think about how a native speaker learns to speak the language. Let’s assume that a Person X doesn’t understand what the word – bull**** – really means; and X doesn’t have a dictionary, or anyone for him to ask. However, suppose that X overhears a conversation between Person Y and Person Z

• Y tells Z: Dude, I got 45 points as my IB final grades
• Z replies: You bull**** !!!

Then by putting the word into this context, Person X can guess that the word – bull**** – means that someone is lying, or over-exaggerating, or just making up stuff. Thus, a native speaker can learn a language through experience & through the techniques of putting words into contexts. And this is exactly what AI scientists working in the field of Deep Learning are trying to do. Deep Learning is concerned with the study of learning algorithms that can help the computers to collect raw data on its own, analyze them, and then learn a new concept by itself. This is actually called machine learning. Of course, it’s extremely challenging to do so, but it’s possible. In fact, Microsoft & Google have put quite a lot of resources on Deep Learning projects. Some computers nowadays can actually distinguish between different concepts (e.g. a dog or a cat, etc) based on its own collected data & analysis. Who says that a computer can’t think!!! (https://www.youtube.com/watch?v=t4kyRyKyOpo)

 

Another thing that you might not have known is that we actually use the results from the study of Deep Learning in our every-day life. Wolframalpha as well as Google Search!!! Take Wolframalpha, for example. You can communicate with it using normal everyday English. In other words, it can understand the English language!!! Not only that, Wolframalpha can also compute a new knowledge in real-time based on existing knowledge, and give out the result as the answer (thus it’s not only a search tool as many people think; it can understand English as well as being able to logically come up with something new). See more details here: https://www.youtube.com/watch?v=60P7717-XOQ)

 

Sorry for this super-long post. But I happened to have been debating with one of my friends about this, so I have a lot of thoughts

 

P.S. I want put here one of my favourite quotes from The Imitation Game (2014): “Of course machines can't think as people do. A machine is different from a person. Hence, they think differently. The interesting question is, just because something thinks differently from you, does that mean it's not thinking?  Well, we allow for humans to have such divergences from one another. You like strawberries, I hate ice-skating, You cry at sad films, I am allergic to pollen. What is the point of different tastes, different preferences if not to say that our brains work differently, that we think differently? And if we can say that about one another, then why can't we say the same thing for brains built of copper wire, and steel?”

 

Don’t you feel persuaded by this argument of Alan Turing???

 

Sorry I am not convinced.

The general idea of thinking is - the process of considering or reasoning about something , with it's OWN though and judgement. The computer can merely follow the instruction given by the programming code. If i put it metaphorically, It has no difference than someone pointing you a direction and you just simply comply, Where no individual THINKING is involved. You said the CPU is nothing more than a human brain, BUT the truth is, the computer could only deal with things that are already programmed or prepared, things that they  can follow the instruction. But when there are some UNEXPECTED scenario, they simply cant do anything, they can't invent or come up with anything new, as they can't think it on their own, which give raise to ERROR. 

 

And about the chinese room, You said that the computer can only understand Chinese in order to reply in Chinese. But actually, they can't UNDERSTAND chinese, They store the information and meaning of the chinese characters ,words and phrases and so on. They store information but not giving it's own though and can only reply with the information stored . They aren't as flexible as human. For instance, As a chinese speaker myself, when chinese speak chinese, we sometimes even create our own language,phrases and words.  As we manage the language fairly well are so we can actually THINK , giving raise to thought and creation, which is unique and can't be done by artificial intelligence.  It is why when we use google translate they just purely make so many mistakes, they are stubborn as they are just nothing more than a group of instruction code.... 

 

 

My argument is definitely not as good as yours, It is not as detailed or even with lots of mistakes . But I am just not persuaded that computer can think. Please don't get mad at me or something of that sort, we are just discussing a matter , nice knowing you anyway.

Edited February 4, 2015 by Guest

[Vioh]

The computer can merely follow the instruction given by the programming code.

 

I would object to the use of the word “merely” here. It’s true that the computers only follow the instructions. But what about when this instruction is not just an instruction for doing things, but an instruction for learning things, & for being creative to create things? In that case, computers & humans are very much alike. Let’s take your example of navigation. Suppose a person wants to go from X to Z; and the fastest way to do that is to go through Y. Now, how would a computer do this job of navigation then? Do we point the computer from X to Y to Z, and the computer just simply complies? Absolutely not! The computer can actually compute by itself which way is the shortest, and thus is able to decide which path to take by itself. So you see, the computer doesn’t blindly follow instructions; it can compute things (i.e. can think) before making any decision in just the same way as humans do. We know that computing algorithms are pretty common in software system nowadays. However, learning algorithms as well as ‘creativity’-algorithms are also being deployed in the field of AI-technology, as I have mentioned in my previous post.

 

But when there are some UNEXPECTED scenario, they simply cant do anything, they can't invent or come up with anything new, as they can't think it on their own, which give raise to ERROR. 

