Artificial Intelligence

[kunekune]

I want to pose a question to the people of this forum that is well known in philosophy and that many people give very little thought to every day. The question was first critically posed by Alan Turing in his essay "Computing Machinery and Intelligence" which queries:

Can machines think?

A machine in this instance is just by standard definition, what you would generally regard as a machine in simple terms. This includes anything that we use to perform basic actions including computers. Thinking is quite a different story, as the definition of thinking and the relation between the mind and the brain is still debated.

To compensate for this, Turing modified the question to this scenario: We have a man A, a woman B, and an interrogator C which may be of either gender. They play an Imitation Game, where C must guess the gender of A and B but both contestants are trying to make C guess incorrectly. Is it possible for a machine to replace A and successfully fool C into believing it is a woman?

So I pose the questions:

Can machines think?
What is the actual process of thinking?
Can a machine play the part of A in the Imitation Game?
Is Turing's test an adequate benchmark for the possession of a mind?

Try to think this one out; one sentence answers will be quite insufficient.

[Aphelion]

Though they function on electrical impulses...
I do not believe they think, for a machine will act upon its programming, which is written by a human. computers are coded for every scenario that the programmers could think of, which may give the illusion that the computer is "thinking".

In other words I do not believe that machines think

As for the process of thinking, I feel that it cannot be answered for since we can do it without Stimulus... Without lacking thinking ourselves we cannot compare the process... and so therefore there is no answer truly.

Yes, though most likely it will not fool another person. For example chat bots, they are built to respond via "teaching" or word association. Should it be programed to respond in certain ways via keywords, it may be possible to simulate a human... although maybe lacking coherence

Mind and thinking though related are very different... also I do not know of this test

[Sharpay]

In a way computers do think. There are programmers who use the language C, C++, Java, and more that creates codes to program the computer or game. For the Man A, Woman B, and interorgator C, it depends on who is the coder to find out which gender C is. They do that by another string of coding for the options of what you think C is.

A computer can think from getting a source of power and letting the BIOS (Basic Input Output System) start up the computer and then it lets the Microprossesor and DOS (Disk Operating System) take over. How I know about this, is because I am doing a course now and I am learning about it.

[Grapes Incarnate]

Processing information is not quite the same as thinking. Machines lack emotion, and emotion is key for anything to be considered living. The concept of thinking itself is something for people smarter than myself to decipher, if anyone at all should, on a moral basis, even attempt to explain the matter. In a sense, I view it as deciding to play divinity and define who thinks and who doesn't and what thinking is.

I've talked to bots, and I've seen people talk to bots. They're clearly not "thinking" on a human level, or for that matter, on the level of so-called "lower" animals. The girls I saw talking to bots were laughing and laughing at the obvious ineptitude on the part of the machines in carrying on realistic conversations.

[kunekune]

I want to express my opinion on the Turing Test (which is synonymous with the Imitation Game posted in the original post) and its validity. If there is anything here that does not make sense, I shall be happy to explain further. Please enjoy, I wrote this for you guys :3


This essay will discuss “Computing Machinery and Intelligence,” an essay by Alan Turing published in Mind. In this paper Turing makes a startling claim regarding artificial intelligence with plenty of reasoning to support his idea. In this paper, I will argue that A.M. Turing’s argument on the capability of computers having minds is flawed.

Turing’s paper was submitted during his time at the Victoria University of Manchester whilst working on the Manchester Mark I, one of the first computers. In context the essay and the subsequent query could not be called strange since such vast progress was happening in the area of computer science at the time. In this essay Turing asks the question “Can machines think?” and attempts to formulate criteria which a machine must meet in order to have a mind. As such he establishes the Turing Test, a hypothetical situation where two players of opposite genders A and B participate in a game where person C communicates through text and attempts to guess the gender of one player. A and B are asked to try and fool C into answering incorrectly using only text. Turing uses this test because of the difficulty in defining the terms “machine” and “thinking,” which is understandable. Thus we see a new form of the question as whether or not a computer can play the part of A and successfully fool C. He then modifies the question further to suit the limitations of computers, notably programs, storage capacity, and speed of action, producing the following: Can a computer with adequate storage, speed of action and an appropriate program (we will call it c) successfully play the part of A in the Turing Test?

