CS372: Artificial Intelligence (Fall 1998)

Week 1: Responses

Readings

Diana Applegate

I chose to focus this short response on the movie we watched in class this week, MIT's "The Thinking Machine". Not only was it extremely entertaining (a classic for sure), it also gave me a great introduction to AI. It provided me with a sense of perspective in terms of exactly when and how AI began, and allowed me to infer just how much AI has grown in such a short period of time. The film also very accurately portrayed society's feelings towards computers back in the 1960's, and this was especially of interest to me.

During the past 19 years of my life, computers have surrounded me, and have always been a big part of my daily routine. I grew up with them, and therefore never feared their existence, but considered them a very positive thing. Computers, to me, are a valuable tool which can help make life easier in some ways.

However, in the 1960's when "The Thinking Machine" was filmed, society could only identify computers and robots as part of some scary science-fiction novel or movie. Technology of this nature was feared, considered highly "dangerous", or "black magic" and even a threat to humanity. Would intelligent machines eventually take over the world?? Since society couldn't identify with these very new and imposing machines, practical applications were overlooked for quite some time. Nevertheless, people gradually overcame the "fear factor" and started putting these mysterious machines to widespread use. The analogy of mind and machine became more and more accepted, and artificial intelligence as a field of research attracted scientists from around the world.

Despite the myraid of advances that have been made in AI since the 1960's, the field remains wide-open in a sense. "Can machines think?" is still such a hotly debated question. And even if machines can think, can they think just as a human thinks? What about creativity? Consciousness? The fact that AI is such a "new" field with so many unanswered questions makes it all the more intriguing and exciting to me. I look forward to learning what the experts think about the above questions, as well as formulating my own answers to them this semester.


Jocelyn Arcari


I found it very interesting that artificial intelligence includes fields such as psychology and philosophy. Fields such as math were not surprising, but were interesting additions that make total sense when one thinks about it.
I found the idea of computational psychology interesting. The article we read explained it as, "The program should do quickly what people do quickly, should do more slowly what people have difficulty doing, and should even tend to make mistakes where people tend to make mistakes." I could not think of any examples of programs that do this already and was wondering if they exist? It seems to me that when dealing with "machines" or programs, our goals would be to try to make them error-free, rather than have them make mistakes as humans do. I do not understand why we would want to create programs that mimic humans this closely. To me, it seems unnecessary. Why not just have humans do the things humans do and create programs or robots (or whatever!) to do things better or more efficiently than humans do. I think AI researchers' awareness of the research in cognitive psychology and vice versa are important for the growth of both fields. An important idea that I learned is that the frontier of AI is constantly changing. As problems are solved in AI or programs are created which fulfill goals, these things no longer belong to the realm of AI but are now entities on their own. The question of whether machines can think is obviously an important focus. I question whether it is the same type of question as "Can machines fly" which was brought up in class and in readings. Airplanes can fly, but do so in a different way than birds do. I fully agree with the movie from MIT, "The Thinking Machine" in its proposal that humans have been programmed from day 1. But, I still don't know if this is the basis for categorizing machines as "thinking" or "non-thinking." I liked the idea of one of the professors from Harvard in the video. He said "We will produce machines that alleviate work for man that is not productive," but he does not believe we will produce machines that are creative. Although the machine "wrote" a country western play, I don't know if I could categorize this as 'creative.' I also liked the comment made by one of the gentleman that he "wouldn't want his daughter to marry a computer" anytime soon. The movie was a really good way of introducing the topic of AI (and the ideas and questions important to it) to our class.


David Costello


The old movie shown in last week's class revealed much about expectations and reality. I found it especially interesting that though computer technology has developed at an exponential rate, our comprehension of intelligence has pretty much stayed the same. Since ancient Greece, the approach toward intelligence has mostly remained a question of philosophy. Only in recent years has science developed an infantile understanding of the biology of the mind. Several method of understanding artificial intelligence, from GOFAI to neurocomputing (from ch 1 and handout), have been researched by scientists with no real breakthroughs except for specialized computers (big blue, for example). Thus, it would seam that much more analysis is needed for truly understanding cognitive science. However, in just a few decades computers themselves have developed into highly sophisticated machines. The scientists shown in the movie had predicted such expansion in computer abilities. Where they went wrong, however, was the fact that the science of the mind would keep pace with the science of the computer. And so, we now are left with highly advanced machinery which try to heuristically mimic our shallow understanding of the mind.


