|
Well, I should try to summarize what I've been doing
for the three years I haven't been working.
I guess I could say I've been looking for what
I want to do next. I did run through a compilers
textbook fairly early to see what was involved
with that, and I got an idea of the general
concepts of those. That was really pretty
enlightening as a programmer, and I have
a much deep understanding of programming
in general. One of my sor of life goals
or interests has been artificial intelligence,
which has been a really hard problem,
and I don't mean for me, I mean for everybody.
So I have been taking some time to think about
it, and I've been watching things that are
being done. Shortly after I was laid off,
I went to one of the annual conferences
on AGI, on of the first ones was in Memphis
while I was at Hilton. There is a guy
in town, Stan Franklin, who works on
one of the big AGI systems. AGI,
an acronym Ben Goertzel came up meaning
artificial general intelligence,
was really what AI was intended to
be at first, but AI as a computer
science discipline has made a lot
of progress in some sense by looking
at programming techniques for particular
parts of intelligence, so we can
use computers to do some of the
parts of intelligence that we know
how to make more mechanical.
And actually, it seems kind of
like we've managed to find what
some people might considered
fairly dumb ways of handling
information that can be a
substitute for things that
take a lot of brains in people.
I went to that one conference,
and I read in on an email list
Ben has on AGI. Someone put
out a list, and there are actually
about a couple of dozen big
AGI architecture projects. Stan
an API for theirs, which you
can download, though I have
done much with it. I've also had a bit of an interest in linguistics, so another thing I looked at was computational semantics. I looked through a textbook on that I found on that in the Berkeley bookstore. One very intriguing thing I found out looking through that is that there is really a big central problem in the whole thing that people haven't figured out. In computational semantics, most of what they work with is how meanings combine together to make sentences. But they really don't have a solid theory of how a single word has a meaning. I think it's called the theory of meaning in philosophy, and I say it's a problem in philosophy because we have such a poor grasp of it, that it hasn't really gotten to science. It's kind of a thing where we don't have a clue. Sort of like heat in the phlogiston days, or more aptly, like the theory of life before cell theory and biochemistry when people could seriously talk about elan vital. Basically about where we are with talk about consciousness. Though, I guess Searle seems to think he has figured out them both. Anyway, I'm listening to Searle lectures right now, though I've gone through them before where he seemed to have some explanation. Searle has a big argument about how computers can't do thinking as a computer program, though I'm starting to think he put some limitations on computer programs thst aren't quite right. Anyway, as a consequence of seeing this, I've become interested in the problem of what it would take for a computer to understand a single word. Shortly after looking at that, I saw a article from some Stanford linguists talking about language learning that had a concept related to the theory of meaning that kind of reversed the idea. It was roughly that words don't really point to particular meanings. The problem of inducing a complete meaning for words is really too difficult, and we just learn appropriate places to use words. And I don't think this idea was from them, but it gave me the view that the problem of language understand is more like a problem of figuring out what happened that caused a person to say what he said, not that there are words that point to things and we just have to go there. Computer language semantics is very clear and precise, and I think people really kind of use that of an idea how language works, but I think it's not particularly right. Language does refer in the sense that there is always something that made the person say the sentence, but the words pointing to the thing precisely is more a matter of being something the speaker or writer is trying to do rather than something the words do magically in themselves. I guess, unfortunately I don't see any money in those kinds of things, so they really only stay at the level of interests. Or at least I haven't seen a way to get money out of them. One intriguing thing happened in there. I went to my 25th high school class reunion. That was kind of encouraging, because there were people who remembered me from having been smart in high school. There I was talking about the thing about understanding a single word. There was one guy who was nice. He tells a story about how my tutoring in math helped him when he was struggling to go from behind to being a little ahead. I think I remember it being more of a frustration thing, and a lot of times you might not really see where you're going on a problem when you start out, and you panic or something. So maybe he was having a mental block and I helped him through. I think I saw that several times in different people. Anyway, he asked me what was my worst subject and I said English. But somehow he had gotten an essay from me, and thought it was impressive. I have been thinking I was not a very good writer for a long time. I even got evidence of that when I scored in the bottom 25% on the MCAT writing section. But