The importance of such a question lies in the theory. Posing the question “Can machines think?” places emphasis on the mind and its definition. This leads to the subject of functionalism and the idea of multiple realizability, where anything can possess a mind as long as its function is the same to a human mind. We need to know if a machine can have a mind, not so much whether they can possess psychological properties, because logically without a mind one cannot have psychological properties. It matters not which answer we receive, for which ever we do get it will be groundbreaking for the philosophy of the mind and will greatly affect the thesis of functionalism.

As mentioned previously, Turing’s attempt to establish criteria for having a mind for computers is flawed in many respects, simply because the Turing Test is not an adequate benchmark for an object as complex as a mind. A standardization of this argument is provided below.

1. If X passes the Turing Test, then X has a mind.
2. Computers can pass the Turing Test.
3. Therefore, computers can have minds.

The first premise is where Turing makes his first mistake in assuming what it means to have a mind. Imitation is the main principle in his test, for if the machine is able to fool a human into believing it is speaking with a human, then it can be regarded as intelligent and in possession of a mind. However, imitation can simply be a result of reactionary programming, that is, programming constructed so that the computer simply responds to certain questions with certain answers. To put it simply, the computer may be programmed so that it responds in an opposite fashion to the questions asked, that if it takes the place of a man in the game it will respond with details of the female figure, not the male figure. Although the premise behind fooling another human is a clever idea, it fails to explain any other sort of rational thought on the part of the machine. So what if the computer can fool a human. Does this mean that the computer (supposedly now with a mind) can now write poetry, be swayed by music, or even be amused by comedy? The Turing Test fails to give sound evidence that any machine can do these activities upon passing because the test is too one-sided. This is not to say that Turing is not on to something, but rather he is focusing on a very miniscule portion of the capabilities of the mind.

Furthermore, Turing states that passing constitutes having a mind, but it is quite possible that an individual with a mind will fail at fooling the other human. Does this, by definition of the test, signify that the individual indeed does not possess a mind? This cannot be so, for it is possible that the individual simply erred in playing the game or rather the interrogator had a greater level of intuition. To mend this complication and at least give some credit to the test, information regarding the intellectual level of the person C would need to be given, otherwise the computer could be playing against any sort of individual, be they intelligent or intellectually challenged. It is possible that the computer is pitted against a gullible individual for the part of C and wins without a proper display of rational tactic. Likewise, the machine could play against a highly intelligent individual that notices the peculiarities in the machine’s reaction times, and therefore can tell that they are conversing with a machine. No matter how much rationale was used during the game, the machine would still be regarded as midnless if they fail. This flaw of lack of information in regards to the level of intellect on the part of C in the imitation game reveals the flaw in reasoning behind the Turing Test.

As the second premise states, computers can indeed pass the Turing Test. I do not believe Turing commits a fallacy in stating this. A computer can most certainly pass the Turing Test, but the question still stands of whether said computer has a mind. Turing mentions that for a computer to adequately pass the test it must have adequate storage and appropriate programming in order to mimic the human brain capacity. As mentioned previously, the computer has the possiblity of failure even with an adequate amount of rationale. Turing, although he does not state it explicitly in the text, claims to remedy this by giving the computer a similarly sized storage to the human brain (which Turing states is roughly 10⁹ rows of data) and an appropriate program of which not much detail is presented. He then proceeds to explain that with these additions the computer would equal a human player in the imitation game and does not take into consideration the method behind rational thought. How do we know that a computer would suffice for a human in the game with simply the same storage capacity? I feel this to be a major fallacy on Turing’s part because he leaves out the most important part of the machine: rationalization. That is the peak of human intellect, the very mental activity which separates humans from other species in the world. How can a machine compare if the only thing similar to the human mind it possesses is similar capacity?

Although I may treat Turing as having left out the part of rationalizing, this is not the case. I am merely illustrating his lack of information regarding the program of which he states. It is explicitly stated later in the essay of a “child program” with adaptability so as to seamlessly integrate the machine into a typical human environment, but the actual program which is the core, the part that helps the machine rationalize, is not spelled out in any matter at all. This cannot be entirely the fault of Turing since this argument was presented in the early twentieth century when computers were still new and fairly experimental. But despite this the jump to the assumption that computers would be able to rationalize upon winning the game is a fault. Thus the jump from premise two to the conclusion is not a logical one since too many holes have yet to be filled.