Sonia Dubielzig

Having taken Intro Psychology (biological and cognitive focus) last year, I think that my approach to this class will be from a "human intelligence" perspective. Certainly, throughout the reading I found myself remembering some of the material I learned in that class, and I could see the ideas and theories of cognitive psychology many times, both obvious and hidden beneath some of the words.

For example, the two approaches to AI, "symbol-processing" and "subsymbolic" really do mirror two approaches to cognitive psychology, and the same words, "top-down" and "bottom-up" are used to describe the theories of processing.

However, what really strikes me, now that I think about it, is how much more respect I will have for intelligence, both human and animal, after taking this class. The functions we take for granted--vision and recognition, learning, reason, language--are extremely complex and intelligent skills, and simply creating a machine that mastered one of those talents would "solve" the AI problem. Psychology taught me a little about how these processes work in the human body, but seeing the challenge of artificially reproducing them makes me realize how incredible they are.

Benjamin Flynn

Hello. This is my second attempt at writing a response to class, the first was lost in a tragic crash of my machine. Perhaps it disagreed with what I was saying. I hope it does not disagree to the point of crashing on this second attempt as well. We shall see.

I was not entirely clear as to the format of this response or what exactly should be responded to. I have taken the tact of discussing my own opinion on the meaning and relevance of computer/machine "thought" in light of the material we've seen and read.

It seems to me that a large part of the worry that "the common man" has with artificial intelligence has much to do with a threat to security and ethics. Could we outsource ourselves by creating beings more complicated and efficient than ourselves? What if, as in "War Games" and "2001," machine that can think come to horrific conclusions that we are virtually unable to dissuade them from? From an ethical standpoint, how should a machine that can think be treated? Does it deserve rights? If it has an emotional life, how should this be looked after? Would we be playing God? If we were, is that ok?

I believe that our own brains are highly complex computer with very sophisticated periphrials. I believe that our experience as humans is entirely contained with the brain, that the brain is utterly responsible for our perception of self and exterior, and that our conciousness is the result of some great "do - loop" thta begins at our conception and ends at our death. This is a straightforward yet faithless response to our existence, and I'm not confident I will always be of this opinion.

Thus I believe that computers do already "think" in that they perform many of the same sorts of actions as our own brains. I do not believe that any machine of today has emotions. I am not clear as towhether a limited amount of emotion would manifest itself into consciousness for, as an example. what if a machine could feel only sadness and anger without ever contentedness and peace, what would the experience of those emotions be like? "Thinking" may imply consciousness, but whose consciousness can we truly understand but our own?

I refute the claim of the Chinese Room hypotheses for supposing that mechanical symbol matching does not constitue intelligence. I would say that the fundamental units of which our brains are composed perform exactly this sort of task. It is not the individual Room that seems Intelligent, but the entire collection of rooms together.

Emily Greenfest

I hate to bring StarTrek into this class so early on in the semester but unfortunately it was the first thing that popped into my mind as I was trying to figure out what interested me about the reading. I have two StarTrek incidents I'd like to just mention -- the first being a little exchange in StarTrek IV: The Voyage Home. I have no idea of the actual lines, but I recall that the incident occurred shortly after the main characters first encountered the probe and Spock deduced that it was trying to communicate with the humback whale. Some other character (probably McCoy) asked why any one would want to talk to a whale and Spock replied something to the effect of "why is it that humans assume they are the only intelligent life the planet has produced?" The other incident I'd like to quickly introduce occurred in a StarTrek the Next Generation episode -- I have no idea of its title, but it was the one in which Data was put on trial and his status as a sentient and possible living being was questioned. It seems to me that these two scenarios sum up pretty well what much of the reading and class discussion/lecture this past week has covered: 1) what consitutes naturally occurring intelligence, and 2) is it possible to create it, and if so, how do we recognize that we created it.

