Course Navigation

 Course Home Page

Robot Lab Wiki

WeeklyReactions

Final Project

Tim Ambrogi

Larry Bomback

I just finished my paper on the RoboCup Initiative. Now I have to finish the other one summing up everyone's data from the Konane tournament. I think I'm going to do my presentation on Thursday on the RoboCup paper because it will be more interesting due to the fact that I have lots of pics and video clips, whereas the Konane one would be much more analytical and mathematical.

Totally unrelated topic. I remember on Tuesday, Deepak spoke about the notion of auto-pilot in future cars and in the military. I read an article a few days before that from Assosciated Press that was about the drone military aircraft that destroyed a car in Yemen that had a number of prominent Al Qaeda leaders in it. The attack actually occured over a month ago. Now the car wasn't being driven by autopilot, but the plane was. In fact, there wasn't even a human in the plane at all. The plane was being entirely run by a computer from take off to landing. I assume it probably received from commands from humans while in flight, but I don't know if anyone can argue against the artificial intelligence it exhibited by finding the suspected car, tracking it, and then waiting to attack it at the optimum time.

Jackie Chew

I can't say that this class turned out to be exactly what i expected. In no way am i dissapointed, i'm actually somewhat surprised that i made it through and learned some interesting stuff along the way. However, i'm not sure that this was the right class for me. While it certainly isn't necessary, i have found that it would have been extremely helpful, in so many ways, to have some knowledge of programming, or just computer science in general... if only for the purposes of feeling connected to the rest of the class. One can't help but feel a bit alienated when it seems like everyone else is a cs major.. or at least has a special interest in computer science. When i registered to take the course, i knew that the information covered would be a little over my head. None of my friends could understand why i, an econ major, would take an upper level computer science class. Basically, i wanted to try something a little different, and be in a class with people who had different interests. In any case, i feel that the class met my expectations in this area. Deepak was incredibly accomodating, and did his best to avoid alienating the "non-cs" people. For the future, however, i would advise having required pre-requisites for this class.

Catherine Chiu

Jason Coleman

I was going to talk about a cool way to optimize our Konane MinMax search, but I thought I would talk about AI in general instead, and the feasibility of AI.

Like many of you, I believe that it depends on how you define intelligence, and it seems that there are many levels or types of intelligence. For many of these, we can say that computers are indeed intelligent, and for many we can say that computers are not so intelligent. What we can say for sure I think, is that computers do exibit intelligent behavior whether or not they are entirely "intelligent". Does that make sense?

Anyway, I think we haven't even seen the tip of what can be acomplished with AI. I think that anything is possible, even recreating human intelligence (the Holy Grale of AI). The thing that is holding AI back is that the brain is a very complex thing, and if our computers are going to think like humans, they are going to be complex also. Don't let this complexity get you down though, because technology is always advancing. Having computers that can represent the neurons in a human brain (IBM's new computers) are going to help, advances in brain imaging, and along with that our understanding of how the brain works, are going to help.

Eventually computers will be able to think like humans, and because they are computers (fast efficient and unflawed) they will think better than humans. Something that I think is even more realistic though is the incorporation of computers with our existing brains/bodies. Think Cyborg. I guess that you will see implants to improve your memory, your calculation speed, and chess playing etc, before autonomous robots. We'll see people acting like computers instead of computers acting like people. The Cyborg theory kind of solves the problem of rampant killer evil robots, in that there will always be a human in control.

So I truely believe in artificial intelligence. And as advances are made and more people believe in AI, the problem of artificial consciousness will become important. If computers are intelligent, and they are thinking, creative, etc., are they conscious?

So, I propose a follow up atrificial consciousness class. I wrote a paper for my philosophy class last year about artificial consciousness. If anyone is interested email me.

Okay... I'm done. Can't wait to see those final projects.

Ian Harrison

Nick Kerr

Lately I have had a bit of resentment towards this class as a whole. My feelings are based upon any of my particular experiences in the class, but rather the experiences of Jamie. Jamie, who I thought was one of my good friends, doesn't much seem to be intersted in our friendship these days. Last week, while we were working on our final project, I made a couple of silly mistakes when we were discussing how to write a function we needed. Jamie starting talking about specific ways in which he thought he could implement a robot that could write final projects for undergraduate cs courses more effeciently and correcly than I could. To be honest, I felt a little threatened. I mean this whole time I have been "considering the AI question" just like the rest of the class, but I never thought that it would relate personally to me. It makes me feel slightly uncomfortable to think that my participation in our final project group could be worth less than the participation of a computer. Perhaps I feel even more jealous after hearing jamie's suggestion than i would have if he suggested that he would rather work with a different human partner. At least then I could rationalize the situation more easily and not feel offended. I mean how can you rationalize that a stupid box that is filled with motherboards and cdroms and other dopey stuff like that could be worth more to a good friend than I am.

What will a friendship be when computerized people join our society? What will a sexual relationship be? Have we already enetered the era in which AI has made computer friendships a reality? It is scary to think about this question after considering the characteristics of the stereotypical computer nerd. Do I pass up real life relationships in favor of computer relationships? I know jamie does. Have computers become appealing enough that people would prefer to pursue a sexual relationship with a compuer rather than another human being? This question seems stupid and borderline inappropriate, but how do we expain the fact that the internet seems to be more than 50% porn? In all this time of considering the AI question have we not realized that the time to consider the AI question was 50 years ago, and that AI already has such a stronghold on society that we can no longer fully image what society would be without it, and thus cannot consider the AI question with extreme bias.