Numerous objections present themselves regarding Turing’s argument regarding artificial intelligence, most notably the objections from Theology, Mathematics, and Consciousness. I choose these three because they are all connected and provide adequate objections to the problem by questioning the very subjects that we have yet to fully understand. The Theological Objection presents the subject of the soul and how God has only given this gift to humans and no other. The soul in this case more closely resembles the idea of consciousness, which is expressed in The Argument From Consciousness as the factor which the machine must have in order to provide flexible answers to questions. Professor Jefferson in his piece “The Mind of Mechanical Man” makes the claim that until a machine can compose pieces of art which are not merely works of chance and feel emotions then it cannot compare to the human mind. Moreover, according to The Mathematical Objection, there are questions which the computer cannot answer and thus are limits placed on the computer that are not found in the human mind. These objections all originate with the problem of consciousness and the degree of rationalization of the machine.

Objections regarding my objections to the arguments could originate in a few ways. For instance, my claim regarding the intelligence of player C in the imitation game could be labeled as setting too high a benchmark for the intelligence of the machine. One could say that even if the person playing C was poorly educated they still possess a mind and would thereby still qualify for the test since no specification regarding the intelligence of C is given. In essence one could argue that C does not have to be smart to qualify. However, this objection would not take into account the fact that if C and the machine’s roles were switched, then there is a possibility that C could fail, and therefore not have a mind. But C is a human and by default has a mind so how could this be? It simply cannot, and therefore that objection to the claim would be inadequate.

In conclusion, it is evident that because of the inadequacy of the Turing Test and the lack of information regarding the method by which the machine is to think that Turing’s argument is flawed. I have first explained how the Turing Test is not a sufficient test for a mind by illustrating how a human can fail at the test. Then I explained how there is insufficient information regarding the method the machine is to rationalize and how many of the specifications are not enough to mimic human rationale. In the end, Turing’s argument, while being deductively valid, instigated the philosophical question of multiple realizability and helped spark the computer revolution throughout the world through advancements in computer science.

I sincerely recommend reading Turing's original piece (unfortunately I don't know where it could be; my copy came from my library and I'm currently referencing a hard copy) as it is rather enlightening and much more detailed that I could begin to be. I also hope that maybe you guys get an understanding of what this thread is supposed to be about: defining what it means to be a "machine," defining the process of "thinking," defining what the machine would need to do in order to be in possession of a mind, and whether or not you think Turing's Test is adequate.

[lloydexphere]

The computer's intelligence needs to approve. Technology still has many flaws and bugs that should be fixed.
Example: Look at my Japanese Fail Thread.

[kunekune]

Sept 11, 2009 11:31:04 GMT -5 @lloydexphere said:

The computer's intelligence needs to approve. Technology still has many flaws and bugs that should be fixed.
Example: Look at my Japanese Fail Thread.

It's painfully obvious that computers in general need to improve as well as the skill of the programmers, but this thread is more about what it takes for a machine to be considered intelligent, i.e. in possession of a mind or something comparable.

[sidewaysj]

A machine cannot be intelligent. It is only what the human programs it to be. It cannot think, as it lacks the ability to go beyond what's programmed in it. We, however, don't have that setup.

[kunekune]

What if we programmed machines with the necessary component in the human mind that allows us to rationalize? That little bit that separates us from other sentient creatures on the earth? Turing mentions near the end of his essay that a "child program" would need to be created, one that could learn from experience and apply its knowledge to other parts of thinking. To say that machines could never be made intelligent is quite a hasty generalization, as we humans learn from what others have taught us, and in essence we are products of repeated "programming" from our earliest days.

Sure, at the moment I'm not sure we can create a machine that could be defined as "intelligent," a term which has yet to be defined in a philosophical sense when considering a machine, but this is not to say that in the future we will still be in the same situation. Eventually I believe we will reach the point of having a machine that is at least more sentient than, say, an animal or even a young child.

[Grapes Incarnate]

Like mud we will. I'll say it again, machines lack true emotion and emotion is a crucial part of intelligence as opposed to being solely its opponent.

[kunekune]

Sept 15, 2009 19:45:47 GMT -5 Grapes Incarnate said:

Like mud we will. I'll say it again, machines lack true emotion and emotion is a crucial part of intelligence as opposed to being solely its opponent.

Machines as of now lack emotion, it's not entirely accurate nor is it wise to dismiss the possibility since it's success hasn't been achieved yet. There's no way to tell, but looking back on its advancements computer science has come a long way since the 1950s, and in the next 50-100 years we might create a machine that has the ability to adapt and think on its own.