The difficulties involved in answering these questions are apparent in the fact that there exist the many subsets of AI (including cognitive psychology and the like) whose goals are the definition and observation of naturally occurring intelligence, in addition to those that focus on recreating it mechanically. And, as Spock notes, too often both of those sciences are blinded by the simple assumption that intelligence is the way human's do it -- as Deepak mentioned in class: are we asking whether a machine (or another animal) can think or are we asking if it thinks in the same way that we (humans) do? Why do we consider dolphins intelligent but not beetles? Biologists may argue something to the effect if brain size and design, computer scientists may discuss neural processes, complexity issues. Some philosophers will argue that we only consider dolphins to be intelligent because we easily identify with them -- i.e. they are easy to personify and mosy people could see themselves befriending a dolphin. This lends them human characterstics ... and human like intelligence. The same is true for most primates. A beetle on the other hand ... well, most people don't see it as friend material. Similarly, a computer like IRAX from WonderWoman fame, or HAL from 2001 A Space Oddessy is more sinsister and less human like -- more computer program and less "natural intelligence" than that encased in a human like form, such as Data, or even Asimovs Norby. Could intelligence then just be a behavior we've learned to recognize in ourselves and then ocassionaly see in the behaviors of animals and machines -- i.e. the mechanical process causing intelligence may be actaully second to the process causing it (which I think is the main principle behind behavior-based studies/evolutionary-based studies).

And I believe I have started to ramble and ended up somewhere entirely different from where I originally intended to go. My apologies.

I think what I wanted to say, put simply, the reading this past week was overall just an introduction to what the various studies affiliated w/AI entail and seemed to suggest to me that the study of intelligence -- regardless of its context (i.e. replication, understanding, observation, etc.) is confused by our lack of understanding what is meant by the term. This complicates the study of AI as it is hard to determine exactly when advances are made in the science as there are loopholes in the field's foundation.


Ayishih Hakim


I really enjoyed the video and Shapiro's reading was a good addition to Nilsson's explanation of computer science.

The one thing that I think is really important to note about the video is a prediction made in the summary of the video. The prediction is that computers will bring about the second industrial revolution. Expanding on this statement it was said that computers will aid the human mind to do things it couldnt do before. No mention however was made about the fact that computers will not only help humans do what it couldnt do before but also to allow humans to become more efficient and in the process minimize human labor. This stood out to me mainly because right before the statement on a second industrial revolution, they mentioned the first revolution and made reference to the fact that machines began to replaced human labor during this period. The only explanation and the most plausible to explanation for the ommision of this very predictable advancement is to ease the anxiety about computers.

Shapiro's reading raised a point that I need some clarification on. He mentioned on the 5 th page that computer scientists were a bit disgruntled by the fact that AI was added as a subfield to computer science. Why is this?

It seems to me that AI is the basis to most of the advances in computer science.


Ada Hogan

I guess I'm surprised to learn how much AI is as much the study of the means through which we'll develop artificial intelligence as it is the search for the final goal. I hadn't considered psychology as much of the process. It seemed odd that Shapiro's point concerning physiology at the end of the chapter wasn't introduced earlier. If in fact the proteins that we are made up of are indispensable to the acquisition of knowledge, and if we are unable to replicate this in computers, wouldn't this change the goals of AI, and the jobs we hope it might accomplish? And unless we are able to whittle down the human learning process to a binary one, how will machines sucessfully mimic us? I suppose that I need to have a better understanding of cognitive science, and how 'human cognitive behavior' can be modeled as computation. In studying computational philosopy and psychology, I would like to read more on how the sequencing of genomes is influencing our understanding of how we learn; and on the other hand, how 'soft computing' works into heuristic programming, giving us the variability and unpredictable nature that we have to allow for in our own actions. After the chapter one readings I was thinking about the top-down and bottom-up approaches that help study how intelligent behavior is to be defined: could soft computing, could fuzzy logic really be used to "teach" moral values in variable situations, to teach reason, to teach what is right and wrong? And according to whom? Are we going to develop french, american, israeli, and chinese computers?

I thought the video was great- especially for the way it introduced human learning techniques, as compared to that of a computer. By concentrating on human instinct and experience, versus probability, statistics and written programs, it betrayed much of the doubt that they seemed to have in creating "intelligent" machines. Little question of human learning being purely computational there!

