Autor Tópico: [SciAm] How I got fooled (and somewhat humiliated) by a computer  (Lida 1452 vezes)

0 Membros e 1 Visitante estão vendo este tópico.

Offline Buckaroo Banzai

  • Nível Máximo
  • *
  • Mensagens: 37.846
  • Sexo: Masculino
[SciAm] How I got fooled (and somewhat humiliated) by a computer
« Online: 24 de Outubro de 2008, 01:48:33 »
From Russia, with Love
How I got fooled (and somewhat humiliated) by a computer

By Robert Epstein

It all started with an online dating service. I was looking for a date. Like most men (we dogs), I made my initial judgment based largely on a photo. Yes, that’s shallow, and when one is online, it’s also fairly stupid because photos are all too easy to fake. But this time, I really blew it.


Google's HTML visualisation of the PDF article (downloadable), on the author's website

Excerpt of the transcription of related podcast episode, download link at the bottom:

Robert Epstein got his doctorate in psychology from Harvard University. He went on to become a researcher and author. He served as editor in chief at Psychology Today magazine. He is currently a visiting scholar at the University of California, San Diego, and a contributing editor for Scientific American Mind. I called him at his home in San Diego.

Steve: Hi Dr. Epstein. How are you doing today?

Epstein: Good, Steve. How are you?

Steve: Good. Good to talk to you again. So, we spoke on the February 14th Scientific American podcast about the perils of online dating—or at least some of the perils of online dating—and you've come up with a new peril of online dating.

Epstein: (laughs) Yes.

Steve: Tell us about the article "From Russia with Love" and, you know, give us a general outline of what happened.

Epstein: Well first of all, the subtitle is "How I was Fooled and Somewhat Humiliated by a Computer" and this is just another crazy online story, but it also has to do with what's happening with computers these days. When I was communicating with, corresponding with a woman, who I thought at first was in California, then I learned she was actually in Russia; and we had a long correspondence, but after a couple of months, I just got suspicious. There was just something wrong. First of all, there were no phone calls, and sometimes she didn't seem very responsive to what I was saying; and then a particular day came. This was this past winter, when she mentioned she had been talking to her friend about me on a walk in a park, and I was thinking, a walk in a park, you know in the middle of winter. And I looked at the weather: The area where she lived in Russia and her city and it was extremely cold, I think it was 12 or 14 degrees Fahrenheit and there was a heavy snow storm; so I got even more suspicious and I asked her about the storm and she didn't respond and I began to notice that anything specific, really specific that I ever came up with, seemed to be ignored. If I said something about Putin you know, the head of Russia there, anything I said that was very specific, she seemed to ignore, and I began to realize what was really going on.

Steve: And one of the reasons that this seemed to have worked for as long as did was because you are working under the assumption that English is not her first language and so when the messages come back to you and the language is a little choppy, in your mind there is a logical reason for it.

Epstein: Well, there are two things. Number one, sure the language was pretty bad, so if I didn't understand something absolutely, I just rationalized it and said, "Okay she doesn't speak English", but the other thing had to do with my motive. I am highly motivated here to ignore feelings, to ignore miscommunication, because I am in that predator mode, I am looking for a mate, and initially I had seen some pretty attractive photos of her, so we've got two things going. We've got my motive going and we've got her language problem, the language issue, and putting those together, I was very, very forgiving of the problems in our communication. But over time—it took me about four months—(laughs) over time, I figured out what was going on, and I used a standard method to figure out what was going on. What I did was I sent in random alphabet letters, instead of sending in English, I sent in random alphabet letters which is a trick I learnt because I used to run the Loebner Prize Competition in Artificial Intelligence which is an annual Turing Test where you try to figure out whether you are talking to a computer or a person. It's the quest for the thinking computer, and I know a lot about that. Actually I used to run the contest, and I have a big book coming out on this issue, The Quest for the Thinking Computer. I've a big book coming out on that in just a couple of months. So in fact I just sent her off a letter with random alphabet letters and she replied with yet another very nice, very sweet letter about her conversations with her mother, and I said, "Oh no!" (laughs) I had been corresponding for four months with a chatterbot," that's with a conversation of a computer program.

Steve: And that's, again, a chatterbot: c-h-a-t-t-e-r-b-o-t.

Epstein: That's right.

Steve: Yeah! And now what does it say? I mean you've covered a little of this in your description of your motives, and what does it say though, that somebody like you, whose got a lot more expertise in this area of, you know, the Turing Test and computers designed to fool human beings into thinking that they are having a conversation with another human being—what does it say that you actually got sucked into this for such a long period of time?

