The Computational Universe

I still do not think Artificial Intelligence, in the sense that a machine will be able to think, assess a situation and make decisions on its own, is possible. I went into this course thinking the same way, and after reading Searle’s objection and Brooks’ comments, I still feel the same way. I originally thought that robots and machines could be programmed so that they could perform tasks that required little cognitive power. Brooks Genghis and the robot vacuum cleaner that we talked about did just that: they could move about the room, and in the case of the vacuum cleaner, figure out when the room was clean. These machines, I believe, represent the extent of AI: machines that [will be] are able to perform tasks that require little cognitive power. Searle’s objections to [strong] AI, especially that “There is a fundamental difference between syntax and semantics (i.e. meaning)…a computer deals with the former, not the latter,” is a very convincing belief because it argues that the machine really is only capable of reading whatever symbol is fed into the machine and is not capable of attaching meaning to it unless the programmer tells it what to do. This raises the question that machines really do not have cognitive and thinking power: they just perform tasks that have been mechanically broken down into a complex algorithm and translate that symbol into an action without knowing the meaning of what it is doing. But Brooks does raise one interesting comment: that we don’t even know what our own brain is capable of and don’t have the scientific knowledge to define 100% of our own cognitive powers, so how can we stay definitively that we cannot create a machine that does think? Personally I believe that emotions require a very high form of thinking which will very difficult if not impossible to capture on a machine. Sure, a programmer could write an algorithm that would get a computer to produce a frowny face, but does that mean that the machine is sad? How would this sadness affect the machine’s body? Loss of appetite? More sleep? More chocolate? Reject the programmer’s attempt to clean it? How would the machine be capable of expressing this sadness? By turning off—a mimicry of sulking away to its room? I personally believe, although I’ve really had very little exposure to the hi-tech world of robots, that it would not be possible to create AI so that they are on par with a typical human, who can express emotions and has the cognitive power to delve into deeper questions which require higher order thinking.

Wow, Sir Tim Berners Lee is more than brilliant. Unfortunately, that means that a computer ignoramus like me gets horridly lost when trying to understand his language. His lecture was interesting to attend and once I realised from the very first sentence that I really couldn’t follow his technical terms I decided to try and understand the larger issues that he brought up. An excellent point that he made was that the World Wide Web is a space that humans create based on applied physical and social rules. These rules create an area in which people not only email but also where information is stored and available at a moment’s notice. Because most information is for public consumption and there is no central “Big Brother” web authority, no one knows how each person will use the web: to Google themselves, to research, to shop, or to blog, etc. Sir Berners Lee’s comment that the web is based on social rules sparked an interesting social implication of the internet and the web: that people can communicate electronically—so email is replacing the telephone. Online shopping, in some cases, is replacing physically getting up, going to the mall and shopping. Because of instant everything—shopping for clothes and groceries, or communication, adults as well as children are slowly losing the inter-personal dialogue so crucial to culture enhancement. Moreover, because everything is a click away, people are leading more sedentary lives—which of course has a negative affect on their health.

Sir Berners Lee mentioned also mentioned URIs, which are anchors for documents and basically a code in a specific syntax that enables all users to work via a shared universal system. I really didn’t understand the Syntax and ontology points he made, but the general concept that I understand is that the web is a universal space (that should be editable as well as brows-able—Wikipedia, but more reliable sources). Since the World Wide Web is for public consumption, its users need to exercise more responsibility.

The Turing Machine provides a logical and exact way to do computations. In his program, Turing essentially breaks down the steps involved in the way that a human approaches computation and analyses them so that a program can do computations. Turing’s ability to analyse what humans do when they compute sums, break it down into the most precise and mechanical rules, create a finite series of instructions and apply them to a computer demonstrates a return to simplicity which is paralleled in Brooks’ idea in creating Genghis. I find this most fascinating: that a program can be inputted into a machine and that machine will output something else based on the instructions. The concept of a universal computer, a “black box,” which can be programmed to perform any possible computation enables processes to be entirely mechanised and blurs the line between data, hardware, and software.