Peter Ingebretson


The impression I seem to keep getting from artificial intelligence is that it's just around the corner, just about to work, almost understood. What seems interesting, after listening to the thoughts of people forty years ago, and today, is that most people still think that intelligent systems will be understood or created "soon, perhaps within 10 years." Even to me, it seems this way. How can we think we are more correct than all the other very talented programmers who predicted that they would figure out intelligence before the 90s, at the latest?

The impression I got from our introduction this week is that we are supposed to believe two things. First, some of the goals of early AI developers were reached, but we don't think about them much because AI continues to be concerned with unsolved problems, not solved ones. Second, we are supposed to understand that AI pioneers did not define their goals well enough; perhaps that the goals of AI are undefinable.

I find this somewhat frustrating. I think the basic goal of artificial intelligence is difficult to define, but intuitive to understand, and although I recognize that many of the early goals of AI have been achieved, I think it is pretty plain that this most basic goal hasn't. To me it seems clear that there should be some way to encapsulate intelligence systematically; it has been done in our brains to a reasonably high degree. I think what I want to learn most in this class is why we have failed so far to figure out intelligence ourselves.



Sarah Klaum

There were several topics both in the reading and in Friday's video that I found interesting and/or thought provoking.

The first is the idea of an AI-complete task. What I find initially exciting about this is the idea that there are so many subareas which, in the event that the question/goal at hand is answered/completed, researchers will have arrived at a complete intelligent system (as Shapiro describes it). The thought that any of the subareas (be it NLP, learning, integrated systems, etc.) could be neatly solved and at the same time wrap up the AI debate is as enticing as it seems almost over-simplified.

I was also struck by the comment that one scientist at MIT (interviewed briefly on Friday's video) made regarding the plausability of intelligent computers: that until the computer produces something new, he will not accept it as being intelligent. I began to see connections between this and the issue of AI-complete tasks. What exactly would be considered new? If a system were to learn, act in an environment, or solve problems and
come to conclusions the same way a human would, would this not be
considered new? And therefore not intelligent? These were a few of the questions I had after seeing the video.

Cheryl Koester


It was rather interesting to see in the movie the other day the expectations of the past concerning artificial intelligence. I wish we could have seen a similar movie about recent advances and contemporary expectations for the future. The article we read and the first chapter seem optimistic, but a little biased on this thread. I also read a little bit from _Mind Children_ as well, and it seems similarly optimistic. I am not sure how to define intelligence in machines. I agree that it is easy to define a human as a machine, but it is necessary to distinguish naturally occurring machines from the man-made. I think I agree with one of the scientists in the film that an important aspect of intelligence is the ability to surpass programming with creativity and originality. It may be possible to program a computer to write a western screenplay, but that computer could not make the robber commit suicide because he was caught unless it had been told that this is a possible conclusion. It seems as if the machines at that time period had been given a program written much the same way as a 'choose your own adventure' book, giving it many possible choices with many different conclusions, but leaving no room for original thought.

Another aspect that I agree is important when defining intelligence in machines is the ability to learn. Learning can be both the response to conditioning as well as learning about things with no relevance to a machine's own station in life and drawing conclusions about them. This ties into creativity in that sometimes all conclusions are not obvious.

Less important, I think, is the ability of motor control and 'senses.' These may be useful for the input of information, but there are many other ways to input information that may be more suited to a machine. The world wide web seems to be the perfect environment for developing machine intelligence because it allows the scientist to concentrate more fully on the intelligence of his software, rather than on the method of input of information.