 

It’s certainly a valid point to say humans might be better than computers at being ‘flexible’. However, what if you feed the computer with the code that tells the computer how to be flexible itself? We can actually do this by giving the computer the instruction to make certain estimations & approximations in its process of analysis. An example of this would be the ‘handwriting-recognition’ algorithm, or the music-recognition algorithm (that can even recognize your humming). In both of these 2 cases, the input (handwriting & humming) is usually very crappy; thus the recognition algorithms are designed in such a way that allows the computer to make certain approximations, so that the computer can get the job done.

 

Btw, don’t you think that humans also make as many (or maybe even more) errors as computers do? I’ve heard some psychology & neuroscience studies that showed that humans are less rational than what we used to think. 

 

They store the information and meaning of the chinese characters ,words and phrases and so on.

 

As I have already discussed at length above, computer doesn’t really reply directly with the raw information stored in the hard-drive. It must first have to process this stored information, arrange it into something that makes sense and finally yields the output. This kinda resembles the act of thinking itself. Humans also do the same. We process the information (in this case, the information is the meanings of Chinese words & phrases) by using the grammar (the rules, or more specifically, instructions) to put these words into a sentence that makes sense. So you gotta remember that the computers actually have to process data, not just simply reply directly with raw information stored in the hard-drives.

 

For instance, As a chinese speaker myself, when chinese speak chinese, we sometimes even create our own language,phrases and words. 

 

Well, I think it’s quite easy to imagine a computer program that can create a new word in English, Chinese or any other languages, by using conventional language-rules & the alphabet. However, the main point is that nobody would even use the word that the computer has created. Why? Well, because language is a matter of ‘convention’ (i.e. popularity). I mean, if I create an English word by myself, nobody would even want to use it. However, the word “facebook” became part of the English dictionary only after facebook.com became popular.

 

It is why when we use google translate they just purely make so many mistakes

 

I think you are a bit confused here. Google translate tool (the one that is currently available nowadays) cannot be called as an artificial intelligence yet. It’s not advanced enough; it deploys very limited use of machine learning algorithms. In fact, I personally think that google translate is pretty stupid (even though it’s so useful to my life ). However, Google is getting there. According to Nicholas Ostler: “Even if you don't like what it [google translate] says, you can immediately make sense of what it gives you or compare it with what you know. It still needs constructive intelligence from the user. But the fact is that it is much better than it used to be and no doubt it will continue to improve.” (source: http://www.theguardian.com/technology/2010/dec/19/google-translate-computers-languages)

 

Now regarding about whether a computer can theoretically fully understands Chinese, I would appreciate if you could re-read through the 2 arguments (summation-room, & the argument that we need to take the whole computer system into account) that I’ve already mentioned in my previous post, because I think I’ve made the case pretty clear that computers can theoretically fully understand Chinese.

 

But I am just not persuaded that computer can think.

 

There are probably 4 possible reasons why you don’t feel that you’re persuaded:

1. My arguments are not persuasive enough: In that case, I’m sorry, I’ve tried my best
2. It may be the case that you lack knowledge about the current progress of computer science (especially in the field of artificial intelligence). The thing is, you cannot just look at the computers that you use every-day to decide on the possibility of whether computers can think. It is because the computers that you use are not advanced enough. And this also includes your smart phones (even though they are called ‘smart’, they are in fact still pretty dumb). Creating a computer system that is truly smart, flexible, & can think is an extraordinarily hard thing to do; however, all my arguments point to the conclusion that it’s theoretically possible. And by considering the recent fast development & some successes within the field of deep learning, machine learning, & natural language processing, I think we have no reason to suspect otherwise
3. It may be the case that you’re arguing from incredulity. It seems to me that you are referring to human as some kind of ‘special’ beings. But guess what? We are not as special as we think. Our brains are just as mechanical as a computer’s CPU. It’s true that CPU of a conventional computer is way less advanced compared to the human brain, but its basic process is just as mechanical as the brain. In fact, many computer scientists are now attempting to build a computer’s CPU based on the human’s neuron network.
4. It may be the case that you are expecting scientists to create robots that think & behave exactly as humans do. Well, this is impossible. As ‘Alan Turing’ (in this case, Cumberbatch) rightly pointed out, humans are different from machines. One is made of living cells; the other is made of plastic & steal. One uses neuron network; the other uses electric circuit. However, that doesn’t mean that computers can’t think. They just think differently. And this is exactly the main point of this thread. The question is: “Can computers think?”, and not “Can computers think exactly as a human being?”.

And sorry again for another long post. Discussing about computers really excites me

[Emilia1320]

I would also say that yes they can, but it's not (yet) as developed as human thinking. Computers don't think same way, nor they should. There are things computers can't do and humans can and vice versa. For example computers still cannot truly create new, or make conclusions based on data if there is something that's out from the expected / previously seen, but have you met the human who can solve a complete 5th degree equation on their head in a second?

Plus. Computers are not something to be afraid of, they are useful tools. Well, I don't know many mathematicians but ones I know use computers on their work. They couldn't do the work they do without computers. and the computers they use would be useless without mathematicians instructing them.

It's symbiosis rather than competition.