Ananya Misra

I have been thinking about our discussions on intelligence and artificial intelligence through the semester. We haven't ever given AI a break, have we? It has usually been considered not quite "intelligent" once you really look at it, or no match for humans, or something...with a few exceptions.

But instead of worrying about whether machines can beat humans (they can, especially in games), we might as well just make the best of them and use them beneficially. If they steal all our occupations, I'm sure we can come up with new ones...and if we don't and the human species becomes extinct, at least no one will be left to mourn it. As long as AI improves some lives without ruining others, that's reason enough to continue working on it, whether or not we ever "solve the AI problem."

The question is, has it ruined any yet? How many?

Jamie Racanelli

I hate to be the guy who has to point this out, but it is the last wiki and this class is all about intelligence and what not, so whosey bluesy...

In A.I. people often talk about whether we are trying to build machines that exhibit rational behavior or robots that mimic human behavior or what have you. It just seems to me that when we compare the things that machines can do with what humans can do, we are always so impressed by how "intelligent" humans are. But there are a lot of dumb, stupid people around who can't do anything a computer can do. Whose intelligence are we comparing these computer's ability to? Granted, any human's ability to physically perceive the world around it by itself is an amazing task that is extremelty difficult to mimick with a computer. However, I think that in certain situations where we would jump to say that a human is clearly more intelligent than a computer, there are plenty of dumb people who are more stupid than computers. I mean take Nick for example. He is a great guy and is in many ways very smart. However, he is also stupid in that he can be very closed-minded and irrational at times. Sometimes, when I try to have a conversation with him, it becomes clear that he is grasping very tightly to certain beliefs or ideas that are dear to him, even if they don't make sense in the context of the argument. That is, he will often not admit that he is wrong and not let go of his idea even if I have made it clear that I am right and he is wrong. Those ideas to which he clings are like his preprogramming. This preprogramming is very well-suited to many situations. However, at times, it just doesn't make sense. But he irrationally holds on to that programming, in the same way that a computer would. So if you replaced Nick with a logic system that was preprogrammed with the ability to realize (through inference) that it was wrong, then it would, in that situation, be smarter than Nick. My point is that many humans are preprogrammed and we try not to hold it against them. Some people are just raised differently. People can change with time though, as they are exposed to new things and let go of olds ones. Similarly, computers can be reprogrammed. So, I think we shouldn't always dismiss Artificial Intelligence just becasue it is preprogrammed, because so are humans.

Juan Ramos

It feels as if we've just started the semester, but lo and behold, it's the end. We've certainly covered a lot of ground in the course, from robots to logic and language interpretation. When I first entered the course, I thought that it would take us a lot longer to start building robots that actually did really cool behavior. I was surprised that we dived right into the subject, and this sort of had a bittersweet taste to it. Though it was cool to start building 'smart' robots since day 1, I felt surprised that a program less than fifteen lines long could perform more-than-trivial tasks. The concept of emergent behavior kind of made me think whether humans are THAT smart after all, or whether our brains are simply wired in an unknown yet simple way that produces emergent complex behaviors.

I think that one of the obstacles within AI is the fact that we're trying to create an imitation of something that we can't yet formally define. Human intelligence is not completely understood yet, and the process of thinking has not been fully defined. Though it is evident that, at the biological level, the brain 'thinks' by interpreting electrical signals, we've yet to completely grasp how the brain does these interpretations. As such, any Artificial Intelligence system could be considered truly intelligent by a group of people, whereas it could be thought of as a billion lines of complex code by another group. My opinion is that until a concrete definition of human intelligence arises (which I think will take a while), the AI problem will not have a solution.

That's not to say I think AI has not produced some really interesting results. Regardless whether our robots were really intelligent or not, it was still nice to see them navigate their way out of a maze all by themselves. The videos on software using GA's also showed that computers can come up with good solutions to problems humans might be incapable of solving (i.e. the stock programs). Working on my final project, it was nice to see the program run through a large set of rules and assertions, make inferences about these, and provide a logical explanation as to whether a query I gave it was true or false. These programs may not necessarily be intelligent, but they do produce some neat and often useful results.

Matt Rushton

Two things: First I completed my final project late last week and it seems to be working well. For the first week of picks, it managed to go 11-5 (and one loss was the Texans over Steelers while another was a last second hail mary that was really lucky), a win percentage of .6875, which indeed is better than any of the "experts" on espn.com. There is still a long way to go to say that it is a good or bad picker but everything does seem to be quite encouraging. One thing that surprised me a bit was that each time the program is run it generally picks the same teams to win time after time, which is also encouraging in that whatever is going on is fairly consistent. I have a lot more to say and could go on a while, if anyone has any questions I would be more than happy to answer them.

My second comment involves the course as a whole. Throughout the semester, particularly early on, I remember that I kept switching back and forth over the whole intelligence question. In the end I guess I'm taking the easy way out and am going to say simply that the question is one that just isn't fair. After all we don't know what exactly intelligence is. So until we understand exactly how our mind's work I don't think we will ever be able to solve the AI problem.

Tina Tan