So I say again, rushing into this discussion with claims of "It will never work" is making too great of a generalization. You may say that my optimism regarding their future intelligence is similar, but it is quite different. Imagine what computer scientists back when their field was first recognized as a science argued about in regards to machines. I'm confident that some claimed that many of the advancements seen today would never come to pass. The future holds the same capacity as far as this subject is concerned, and I believe that emotions will be a realizable goal for machines.

[Tony Redgrave]

Ah.. philosophical questions.

I suppose that one day, technology could progress to the point where we can play uber god and give machines emotions and the means to learn from experience and all. It will allow them to stimulate human feelings and all that was discussed above but one cannot forget that it will all be programed. Intellect comes from natural means of evolution and human beings are born with the ability to rationalize, feel emotion, and think all the same. A machine can only do what it is programmed to do and if we were to create a "smart" program that bestowed the ability to learn from experience and apply it to their own "mind," It is only because we have made it as such. You can call it what you like but in my opinion, programming is programming no matter how much it could mimic a human being.

[kunekune]

And you believe that what we learn has humans isn't synonymous to programming in a sense? We are taught what to do from day one, rationalizing is just an additional element. I know that many of the things that I do are a direct result of how I was taught as a child, how I was told to act when certain situations occur, so I, no, humans themselves are programmed beings, it's just in another medium.

[Tony Redgrave]

Sept 16, 2009 12:26:01 GMT -5 kunekune said:

And you believe that what we learn has humans isn't synonymous to programming in a sense? We are taught what to do from day one, rationalizing is just an additional element. I know that many of the things that I do are a direct result of how I was taught as a child, how I was told to act when certain situations occur, so I, no, humans themselves are programmed beings, it's just in another medium.

Its not really programming Kune. What your referring to is social conditioning. Like you said, Humans pick up on social norms and the "correct" responses to certain stimuli by observing action. Everything we experience, we soak up like a sponge and even before we learn language, we remember the past and the results of an experience along with some kind of interpretation. I don't remember my own mother telling me not to touch a stove but I do remember when I was three, I touched that piece of metal that supports pots and all when they are over the lighter and burned my finger. From that day on, I simply knew not to touch the stove.. case and point.

I dunno. I'm a firm believer in the fact that only what we do effects us and what you refer to as programming could have as little effect on a person as could be. Some people are just more affected then others. Likewise, some people are more "modified" by society while others believe what they personally think suits them.


But in other words, your mother and father and others can have little impact on you. I'm just one of the people who just formed their own views on the world -shrugs-

[kunekune]

I'm sorry, but social conditioning is just a term that we use instead of programming because the latter makes us feel like less than humans; there's nothing different about the two.

If you are involved in a funeral and a relative dies yet you feel no sadness and express it openly, then you are treated as though your reaction is unnatural and are urged to feel some sort of loss from the situation. This, much like your stove burning example, is something that you were told but simply do not remember. Now, that doesn't mean that your initial lesson of not touching the stove or learning to at least appear sad or remorseful when people die do not affect things later in life.

A group of students is giving you a hard time because you dress differently than they do. Suddenly you begin to dress the way they do because you do not want to evoke the same kind of response out of them. Or, you deny their style and you seemingly make your own and simply ignore their antics. Did you think to do that on your own, or did you see someone else stand up to their attitude and you wanted to emulate their strength and determination? Either way, you've been ingrained with a mindset that will follow you all of your life.

The only difference between a machine and a human in this sense would be the amount of choices one can make. Humans can make several choices depending on certain stimuli but a machine can only make a few. However, the human mind has a limited capacity while the machine's "mind" is virtually limitless (or at least can easily exceed the human's 10⁹ rows of "memory data"). So the concept is not so different, what matters is how we are able to transcribe the workings of the mind into code that a machine can understand.

Semantics is a major part of this discussion, as we need a proper definition of thinking and intelligence in order to proceed with the topic. I don't believe that the capacity for a thinking machine is none, in fact we may get closer than ever within my lifetime, but this discussion will be all for naught if we do not have strict definitions of what these two things are.

So I ask now, What does it mean to think? How do we define thinking? Is the mind even a part of the body?

(For the record, I think Dualism is the silliest thing I've ever heard of besides solipsism)