Epstein: I don't know. I guess it means, I am not as good as I thought I was, but also it means, that again, this is a situation where anyone could be fooled and even an expert obviously can be fooled. It says, in part the computer is getting a little more facile in conversing with people, so there are a number of things. I mean, you know it has to do with advances in computing for sure, it has to do with these special circumstances, has to do with my fuzzy thinking and my fuzzy motive, all in one big jumble. But clearly we are moving gradually toward a situation in which computers are going to become much more proficient in communicating and conversing with people. They are still not very good, but eventually they are going to be really good, and at some time we are going to cross the threshold into a whole new world.

Steve: As a psychologist, what do you think the person who is designing and operating the chatterbot—what do they get out of this whole thing?

Epstein: In this case, there are two possibilities: One is it's just another hacker, it's just another person having fun with computers and kind of keeping tabs on these conversations and getting a little thrill out it; or it could be a serious programmer, and you find both of those online. In fact, as a result of my sad experience, I've actually been now serving chatterbots, known chatterbots on the Internet. There are more than 80 of them that are known and about 20 of the 80 are serious projects in artificial intelligence. The other 60 seem to be, you know, just kind of pet projects of people kind of having fun. But there are some serious efforts underway to create serious chatterbots, a couple of them are even tied to big databases; one of them is even tied to Wikipedia. So imagine that, if you are conversing with a chatterbot, and that chatterbot is tied to a big database of information …

Steve: You'll think they are a genius.

Epstein: You could think they are a genius. You could also think they are up on current affairs, for example.

Steve: What are we looking at in the future? I mean to a certain degree, this is the same kind of question you could ask about somebody who gets one of these robotic dogs. But if you develop an actual emotional attachment to this ephemeral being who doesn't really exist, I mean, is it any less real to you?

Epstein: Oh! It's entirely real. In fact, the piece I did for Scientific American Mind a few months ago—that was on the android I met in Japan—conveyed that, I think, very strongly, and that's still my feeling about that experience that I had. I mean, I was very emotional when I met that android. She was very attractive. She just had certain kind of, you know, movements and facial expressions that were very impactful. Oh! The attachment is quite real, and as I said at some point we are going to pass the threshold wherein, you know those entities are going to be indistinguishable from humans, and at some point they are going to pass us by in some respects but, now, we are heading to a very new world, and I think its coming sooner than people think.

Steve: You know, you are raising all these science-fiction-like issues, but you make me think, it's inevitable if these bots get good enough, 100 years from now, somebody is going to bring one home to their parents.

Epstein: I don't think its going to take 100 years. I think it's going to be much sooner. In fact, in the book I have coming out, in the introduction, I actually went through a vast literature that is growing in several fields now and kind of summarized what's happening in many, many, many different areas related to artificial intelligence, and I'll tell you, you read those few pages in the book and it is either extremely exciting or extremely frightening, but basically a lot of the components that are going to be needed to create, you know, true AIs and human-like robots are being developed right now around the world by companies, by laboratories, by universities, and there is a lot of movement. There is especially—in speed of computing, we are only now a couple of years away, 2011 will probably be the year. We are couple of years away from having a computer that has reached the processing speed of the human brain, and that's just one of many, many, many developments that are in so many different areas, that are going to bring about this radically different world. We could reach it by, I would say, before the year 2030 and which means a lot of people, who are alive right now are going to be around to see what happens, and I also talk in the book about this concept of not the Internet, but the Internest. I think the Internet is actually a nest that we are building for AIs. I think we are little worker bees, little worker ants, and we are building a nest for AIs and I think that's all the Internet or the Internest is going to turn out to be as one huge worldwide nest that's not going to be controlled by us, and as I say, this is going to happen so fast, that I think most of us are going to be unprepared for what happens and frankly what will happen next, I don't even know. I don't think anyone knows.

Steve: Wow! What a creepy, fascinating thought. You remind me of a cartoon that I must have seen many, many years ago. These two robots standing on a corner and one of them says to the other, "Do you believe in man?"

Epstein: (laughs) Well, one of the final chapters in this book I have coming out, actually, it's called the Gnirut Test, which is, Gnirut is Turing spelt backwards, and sure enough that's a contest in the future that's arranged and organized by computers and they are trying to figure out whether people actually are intelligent or not.

Steve: (laughs) That sounds excellent. So, let's talk a little bit about. You have a second shorter article in the same issue of Scientific American Mind called, "Smooth Thinking about Sexuality"; this spectrum of straight and gayness.