Maralee Labarge

One of the best things that AI has done for human beings is to cause us to make an honest, earnest search for what makes us tick. Whatever other sciences that have been involved in this field, AI constantly pushes the boundaries of human perception, understanding, and intelligence, not just because we want to know why, but because we want to replicate that. Also in this search, certain questions cannot be avoided: must all intelligent beings think like humans think? Must machines learn the way humans learn? Not only are we studying our own processes, we are wondering if how we operate is the only viable way available to intelligent creatures. Without question, humans are the most intelligent species on the planet and it seems we're not quite satisfied with just reproducing the "normal" way. The idea of Data from Star Trek: The Next Generation is an exciting dream, the realization of which would alter the way humans think of themselves and what it means to be human. In a way, AI scientists get to be kids again, building machines to do our work for us (who didn't want a machine that could clean the room, or a make a bed?), machines to interact with (how great would it have been to be able to build yourself a friend when you were young?), and machines to teach us more about ourselves. AI is a field of myth and mystery, postulation and conjecture. It's provided more than a few interesting and worthwhile applications, but what we really want to know, why we are investing so much time in the pursuit of this field is the wonder: can machines think? Can they think like us? Can they think in such a way that we might have to redefine the word? Can a machine do all that a human does--better? faster? And aren't we a little bit afraid of that knowledge? And as far away we are from answering those questions, or providing an example for them, some people are. It remains to be seen what will come out of AI fields in the future.

David Rothstein

The difference between "symbolic" and "subsymbolic" artificial intelligence seems to me to be related to broader issues in the overall field of computer science. With the programming experience I have had, the main stress in the programming philosophy has always been to break up an assignment into many small tasks and then write individual functions to deal with each of these tasks. In the realm of artificial intelligence, this type of programming could be considered as the creation of many different intelligent agents (the functions), each capable of dealing with a very specific problem and only that problem. If the input to that function is different from what it expects, the function fails; in this respect, the function is incapable of "learning." Because symbolic AI deals with transforming the process of knowledge-collection and decision-making into a series of symbols, it seems that symbolic AI programs would be similar to those I have described above; once the problem of intelligent behavior has been transferred to manipulation of symbols (the language of computer science), the best way to deal with it is to use traditional computer science techniques. Subsymbolic AI, which forces a computer to learn from information fed into it from the outside world, seems to require a different programming philosophy. Many different functions, none able to communicate with any of the others, do not seem to have the ability to learn and modify their behavior. Therefore, my guess is that a different programming style is necessary for this type of AI, although I am not completely sure what it would be. As an example of this difference, I started to think about the computer program depicted in the movie last Thursday that was capable of writing Westerns. As described by the programmer in that film, the computer worked along the lines of traditional computer science -- various functions were used to decide the next course of action in the script based on several input variables and some degree of randomness. However, programming a computer to write Westerns that demonstrated real originality would require huge amounts of programming under a traditional, symbolic AI approach, since all sorts of unexpected possibilities would have to be programmed in. A subsymbolic AI approach might somehow allow the different "functions" of the program to interact, producing more possible scripts without requiring extensive programming.

Frank Rusch

I thought the movie was somewhat anachronistic, but it did bring up some interesting points that (to my knowledge) have yet to be denounced. The question the Harvard professor raised about great theories being formed by "breaking the rules", while computers apparently are in essence rule-followers. This raises some interesting questions. For one, a computer that was somehow given the "free will" to break rules (as humans have), it would bge interesting to think about where the computer's free will would be cut off. For example, an intelligent robot might hurt people in the process of breaking rules. This also raises some questions about human psychology-- we can certainly conceive of breaking any number of rules, but supposedly it is our morality that restrains us. hOw could morality be implemented in a machine, or could a computer learn to theorize about its own sources of morality (e.g. god), as humans do?


Some of the movie's speculations about the future were surprisingly accurate. It hadn't occurred to me before, but semi-intelligent machines operate automatic transmission systems in our cars. Other such machines are more common than I normally think about. Of course, they don't do it because they WANT to, and they don't apparently understand what they're doing. Or do they?


As was brought up in the text, computers are just symbol manipulators. And, at some level, humans are symbol manipulators, too. Collections of electrical impulses represent our sensory perceptions-- when we sense heat, it is because a heat receptor cell is sending an electrical signal to our brains. The text states that our brains process information in parallel, while computers process information serially. Are multitasking (or multiprocssing) computers processing information in parallel? For example, a computer could be designed that, when examining "visually", has one processor to handle, for example, texture, and one to handle color. When we perceive objects, we feel as if we're taking it in as "one big picture" rather than a unit of space with varying levels of depth, color, reflectivity, etc. Are our subconscious minds as analytical as computers, while our conscious minds present a general idea with the loose ends trimmed off? Though it is theorized that humans may perceive by comparing what they see with examplars, it seems that, at some point, base-level, perhaps binary, comparisons must be made.