The Turing Machine is very efficient since it reads binary code; this universal method of counting can be applied to instruct the program to do computations from an endless tape, such as doubling or tripling numbers. Although the Turing Machine seems like the perfect solution to creating universal mechanised computations and automated “intelligence,” the Turing Machine does have its limits and cannot be halted. This means that we cannot test the program to see when it will crash. Yet, because the Turing Machine can perform the essentially human function of computation, pushes the boundaries of artificial intelligence and probes at the question of what intelligence is.

Although Scribbler is a complex machine—he can be programmed to move, turn lights on, emit sounds, and simulate a coin toss—he is a rudimentary performer of tasks requiring precision. Scribbler can’t even draw a precise square: it had circles in the corner from when Scribbler changed direction. Drawing the star was more of a disaster. By the second or so line, Scribbler was slightly off-course, which meant that the star would not close. This imprecision speaks to the ease with which Scribbler’s wheels lose their calibration. Drawing the circle was probably the easiest task for Scribbler to do—his circle actually closed and did not resemble a lump. God forbid, though, Scribbler attempt a more complicated task. He’d have to be so much more precise. Precision, of course, can be bought with more money. Overall, I’m surprised by the imprecision of Scribbler’s calibration and obstacle sensors. He is farsighted and cannot detect objects that are placed too close to him. This renders Scribbler an “ok” sensor of the world around him.

Brooks deals with this problem in trying to create an economical robot which will perform household tasks. He proposes several ways to get the robot to orient itself: it could rely on a programmed map of the room, beacons installed to provide a 3D vision system, or simply “odometry” (keeping track of the distance covered). Of course, random movement is also a possibility: one would wait for the robot to (as the theory goes) cover the room. However, these solutions are neither financially feasible nor efficient.

Essentially, Brooks believes that to remain economical, we must skimp on the robot’s precision—we must be content with a robot that will barely “know” how to carry out its function (see Brooks 117). Brooks discusses robots doing dishes, ironing clothes, or setting a table. These are more complex functions which not only require “programmed” knowledge of how to mechanically do these tasks, but also require perception, skill and most importantly adaptation. True, a robot might be able to iron a shirt, but for it to process the subtle differences between textures or to recognise when a shirt has no more wrinkles requires specific cognizance and discernment.

My closing observations include the fact that at this time, economical prototypes of household robots lack the perception necessary and are not perfect enough to perform tasks which require adaptation; until a robot can be programmed to be better adept at acclimatising itself, an efficient robot will not be available for mass consumption.

Hello All,

I’m Christine, a sophomore majoring in English and pursuing a Theatre certificate. Some of my favourite playwrights/authors include: Shakespeare, Hugo, Dickens, Dostoevsky, Fitzgerald, and Arthur Miller. I love creative fiction and non-fiction writing. I love music. I love Rodgers and Hammerstein musicals. My favourite movie, without a doubt, is The Sound of Music. I like computer graphics and film, especially the programs Photoshop and Final Cut Pro. When I was a freshman in high school, my peers became obsessed with creating websites using software provided by AOL. I hopped on the bandwagon, and designed what I thought was a really cool website; I filled it with various quotations and images, especially the cartoon dolls which I made (another obsession of my peers).

Admittedly, I really don’t know that much about the concepts behind computers, etc. I’m taking the course not only to fulfill a ST requirement but also because I’m interested in learning more about the history of computers. I’m absolutely fascinated by the question of just how intelligent machines can be, the distinction between human and machine, and how close humans can come to playing God and making AI (ps, I love The Matrix—only the first one!!, since it raises so many interesting questions and concepts that combine philosophy, religion, AI, etc). Looking at the syllabus though—I’m interested in our lab on computer music…I’ve toyed around a bit with Garage Band, but the concept of making electronic music on the computer sounds really exciting. Also, what’s listed for week 10 interests me—cryptography and viruses and worms, etc. I don’t really know what to expect from this course, but I’m excited to be taking it.

Update 11 pm on 8 Feb 2006: My school laptop is a Dell (Windows) and at home I use a Macintosh