Epstein: Yes, I am presenting a paper in a few weeks at the fiftieth anniversary meeting of what's informally called Quad S. Quad S stands for "Society for the Scientific Study of Sexuality", which is a wonderful organization, that was founded by Dr. Albert Ellis, who just passed away a couple of months ago, and I am very pleased to be able to do this there, and I am presenting a data from 18,000 people who took a test that I developed— which I developed, by the way, in part because of work I was doing with Scientific American. It kind of grew out of some writing, I was doing for the magazine, and a lot of people who have taken this test, which kind of pinpoints where you fall on what I called the sexual orientation continuum, and sure enough I found what Alfred Kinsey predicted many, many years ago; I found that there are very few people who are truly straight or gay. And in fact those terms are highly misleading terms, and I found that more than 90 percent of the people who take this kind of test end up somewhere in the middle of the sexual orientation continuum, so that in fact, you know, sexual orientation, number one, does fall in the continuum and very few of us are out at the extremes, either the gay or the straight end of the continuum.

Steve: So, according to your research, lets say, most straight people are actually just mostly straight.

Epstein: That's exactly right. In fact, I have generated many, many different kinds of graphs which break down the data in various ways, and if you just graph the frequency distribution of people who call themselves straight, that's exactly what you find is that yeah, most of them are towards the straight end of the continuum, but believe it or not, there are straight people all the way at the gay end of the continuum as well and the reverse is true. Among people who call themselves gay, yeah most of the people who are gay are toward the gay end of the continuum, but there are people who call themselves gay, who are almost at the straight end of the continuum and the same is true among people who call themselves bisexuals and so on. So sexual orientation is actually a continuous variable. It's not what people think it is, and I think we get into a lot of trouble socially because we have this crazy idea on our head that sexual orientation is an either-or thing. It absolutely is not. This is exactly what Kinsey said, more than half a century ago.

Steve: What do we actually mean in your survey by gay and straight then? What if somebody calls themselves gay, but according to your survey they are actually way over on the straight end of the spectrum? What does gay and straight actually mean here, is it behavioral? Is it attitudinal, what does it mean?

Epstein: Well the gay and straight labels are just very approximate labels for, almost for what you want society to believe you are. But they are not actually accurate labels for what you are. To describe your sexual orientation, you need two numbers basically, two values, one number saying where you are in the continuum and on my scale, the number ranges from 0, which would be pure straight, up to 13, which would be pure gay. So you need a number that says, basically, where you are kind of centered on that continuum; and then you need a second number, which is sexual orientation range, which is how much flexibility you have in expressing sexual orientation. With those two numbers, that actually tells you what your sexual orientation is, but the label does not. The label meaning straight or gay, that really doesn't tell you much at all. That tells you what you want to be or what kind of label is acceptable to you, but it doesn't really tell you about your sexual orientation.

Steve: Right, like Larry Craig says he is straight.

Epstein: Perfect example.

Steve: But some of his behavior may indicate possibly otherwise.

Epstein: Well that’s a perfect example. Because Larry Craig basically is saying, "I call myself straight. There are reasons I call myself straight. It's a word I use to kind of describe some aspect of what I am, but you know, my behavior doesn't fit that in some respect." That's exactly right, and you see the problem here: The flaw is with the label. The label, those labels are, I have come to believe that those labels are not just useless, but they are actually problematic, they are troublesome. They cause trouble as a society literally based on what I've learned. Since I've been studying this, if I had the power to just obliterate the labels and instead get people to think in these continuum terms, I would do it in a harpy[heartbeat] because I think, frankly, our society would operate more smoothly if we had more accurate information about sexual orientation, if we are more realistic in the way we talked about it and the way we thought about it, I think, we would have a much more, I would say, peaceful society.

Steve: You know, very interesting. You make me think of a speech in the film Angels in America that Al Pacino as Roy Cohn has, where he has just been diagnosed with HIV and his doctor is explaining to him that he is gay, and Al Pacino as Roy Cohn says, "I am not gay. I have sex with men, but I am not gay."

Epstein: Right, right!

Steve: Because, and then he goes into it some more and it's really quite interesting; he says, "Gay people in this society do not have any power, and I am one of the most powerful people in the country. Therefore I cannot be gay. I am a straight man who has sex with other men."

