Unique

In this section of the reading, Hofstadter continues discussing Jumbo. He details its purpose and writes about the importance of the thought processes that Jumbo was aimed to model. When I first read about Jumbo, I was initially thinking about how much effort Hofstadter went through to create a program that does something that, on the outside at least, seems like a very trivial thing—solving “insignificant word games.” However, he makes sure to remind the reader that the real purpose is much different than that.

Attempting to model these thought processes is very interesting, and that might be because I did so many of these types of puzzles when I was younger, as I am sure many other people have as well. I also suppose that not many of those people actually think about their own thought processes either during or after the solving of the problem at hand. Clearly, he is interested in things that most people do not even think to think about.

I find it interesting how he points out ideas that he is currently discussing in the reading by referring to text that the reader has just read in the very same paragraph. Occasionally he points out how he worded something in a previous sentence, which I feel helps to connect the reader to the material. It also shows how he truly does think about these kinds of things quite frequently.

Introduction to Jumbo

In this section of the book, from page 87 to 95, Hofstadter introduces the reader to the basic ideas of his Jumbo project. He talks much about anagrams and discusses the types of methods used for solving them. He focuses on the idea that these thought processes are not actually something that happens consciously, but happen under the surface, at least for people solving these problems at an expert level--not novices uses a brute-force approach.

He also discusses the issues of perception of words while reading, and how it is surprising that people do not make more mistakes in grouping letters in a given word, given that the process may be fairly complex if one actually thinks about all of the possible ways the letters of any word can be perceived.

This all reminded me of something I read a long time ago about reading, although it is not quite the same idea:

"Aoccdrnig to a rscheearch at Cmabridge Uinervtisy, it deosn't mttaer in what order the ltteers in a word are, the only iprmoatnt thing is that the frist and lsat ltteer msut be in the rghit pclae.

The rset can be a taotl mses and you can still raed it wouthit a porbelm. This is bcuseae the human mind deos not raed ervey lteter by istlef, but the word as a wlohe."

I always thought this was interesting, and I think it somehow connects to the mental processes he is discussing in the book. In fact, it almost seems to go against what he is discussing, in that it completely dismisses the idea that the mind actually even processes the groupings of letters at all.

Me, too!

This reading mainly discussed the idea of conceptual spheres. Hofstadter describes these as being formed by the central idea or theme at the core, with variations of the theme, or generalizations, making up the outer layers of the sphere. The examples he gave were very interesting. He was trying to demonstrate how it can be difficult to judge when an idea gets stretched so far that it no longer actually has anything to do with the original theme or event. The examples mainly dealt with how people react to given events and if their reactions are really appropriate. The best example, I thought, was how the FDA responded when there was a serious of deaths due to tampering with Tylenol. The FDA, at first, only decided to mandate new regulations for drug bottles, with deadlines given sooner to drugs that are most similar to Tylenol.

At first glance, this seems completely understandable, but does it actually make sense to target drugs that are simply more similar to Tylenol? Surely food or drugs not related to Tylenol could have been tampered with as well, but this is how the FDA reacted.

Apart from this, the most interesting part of the reading was his discussion of how “Me, too” fits into human tendency to generalize. Often times, when someone says “me too,” they do not actually mean they will be doing the exact same thing, as shown in many of the examples he gave. This is of course interesting because everyone probably hears someone say this at least once every day.

Mountains and Islands

In the latest section of the book that was read this weekend, Hofstadter discusses how difficult it is to represent rules realistically in a program. He mentions, as part of this, giving templates the ability to have parts of them changed--what he refers to as "fluidifying." He also talks much about the importance of perception in recognizing patterns, and walks through the process of discovering different kinds of glues that bind certain elements of a sequence together, and developing those into islands that are used to understand the relationships in a sequence.

