Open Thread
Reminders:
Post 2 due Thursday night @ midnight
Screening this Thursday at 9pm in the main lounge. Film TBA.
Use this thread to talk about the movie today, The Great Robot Race. Also, here’s the website for the 2007 Urban Grand Challenge. Any substantive comments you make in this thread about the robot race will count towards your comments.
Also,
10 Responses to 'Open Thread'
Leave a Reply
You must be logged in to post a comment.
A few points of discussion:
1) If Turing is right, then the robots in the race are NOT intelligent, because they do not behave indistinguishably from humans. Some of the behaviors are erratic and seem artificial. They are not the behaviors of nonhuman machines.
2) Searle would argue that the robots do not understand anything. I find this hard to swallow. They clearly understand how to drive, which they demonstrate by completing the race track. If they didn’t understand, they wouldn’t be able to accomplish that task. Granted, they do not understand the same things we understand. They don’t understand that it is a race, maybe, and they don’t care about winning or losing. But they do understand driving. They understand how to avoid obstacles; how to adjust their speed based on road conditions; how to turn and follow a course. What else is there to understanding than being able to perform the right actions? Furthermore, their understanding demonstrates a commitment to driving: they will keep doing what it takes to finish the course.
3) The real point in showing you this video is to provide a point of contact for discussing robots. I think that many of you deny the possibility of machine intelligence not for any principled reason, but just because you lack imagination: you can’t see how intelligent machines is possible. This video shows you the cutting edge of robotics, and what machines are capable of. It also shows you how designers, engineers, and programmers treat their own machines.
The contrast between the Red Team and the Blue team could not be more clear. The Red Team planned everything in advance, down to the minute details of speed and direction for every inch of the course. The Blue Team gave Stanley some waypoints, but explicitly left it up to the car to make decisions on its own. They treated Stanley as an intelligent decision maker, and trusted it to make the appropriate judgments.
The real question is the question of autonomy. Maybe autonomy isn’t all-or-nothing, but is on a scale. Some things are more independent or autonomous than other things. The Red Team’s robots were less independent than Stanley, but Stanley is still less independent than a human.
But if that’s the case, then maybe there are NO principled arguments against the possibility of artificial intelligence. Autonomy is not an all or nothing property, and some machines are more or less autonomous, depending on how far they are allowed to stray from their designer’s original intentions. Stanley, and in fact most of the machines in the race, end up straying pretty far from their designer’s intent, and do things that no one, not even their programmers, could have ever expected.
Is that enough for autonomy? Is that enough for intelligence?
Daniel Estrada
30 Sep 08 at 2:59 pm
I would have destroyed those robots.
…maybe this lack in the willingness to believe robots are autonomous is derived from the same belief applied to the slaves many-a year ago.
…I don’t know; I’m just talking.
Roy Bell
30 Sep 08 at 4:17 pm
“Stanley, and in fact most of the machines in the race, end up straying pretty far from their designer’s intent, and do things that no one, not even their programmers, could have ever expected.”
What, barring erratic movements most likely attributed to sensory errors, do the machines do that is unexpected? When Stanley swerved around, it was probably because the color of the road didn’t quite match up with its predictions.
“What else is there to understanding than being able to perform the right actions?”
Everything. You called Searle’s Chinese Room example a “terrible argument,” but I think it holds an incredible amount of merit. The translator’s lack of understanding despite his ability to “translate” is based on the same principles employed by teachers who tell you not to copy your homework from someone else because you won’t learn anything. Bartleby, fictionally, was an incredibly proficient scrivener, but he had no understanding of the words he was transcribing.
“…maybe this lack in the willingness to believe robots are autonomous is derived from the same belief applied to the slaves many-a year ago.”
I don’t think so.
Luke Kaiser
30 Sep 08 at 10:55 pm
Frankly I think that the robotic cars in the race that “strayed from the designer’s intent” simply malfunctioned. Their errors were due to programming errors or faulty equipment. If by “strayed from designer’s intent” you mean “failing” — because all designers intended their machines to win — then that I understand. But I fail to see where you think Stanley strayed from its designer’s intent. It followed the rules of the programming, including the rules that allowed it to be adaptive and use the input data to choose an output (driving decision) and apply it.
But is input & output enough for autonomy? Not for me. The car may drive itself, but if you don’t give it a destination, it does nothing. Perhaps you feel that making a driving decision, like ‘choosing’ to avoid an obstacle in the path, is autonomy. I don’t. The car was given a programmed set of rules to follow about how to drive. It CANNOT deviate from said rules. It’s lasers detect an obstacle, so it goes around. You say “it made the decision to go around” but it didn’t! The programming rules say “see object, go around” so of course that’s what it will do! It does not have the “will” to outright ignore the rule.
To me, autonomy is in part that freedom to break the rules.
Rebecca Spizzirri
30 Sep 08 at 11:05 pm
I agree entirely with Rececca and Luke. When the car “strayed from the designer’s intent.” when exactly did it show that it learned something? Now if the car would of started playing a song with it’s horn, or gave a good speech, or simply turned around and refused to go by giving a legitimate excuse, that would have been straying from the designer’s intent. The cars all did exactly what their creaters’ intended. They all drove and attempted to follow a path that was laid out before them by GPS. If someone was able to just walk up to the car point with their finger and be like ” hey Stanley, now what I need you to do is drive through this desert and follow this path and come back here in 10 hours. Ok?” Now if the car could of done that then I might start to believe that robots are autonomous.
Austin Maske
1 Oct 08 at 1:48 pm
If I ask you to run to the store and get me some eggs, and these are the only instructions I give you, do you think it requires intelligence on your part to carry out this task? Why or why not? If you succeed in carrying out this task, do you get credit for completion of the task, or do I?