I've been blogging for a long time now, and I finally have started thinking about writing a book. I've worked a little on an outline, but it looks like the money has been winding down to where it would run out before i finished, maybe, so I really kind of felt like I need get some work first. Maybe I haven't been looking too hard yet, because it hasn't run out, but now it's getting there. A computer reading in a sensor that shows if a light is shining or not has semantics about the world, and this is not something a computer program on a turing test can do. So the Chinese Room, in arguing about pure syntactic programs is not talking about the really computer machines we have in the world. They already have at least some kind of semantics that turing machines do not. Maybe they can understand. Transducers. I'm not sure what he had to say about them. He doesn't say what he means by understanding, but I wouldn't be surprised if it meant some consequence of some kind of sensory or motor transduction. It may be that the only way a Chinese room could work is by giving you an English/Chinese dictionary plus some other instructions, so you will understand word of Chinese inside it. For a computer, of course, I would mean that it has the equivalent to a dictionary to whatever the computer actually understands, and it has to have some other kind of understanding in a programs of its own. This does bring up an issue of a "language of thought". I saw a claim somewhere that knowledge can't be encoded in language, and I think that's actually correct, though it's a very deep idea. Listening to Searle lecture 12 some more. He goes through the whole mood, emotion, affective character thing that I've seen in other places. Actually, I think I saw Stan talking about this relative to his Lida system. But one thing has stood out I wanted to comment on. He said you could have moods, like you're happy or anxious but they don't have any intentionality, which means they aren't about anything specific. but psychoanalysts once heard this and told him their job was to take undirected feelings and find out what they are really directed about. And it occurs to me, from something Minsky talked about in _Emotion Machine_ that you can have thoughts that you repress or censor and don't think about, but they color your emotional life, and that kind of thing is just what they are talking about here. And a little more generally, the thing about emotions is that can persist a little bit longer than just the awareness of the immediate situation that caused them. This is a valuable feature for learning as it's how emotions get tied and associated with now things--we couldn't learn at all if emotions just died off immediately after the end of the thought that first caused it. So, consciousness is sort of a perceptual, almost sensory experience. For Searle, undesrstanding is a conscious experience, so a computer, if since doesn't have that experience, doesn't have understanding and isn't intelligent. I think that's the point of the Chinese room argument. i guess I think it's a kind of chauvinism about intelligence. Argh. There's a bar puzzle at Bardog, calle something liie freedom rings, that I have just not been able to get. It's been two times now I've been down there and spent most of the evening just messing with it, but nothing. There's a video on how to solve it, and Melissa has watched it and solved it. I watched it maybe a week ago, and I thought it looked pretty straightforward, but I got in there, and just messed it up somehow. I say it looked straightforward. The puzzle involves a chain, and I did not actually visually follow how the chain gets kind of knotted then is freed. I just kind of saw what the general idea was. But playing with it, it looks like there are a lot of ways this chain can be tied into knots. Melissa says they say this is their hardest puzzle. I can believe it. But the worst thing is that Melissa has clearly lost respect for me because I haven't been able to solve it. And she gave me advice about it. Something about how you hold the thing attached to the chain and have to keep tension on it. And I just don't see what that is supposed to do, and I guess I don't really do it, but it doesn't help me. And she must think that's totally the secret, so she thinks even worse of me because I'm not listening to her and can't get it. Grr. So I'm a little frustrated by the puzzle, but I'm more frustrated by how Melissa seems to be reacting to me. And she did for a little bit try to discount it and say it was the hardest one they had. But now I think she just thinks I'm not as smart as she thought. I've been able to get all the other ones, with not too much trouble. Maybe not as easy as some people. Some people are quite a bit more visual and mechanical than me so they can actually do better. But I am at least pretty sharp, even if this spatial stuff isn't really my strongest area. There are some puzzles, ones with rigid pieces most especially, where I can see the solution in my head. It's not all that easy to manipulate these little wire mesh things in 3-d in your head, but sometimes it's possible. Things with chains, though. That's a little too touch for me, and I have to go with more of a general sense of what kind of manipulations to use. Over this, throught that. There is a knot theory book that I was hoping would give me some tools for dealing with this kind of thing, but I haven't gotten through it yet, and I'm not sure if it would really give me tools to do it. Mom took out some deer meat, which she