Another thing he touches on in this reading is something he already discussed earlier, but with the examples given in this section, it becomes much clearer how important the realization is. That realization is that sequences are truly difficult to interpret when given a set of elements that is too small. His analogy of trying to guess the future personality of an infant compared to guessing the future personality of a teenager was very powerful in displaying this important concept.

The reading was interesting for me, because I used to be very fond of this type of mathematics game when I was younger, and I spent a lot of time trying to find patterns, although I suppose not to the same extent as Hofstadter did. The process of finding patterns based on successor-ship, predecessor-ship, sameness, and run-ups/run-downs was interesting because as I did these types of puzzles, I never actually thought about how I was thinking about them, but after reading this, it seems very accurate. The fact that no one ever told me how to think about these puzzles and that most people seem to solve these things in the same way is interesting to me.

Esthetics

For this reading assignment, I was most interested in how Hofstadter regretted his attempts at creating a sequence-analyzing program as part of a competition with his university students. Mainly, his reason for being upset was interesting because it marks the difference between true artificial intelligence and the illusion of artificial intelligence, as I also believe it to be.

As I see it, most artificial intelligence is truly just an illusion of genuine artificial intelligence, because the software that calculates decisions based purely on mathematical criteria is a program which does not actually exhibit the types of thought processes that a human being does when making a decision that requires any real thought. As Hofstadter walked the reader through an example of how simply knowledge could not be the only tool employed in solving problems requiring an advanced degree of intelligence, I experienced the same sort of problems that he had. Namely, this problem was one relating to esthetics. It was difficult for me, as it was for him long ago, to actually see a pattern where it should have actually been quite easy.

I thought the main idea of this was to show how heuristics we use when attempting to solve problems as humans can either help or harm us. In this example, it was difficult to discern the pattern because of how most people, I think, are used to recognizing certain types of patterns, such as the same numbers being side-by-side. However, I also believe this can work in the opposite way—some sequences are also much faster to decipher because of this ability to instantly recognize certain patterns, to the extent that they require very little knowledge of math and calculation. Because of this, I do believe that artificial intelligence systems should attempt to incorporate the same types of heuristics in their attempts to problem-solve, in order to achieve true artificial intelligence.

First Reflection

Douglas Hofstadter's breakdown of the process of analyzing a sequence was, I thought, very interesting.

First of all, I thought it did a good job of really demonstrating the thought processes he employed in solving a problem for which he knew no clear or set method of finding a solution. As expected, this seems to have involved a great deal of guessing, and even the guessing itself was examined as an interesting phenomenon--studying how the guesses may change as each successive element in the sequence was calculated.

Of course, this immediately reminded me of the first exercise in this class, being the Crypto Problem assignment. This is especially so because I was asked how I attempted to solve a crypto problem and had, at the time, realized that it was a bit difficult to explain exactly what I had tried to do in my mind. However, my method may have been fairy similar to Hofstadter's method. Mainly, it involved some guessing, along with some deeper analyzing, followed by backtracking, and then further followed by more guessing and analyzing. Of course, this basic process was repeated, in my mind and occasionally on paper, until a solution was found to all but one crypto problem assigned to us.

Probably the most interesting part of the assigned reading for me, was how one of the ideas was so similar to my strategy in solving the crypto problems. That was the strategy of reducing the sequence down into more basic sequences, or children sequences. In the case of the crypto problems, I did this, in a way. If nothing was obvious from first glance at a given problem, then my first course of action was making the data set smaller. I did this by actually writing down a new number set and new goal by working backwards from the originals.

For example, if my goal was 35, and one of the given numbers was a 5, I would reduce my goal to 30 and remove 5 from the number list. This created a smaller problem to work on, and once I solved it, I could easily get back to the original problem by using one operation and including the 5 once again. This would be, I suppose, an example of a depth-first analysis strategy, as I chose one number to remove from the set, and worked further based on the assumption that it could be solved from there. Sometimes this did not work, though, and I would choose a new number to remove from the beginning.