It seems that by the arguments being made in this thread, it does NOT take intelligence to complete the task, and once you’ve completed the task I get all the credit, since I was the one who assigned you to the task. Does that seem fair?
Turing’s paper is based on what he calls “fair play for machines”: namely, that we don’t hold machines to standards that we wouldn’t hold a human to. It seems in the human case, it is the person who completes the task that demonstrates intelligence (or competence, or understanding), and therefore it is you who gets credit for getting the eggs. Why should we treat the machines differently?
You might object: the machine has more instructions than just the high level rule to complete the race. Well, in a sense it does, and in a sense it doesn’t. It has 2900 GPS way points it must follow, which over a 140 mile course is about one marker every 250 feet. It would be the rough equivalent of me telling you which streets to take in order to get to the store and back. If I tell you all the streets to take, does that make your journey any less a demonstration of intelligence? Why or why not?
Daniel Estrada
1 Oct 08 at 2:11 pm
I don’t think it’s fair for you to take credit if I get the eggs, but I also don’t think that’s the same as this robot race. If a robot can use those GPS markers, along with checking the color of the road and using the lasers, to find a safe path, that’s fabulous! I would say that robot is well designed to complete its task, not intelligent.
If I can go to the store and get eggs by using what I’ve learned about the world, then yes, I’m intelligent. Do you feel that my learning and a machine’s programming are the same? In some ways they are, but to me they lack a fundamental difference. I don’t doubt that one day we’ll have “intelligent” machines, but presently I don’t feel that we do.
I think it is very hard as a human to accept the idea of intelligent machines. As I alluded to in the title of my first post (”now don’t you feel special, human?”) intelligence is one of those “unique” traits that we as humans don’t want to capitulate! Now that we understand animals use tools too, what’s left for humans? Language was ours, but I feel that computers have communication with their various programming languages, so that’s shot! What’s left?
“Well, WE’RE smarter than YOU are, computer!!” says humanity.
It sounds like a playground argument between second graders. And it had just as much merit.
Rebecca Spizzirri
1 Oct 08 at 10:22 pm
The thing about you giving me those instructions are of course getting the eggs would be no problem. I could walk into the store, go to the shelf, grab the eggs, pay for them, and bring them back to you. Or theres a possibility that I could tell you to piss off and go buy your own eggs. Computers don’t have the ability to disobey the command that they are given. When robots don’t do what they are told it is because it was an error on the part of the person giving the command not the robot. The robot does exactly what its told every time it’s given a command, unless another command was given to it that hinders its ability to do so.
With you giving me the streets I could take the exact route you gave me. There’s always a possibility though that I could forge my own path, with out any knowlede of what lies ahead of me. I could experiment, go down streets that I have no clue where they lead, of course there’s a possibility that I could get lost, or there’s a chance that I would find my own way to the store.
Lets say that you were invited to a party, and you asked if you needed to pick up anything for the host. The host simply says yeah, grab your drink of choice. You agree, then you tell me to go get the drinks from the store. I go and get them and bring them to you. When you get to the party you’re going to be the one that gets the credit from the host for getting the drinks, because you were the one that was asked to get them. In my mind I will be the one getting the credit for getting the drinks because I was the one who did the work. Machines aren’t able to take the credit themselves though because they don’t have that capability of feelings, or pride, or any emotion.
Doing the task by a human requires not so much intelligence as it does common sence. For me all you would say is go get the eggs from Wal-Mart or where ever. I would go and get them with out anyfurther instruction. If you told a machine to go get eggs. It wouldn’t be able to simply go out and get them. It would need a command to walk, find them, move its “arm” or what ever it uses to grab the eggs, a command to be able to calculate the appropriate amount of money, and pay, then will need another command to be able to find its way back home.
No I don’t give the machine credit for getting the eggs. I don’t give you credit for getting the eggs either. I would give you credit for being able to create a machine that was able to go get eggs for you.
Austin Maske
2 Oct 08 at 10:24 am
I’m making a comment here because of the video of big dog. I had been showing some of these robotic videos to my friends, and one of my close friends wrote me back saying, “the [video about big dog] frightens me a great deal. it is amazing, and i couldnt believe it when the guy kicked it and i stayed up. totally blew my mind. but it still makes me very uncomfortable….tis so bloody creepy…..” then a few paragraphs later in the e-mail he said, “aw man…..im still thinking about that creepy dog….how the heck did it stay upright?!?! Even a human would have had trouble staying up……..my mind is thoroughly blown…….”
I’ll agree with him, it is crazy how amazing it is that big dog was able to stay upright. the only way a human would have been able to stay upright would be if the human saw the kick coming, and braced himself for it. Big dog didn’t know it was coming though, and that’s why it is so amazing.
Also, the big dog beta parody is pretty hilarious.
Brad Thompson
6 Oct 08 at 2:18 am
Austin, you said “If someone was able to just walk up to the car point with their finger and be like ‘hey Stanley, now what I need you to do is drive through this desert and follow this path and come back here in 10 hours. Ok?’” As far as I can tell, this is closer to what the Stanley team did than what you think. Stanley set out on the path, and using the combination of laser mapping of terrain and the visual determination of where it was safe, he followed the path he was set on. He followed the path from waypoint to waypoint, but I am willing to bet that if you removed half of the waypoints, he still would have kept to the path through the desert. If you then again removed half of the remaining waypoints, he still would have completed the race. If you continued to remove waypoints until none existed, I bet Stanley could have completed the path. This is because he actually figured out what the path was, and didn’t just use a compass and his gps to drive from waypoint to waypoint. The Carnegie Mellon cars did just this, they followed waypoint after waypoint religiously. Stanley simply used waypoints to make sure that the path he was on was correct. He didn’t choose his path based on the next waypoint.
Joel Ferm
7 Oct 08 at 2:52 pm