Although when I think of AI, I tend to leap right towards emulation of human thought, it definitely makes sense to me that one does not need to be writing novels or conversing about abstract art to be considered "behaving intelligently". For example, a squirrel behaves intelligently because it readily does what it needs to survive. A robot might be made to run around, getting nuts, etc. Though a computer programmer might not bother to add in every detail about what a squirrel actually is thinking during the process, supposing that information were known. The question about whether a robot that acts like a squirrel is less intelligent than a robot with a precisely emulated squirrel mind is still unsure to me. To think that a robot that can adequately act like a person without actually being constructed like a human is somewhat disquieting, because in seems to take away some of the realness and uniqueness of humanity. I remember reading in National Geographic about robot bees that were constructed to "dance" in the same way that real bees use to communicate to each other. The calculated movements of the robot bee successfully communicated the location of a food source to the other bees. The real bees were, sure enough, reported at the food location, whether there was actually food there or not. The robot bee was in fact capable of influencing and even deceiving other bees. A human android capable of influencing and deceiving humans would most likely be considered intelligent (not to mention evil), even though its programming may just merely tap into some sort of "dance" that humans respond to.



Edina Sarajlic

Can computers think? This seemingly simple question posed in the early stages of the development of AI still does not have a single widely accepted answer. It was very interesting to hear some opinions on computer intelligence from thirty years ago presented in the MIT video. One of the attitudes was that artifacts cannot think, because they are only capable of following human instructions, not of producing genuinely new ideas. To refute this, few AI scientists presented examples as "evidence" that human-like intelligence in computers is possible, if not already attained. While watching, I tried to place myself into the position of a viewer from thirty years ago, encountering theses questions for the first time. I found the example of computer originality very unconvincing, even to the point of aiding the opposite side. Namely, the designers claim that their program functions as a human playwright in that it follows a certain set of rules for determining reasonable behavior of characters and
rules for modifying that behavior. In their opinion, the rule following is an indication of intelligent behavior. I think that following rules is far from not enough for creating a humanly intelligent machine. In this I agree with another scientist featured on the video, who said that the machine needs to be capable of modifying the rules governing its behavior, even some of those set by humans. However, while this scientist believes this impossible, I see it as a part of the AI future. Theoretically, replicating human neural pathways with electronic
components is not the only way to achieve a generally intelligent machine akin to a human being. In fact, from what I have read so far, the task of understanding the physiology of thought processes is far from completion. One of the approaches to AI mentioned by Nillson states that the ways in which humans and animals achieve intelligent behavior might not be suitable methods for machine thinking. Designing the implementation of intelligent behavior with the priority of optimizing the machine resources, rather than just replicating natural intelligence methods. seems like a reasonable course for AI.

Ben Sprecher

For me, the readings in the book and the article didn't answer any questions for me, but they did get me more excited about the different possibilities for projects in this class. I'm now excited by the idea of possibly building a chess program that I couldn't beat, or doing something with artificial life or genetic algorithms.

I'm even more excited now than I was last week about where this course is going. In fact, Tim, Ben and I left the lab at 10 o'clock because we were having so much fun building our robot's chassis. So I'm already eagerly anticipating class tomorrow.


Emily Sweeney-Samuelson

I am very interested in the field of Artificial Intelligence and its effects on modern society. I don't yet know very much about these things, but the more I learn, the more intriguing it seems. The movie this week was a glimpse into the early days of computers, which made it worth watching anyway, but it also addressed fascinating questions about artificial intelligence from the viewpoint of that time, which was intriguing to hear. It's eerie to think of the advances we are predicting now! I'm very curious about whether an android-like robot will be developed in my lifetime, or a really excellent natural-language parser. That could be an incredibly useful tool, I think.