Epstein: Right! It's you know, what's absurd here is how these labels are used, and it's very unfortunate because you know you could think of this in terms of eye color versus height. You know eye color, we think of as existing in discrete categories, and it does more or less. There is some continuousness in eye color, but it's not unreasonable to think of it in discrete terms—someone has blue eyes, some one has brown eyes. Now with height, it's obviously on a continuum and we don't fool ourselves thinking otherwise. We don't automatically call half of us tall and half of us short, okay? That would be absurd. If you say, "I think from now on we should call half of us short and half of us tall," right, no one would go with that. They'd say, "No, that's crazy, it exists on a continuum." Now that's you see, that's the point I am trying to make about sexual orientation. It's exactly like height, it's not somewhat like height—it's exactly like height. There is a continuum period and it is ludicrous to try to draw [a] line and say half of us are tall or gay, half of us short or straight. It's absurd; it's wrong and, as I say, it also causes a terrible trouble in terms of misunderstandings and hatreds, and you know it's just bad for the society to have this false information running around. By the way, the paper I am presenting is with 18,000 subjects, but well over a 100,000 thousand people now have taken that test and the more data I collect, the smoother the curves gets; and not only that, I get the same curves in the United States and in other countries around the world, so what I am talking about is universal. Sexual orientation lies on a continuum like height does, period.

Steve: That's very interesting as you are getting cross-cultural replication there.

Epstein: Oh! Absolutely; and I am getting slightly different curves for female versus male. Female skew a little bit more toward the gay end of the distribution. I actually get slightly different results for people from different subcultures, but they are just slight differences. I mean, it's a fascinating thing, but no matter how you look at it, those labels—straight and gay—are problematic; they are wrong.

Steve: Very interesting stuff. Let me let you go, but first lets do a few plugs you've got. You've got the book coming out in a few months that you were talking about, about computers speaking with humans.

Epstein: Yes. That book is called, Parsing the Turing Test and it has a long subtitle about the quest for the thinking computer, and that's coming out in January with Springer.

Steve: And you also have a book out currently called The Case Against Adolescence.

Epstein: Yes and the Web site for that is the, and I did a piece for Scientific American called "The Myth of the Teen Brain" that is related to that particular book.

Steve: And briefly, what's the whole book about?

Epstein: Well that book is about the artificial extension of childhood past puberty and the enormous harm that has come from that extension, and it's about treating young people like individuals instead of just automatically assuming that because under a certain age, they are all incompetent. It's about the extraordinary competence actually of young people and the fact that we need to, we need to stop judging them by age and find ways, you know, one by one, case by case, to integrate young people into adult society, rather than trapping them in this absurd world of teen culture, which causes a lot of grief, causes a lot of, basically, mental health problems.

Steve: Puberty currently ends after the second postdoctoral fellowship.

Epstein: (laughs) So I have heard. Just among some subcultures in America, that's possibly true.

Steve: You have any thoughts about this Kid Nation show on CBS?

Epstein: Oh! I have a lot of thoughts about it. In fact, I had the vice president from CBS called me up about that show recently. I have a lot of thoughts, you know, it was just a grand idea which was getting skewered by the press and by the pundits even before it started to air, but the idea of taking a bunch of young people, giving them responsibility for running their lives and running a small panel, it's actually a sound idea. Actually I show in my book that there were a number of small communities run by young people, both in the United States and Europe, believe it or not, just less than a 100 years ago, and they were very successful. So, the concept the CBS had was sound. Unfortunately, having viewed so far the first episode, I can see why it is getting skewered. They kind of got off on some crazy tangents and you know they are actually not letting these young people run the town; and they turned into almost a kind of a game show, you know, with four different groups, with different colors and different status in society and it has turned out to be, so far, somewhat silly, but the basic concept is sound. And I am actually working with a production company in England right now, to try to do the same kind of thing in England with, again, a number of young people running a small town on their own. I have no doubt based on the data, I have collected over the last decade, that young people have enormous capabilities that are pretty much varied in modern society and that they certainly are more than capable, not only of running a town on their own, but probably doing so far better than we adults run our own towns.

Steve: (laughs) Well, that wouldn't be saying all that much.

Epstein: Exactly!

Steve: And why don't you tell you everybody, where they can listen to you on Sirius satellite radio.

Epstein: Yes! I have just now starting my third year on Sirius, my channel, 114 on Sirius, and you can get more information about the show on my Web site, which is The show is called Psyched.

Steve: Sounds good, Dr. Epstein very interesting talk. Thanks very much for your time.

Epstein: Thank you Steve, it's been great fun.

Steve: The new issue of Scientific American Mind hits the stands, October 9th. It will also be available at And check out the weekly Mind Matters Seminar blog, where psychologists and neuroscientists write about studies and trends they find exciting. That's at and Robert Epstein's Web site is simply; that's

Offline Jeanioz

  • Nível 33
  • *
  • Mensagens: 2.501
  • Sexo: Masculino
Re: [SciAm] How I got fooled (and somewhat humiliated) by a computer
« Resposta #1 Online: 24 de Outubro de 2008, 09:32:36 »
n00b... ::)


Do NOT follow this link or you will be banned from the site!