thought was ground, so we could make some tacos. Then she saw it was fillets, kind of at the last minute. So she switched right over to making something fancy. She sauteed some mushrooms and onions, and used some onion dip that we didn't eat to make a creamy mushroom sauce, and we had it with potatoes, peas and a salad. Whipped up in a jiffy. It was quite tasty. My back is still hurting. I think it must be something with the mattress, though just generally lying down must not be so good. Seems like the mattress has a good section on the edge and it doesn't hurt to lie there, maybe. I need to do something. I'm listening to lecture 12 now in Searle's class. He's talking about mysterianism, an idea most strongly put ofr by Colin McGinn, that there is something mysterious about consciousness and we will never understand it. And Searle talks about how there used to be a theory of vitalism, that there was some mysterious something that caused life, and elan vital. And there was a big dispute, but the vitalists lost. Or at leaast, we don't think like that any more. We understand life as chemical processes. It occurs to me, though, that understanding biology was much easier to do because we had animals and plants and things to do research on, and we could cut them up, and look at them while they were alive. Unfortunately, we don't have that easy availability of working examples that we can experiment on for consciousness. Maybe we have it a little bit with difficulty, with things like fMRI, but not so super easy with biology. So it's only natural that it has been a much tougher thing to find out about. There's also the thing that the mental stuff is so much about having the experience, and not so much about looking at what is happening in the brain of the person having the experience, so that makes it harder to look at. The thing that is making consciousness really tough to study is the subjectivity, or stuff that is only available to the person having the mental experience. They call it the "hard problem of consciousness". I'm thinking it's hard largely because it's hard to have good procedures about how you would even study that kind of thing. There's not any data that would transfer from person to person. I don't know. I'm listening to them, and starting to have reactions like I don't really care about this idea of understanding that he seems to have. And I'm not so interested in the questions they seem to be arguing about. Maybe I'm not following them that well. I'm playing freecell while listening, so I'm not giving it my complete attention, but it doesn't seem interesting enough to hold my attention. So maybe I just wouldn't be a good philosopher. Having a reaction to the little futurama show where they tell the story of how Hermese saved bender when he was a baby. And there's a song they play, "little bird, little bird, fly through my window, x3, find molasses candy". I think I saw it much earlier this year too and it got me. It seems like it must be about other feelings. Maybe some stuff from my dad. But could also be other stuff from things with people. I went to shadowcon today. I was thinking I would see Trena, but didn't. I think that was the main reason I even went at all, but I didn't make much effort to see it happen. I saw Roy, but we really didn't say much. the first time I went down in the afternoon, I was tired because I hadn't really slept enough, but honestly, maybe it was boring, or maybe everyone else was tired to from their night before, it just made me feel really tired, and I went home and slept for like four hours. I went back, and there just wasn't much going on, so I left after maybe a couple and a half hours. Didn't stay for the dance. Didn't watch their little costume party contest thing. For the dance, they had a band, but from what I heard in their practicing, it was some kind of metal something that I just didn't want to hear much more of. It must have been important for humanity that we not be very monogamous. And we are not. Even having one mate at a time, which at least does happen now, isn't real monogamy. It is biologically possible to have a creature that "pair bonds for life" I think is how they would call it. That is not us. And one little psychological aspect of it that I'm thinking of. If you wanted to have that in a species--really solid monogamy-- you'd have a positive feedback loop, where if someone found you attrative, you would find them more attractive back. But from my experience, that's not what happens to people. If someone shows interest in you, shoot if you don't feel a little bit less interested in them. When I was little, I managed to really get this from the wrong side. And I guess I also have to admit, there was someone else that I unfortunately on the mean side. Yeah, this is a lose, lose situation where on one nad side you lose something you want, and there's a mean side where you don't want something, and you end up being mean to someone who could be good to you. Anyway, it's a fairly cussed part of human nature. I can think of a couple advantages to where it can be biologically helpful. It probably necessary for people to be selective. If you just got tied in because someone liked you, you'd get stuck with pretty well the first person to come along, forever, whether that was the best choice or not. And you wouldn't be able to change later, so you would lose some mating diversity. So maybe. It is a fiarly nasty part about life, though. It's not that super strong a tendency, and people do get together. But I don't think I even hear about