As for the question "Can machines think?", I have no definite opinion yet. I am inclined to define the question as one man in the movie did; he said he would not agree that machines could think until he saw one produce something really original. Many robots in science fiction are barely discernible from humans, and it is difficult for me to get past this impression of the thinking machine and define the intermediate stages of machine intelligence. My instinct, therefore, says that there are no machines today that can think, but again, I don't know enough about the field yet. I don't know about all the artificial intelligence projects that are going on, all the machines that are coming closer to intelligence. I'm certainly not closed to the possibility that someday, machines may think independently of humans, but I'm looking forward to exploring the field a little in this class, and finding out just how close we are to producing a thinking machine.


Saskia Tan

I didn't realize how many things in my life that I've heard about relate to AI. Over the past two summers, I worked at a local natural history museum. Although it was a small place, we were fortunate enough to have a replica of the land rover that was sent to Mars come to our museum for a presentation. I think it was called the Sojourner. And then this past summer, we read an article about the robots that give tours around the large natural history museum that was created by the Carnegie Mellon Institute. As museum workers, we discussed what type of environment they would be most efficient in and also how they are replacing us as workers. These robots would probaly be able to give better tours than us but the issue that comes up is that if things like this keep continuing, then soon people will interact more with computers and robots, etc than with other people.

It is interesting now to look at the same issue from a different perspective. Before watching the video, I might of said that computers can think but right now I won't. Maybe by the end of this course I will change my mind. What changed my mind is how the computer is programmed, it is given a set of directions and then it just follows them. Humans are like this too but they have the ability to change and be creative in what they do. When computers are created to follow the same patterns as humans, that doesn't mean that they can think. Again, they just follow their programming. But then there are machines like Deep Blue which is why I'm still not quite sure of my answer.

Tim Waring

my current thoughts, after having read both the first chapter in the book and the article, as well as other related books, Galatea 2.2, Kevin Kelly's Out of Control (a serious discussion of everything from AI to ALife to Emoney) are that The simple problem solving strategies, striaght forward algorythms, and even heuristic procedures are not the most promising route. I have a lot of faith in the learning capacities of neural networks, and also trust the idea building something that learns, and then having teaching it the rest of the world. I think that "Android epistemology" when the creators define an entire web full of inter-related meatings, so that the AI will know that having cereal usually means a bowl and milk, are ultimately a stumbling block. To build in everything that the AI is expected to know is dangerous because a small flaw in your entering could mean a misuderstanding of the world in which it is expected to operate. A neural net on the other hand, will learn as it experiences.

I also am not totally convinced that the symbol system hypothesis is valid. There I do manipulate symbols in my brain, but i have no evidence that all my thoughts are composed of are symbols. very many of my thoughts are very abstract, and more of a "feeling" than a distillation of meaning, a single unit of meaning. I currently am putting my chipps in the Neural Network strategy, especially those using a different processor for each A-neuron.



Sarah Waziruddin

Artificial intelligence is a recent offshoot of computer science and is comprised of a variety of different fields including cognitive psychology, philosophy and computer science itself. Artificial intelligence utilizes heuristic programming and fuzzy logic because thinking machines also utilize this sort of logic and do not follow a traditional algorithm. The main issue in AI is whether or not machines can think. This question is met with a number of reactions ranging from paranoia to enthusiasm. There is a sense of intense paranoia and panic in the film, The Thinking Machine. The film also exhibited an 'us against them' sort of mentality, grouping scientist and machine against ordinary man much like in old horror films. Among scientists, the question of machines thinking is met with many reactions. One of these reactions is Searle's. Searle takes the view that machines cannot think because they are not comprised of proteins. For Searle, being comprised of proteins is essential is order to think. Newell and Simson do not agree with Searle and believe intelligence is based on the ability to process symbols. AI works to determine what exhibits signs of intelligence and how.

Leslie Zavisca

So far AI is pretty much just what I expected and I am pleased that we are getting right down to it. I think that am really going to enjoy the labs and building our robot. The movie and readings were very interesting. I did not realize how far AI had come in its early stages, and I certainly did not realize how much of our daily lives it encompasses. Although the question 'can machines think?' at first seems silly and dramatic, I have honestly started pondering it myself during the past week, and also which of the different AI complete tasks I tend toward the most. I think this class is going to show me an entirely knew perspective on the everyday world around me. I am particularly interested in learning more about the hardware of the robots. I know more than most people about programming and software, but the hardware still stumps me.

Back to CS372 Course Materials