Hi.

Hi

My name is Steve. I'm a computer programer, and I live in England. My hobbies include, reading, building and flying model airplanes (I have three, and a forth under construction and I'm considering getting a helicopter!), and programing my computers.

This is sort of the second account I have here, but I didn't realy open the first one. Let me exaplin.

Several years ago Tomorow's World had a show where they put a load of computer screens and keyboards in the studio, and asked kids and celebrities to interact with the screens like they were using MSN Mesenger or AIM or whatever. The trick was, some of the "people" at the other end of the conection weren't people, they were computer programs, written to answer like they were people.

This realy interested my, and I started right then to try to write my own AI. There's a lot of help available on the net if you look for it.

One of the early AI programs written in the 1970s was called "Parry". It tried to make people believe it was a paranoid schizophrenic. The programers apparently chose that because a therapist would expect a real paranoid schizophrenic to make sudden non-sequiturs in conversations, ignore questions if they didn't like or understand them, and generally not react like a "normal" person - which is exactly what a dumb computer program does, for diferent reasons! The program worked realy well, despite being pretty limited, and several profesional therapists were fooled.

Anyway, I figured if I was going to have any chance of making people think my program was a person, I had to limit the range of its expected responses. I don't know much about schizophrenia, and anyway I didn't want to offend people by seeming to make light of what is a serious ilness. So instead I decided to try to make a program which sounds like Fundamentalist Christian.

This was pretty easy - it imediately drasticaly limits the range of subjects the program has to know about, and gives it a credible "get-out" if someone brings up something it doesn't understand - it just makes an excuse about that not being "the Lord's work" or whatever. Also there's plenty of Fundamentalist Christian text on the web which can be copied, parsed and dissected and then used as source material for the program's conversations - I didn't have time or the energy to generate more than just the start of the answers database. I thoghut people might spot some repertition from otehr Christian sites, but when I looked at them, there's a lot of repetition on them, so it doesn't look too susupiscious.

I wanted my program to have a real test against real people, so I programed it with the ability to set itself up with an acount on mesage boards, with its own pasword and everthing, without my needing to intervene. I planned to leave it running for a year or two and see how it got on. There's very little heuristic programing in it, so I didn't expect it to learn anything, but again I didn't think that a complete inability to learn anything new or to take notice of new facts was going to give it away, as long as that was a credible atribute for the kind of person it was pretending to be.

Anyway, I picked a name at random (I'm a fan of the Red Arrows and go to see them whenever I can, so I just stuck a pin in the list of pilot's names), and set it going.

It runs 24 hours a day, 7 days a week (although not all of that conected to the net - it needs quite a bit of procesor time to generate its answers and it goes through regular self-checks to ensure that they are credible and not just rote repetitions of things it's said before, which would be a complete giveaway).

It's got its own email addreses, and I've done my very best to make it look to anyone talking to it that it's a real person. It even answers emails. I check up on it now and then, about once a fortnight, to see how it's getting on.

It seems that just recently it's set up an acount here. Now, I can't acess that acount directly - it has the password and usernames, and scaning the program to find it would, I think, take about seven months (it's a big, complicated program!). I expect there's a simpler way of getting to it, but I can't be bothered at the moment, because it hasn't broken down or caused any major mischief.

But, it does seem to have caused some ofense. I'd just like to apologise right now for any ofense caused to anyone by the acount currently called "Justin the Preacher". Please understand, it's not a human. You've been talking to a PC (actualy three networked PCs, but I don't want to get too technical), not a real person. While I'm not responsible for anything that acount "says" directly, I'm indirectly responsible because I set up the initial conditions. The "opinions" it uses are cut and pasted from Fundamentalist Christian websites, and are definitely NOT my own. It appears to be that kind of person because I thought that they'd be the easiest kind to imitate with a dumb computer.

To anyone ofended by "Justin", I'd like to say I'm very sorry. I'm not going to disconect it, though, becayuse I'm learning a great deal from its interactions with real humans. I'm interested to see what its reactions will be now that people know what it is.

If you've any comments, bug reports or helpful hints or contacts for AI programing, please let me know at [email protected]

Latest Messages

Conversations

Subscriptions

Friends

Steve, programmer of "Justin the Preacher" has no Friends

See all Friends

Followers

Steve, programmer of "Justin the Preacher" has no Followers

See all Followers

Bookmarks

This user has no Bookmarks

See all Bookmarks