people any more who have only been with one person. It was a big deal when religions made it a policy to encourage. One thing about monogamy--one mate for life, not one mate at a time which is the compromise we have let that concept drift to-- is that it will kind of push you to getting together young and having a lot of kids. That's something a religion would like, if it got you to do it, because it would spread that concept around to new generations. Birth control really messed that up for them. Recently, the Catholic church started putting ads in. Theu haven't been mentioning the birth control thing, though. Protestants, apparently, dropped opposing birth control, but of course they still have the monogamy stuff. I guess that's good. And one at a time does seem at less to give more guys a fair chance. Since men are the ones that would collect and restrict access to women, it's been reasonable to have the cartel thing. But I guess it's also true that women shop around. I've been listening to John Searle's class on philosophy of mind. This is a very recent one of it, from the last couple years or so. And one of his big things, the thing that his Chinese room argument disproves according to him, is a theory that the mind is a computer program run by the brain. OK. So he says he shows it isn't. I just guess by now, I'm not sure if anyone really thought that, or if anyone still does. Maybe they did. I guess it was a possibility. Doesn't really seem so convincing. Actually, it seems like fairly often, I don't buy things that Searle says. How do we know Searle understands English but not Chinese? Does that mean something to say? Maybe he doesn't understand what it means to manipulate symbols. maybe a chinese room has to work by having a chinese dictionary, in which case he will understand some chinese. Of course, my feeling about it now is as someone who has been pondering the question, how would you get a computer to understand a single word, and since I've been starting to think of some ways that might really be done, it looks like a different situation. So, we have a conventional idea that words and sentences have meanings. That there is something that they point to out in the world. Does "hello" point to something out in the world? I would say that quite often though, particular things call up particular words, and because there was an instance that invoked a word, it did mean specifically that in this case, and we learn names for things. Maybe it works more or less. But it's really much more the other way. People just trying to find the right thing to say. So it gets tied in with what people want to do. Happy Tenth Day of Christmas! Back getting better, cold getting worse. Happy Ninth Day of Christmas! Happy New Year! I guess I'm getting to this a little late. Seems like I've been switching the year over on this blog on the first, but this time it's the second. Just in a thing where I'm not going to the blog as much. Too much Facebook. The kids managed to break two glasses yesterday. Each broke one. One actually fell on carpet, and maybe hit the chair leg. Colin was saying their glasses are stronger and wouldn't break so easily. I guess it's not a good sign that he would know that. But it gave me the feeling that they just aren't that careful. And then Aaron broke one later that day. I'll admit, we have pretty cheap glass. And we have a stone floor that will easily break anything it gets. And you expect kids to be more clumsy. And I wouldn't want them to get neurotic. But seriously, knocking over glasses is pretty pitiful. Do we need to get some sippy cups? Yeah, too much Facebook. I got on another lame Facebook game. Sims Social. There's a person, Trena, who is on there, and I got on because of her. Like all those things, they are always pushing you to get other people involved. And there are some "romantic" things you have to do to make progress. It'd be nice to have someone to at least be able to pretend to do them with. Or I might just give it up. Probably will give it up. So I did post about the dream thing. DaveDaveDave was encouraging on Facebook. I really should get moving on it. But not so likely. Hurt my back again. Chopping and moving wood. And the bigger thing is that my mattress is not so good, so if there is a problem, it makes it bad. During the day, moving around it gets better. Actually, several chairs are also bad for it. I need to figure something out. And it's especially bad because I have some kind of cough, and it hurts to cough or sneeze. Trena. She's in some kind of committed relationship, and yet she was on OKCupid looking for friends. Whatever. Kind of nerdy, seemed like, but she does or teaches some kind of martial arts, Taekwondo or something, and seemed like we have some common interests, so I'm going to try reaching out. Still haven't officially met in person, but I'm sure I've seen her around at cons. Seems to be on the edge of the SCA crowd, which I also hang around. We had two people in common on Facebook already. And Shadowcon is this weekend, so I'll probably see her there. One thing was that I was hoping we could inspire each other to get out to a Krav Maga class. And she goes to Bardog. I don't know when. But apparently her SO works nights and weekends so she's free to do things then. But just friends? I don't know how much I'm especially up for that. Grace is talking about jobs in the valley. They all know people all over. She was talking about people at Google, and said something about Java stuff at IBM. I guess I need to look into stuff like that.
|