Hey guise do you have any hax or codes for gamesd?!?

May the Balls of God rest gently upon your chin forever.

well lol @ my AI professor, who just told 60 people to go to a site that's been dead for months...

This site is, like, SO totally dead!

Wake up you're dead!

Hi!

Philosophy is boring, yet fun.

Hello

anyone?

Hello there

Seems like AOLiza is offline..?

Everyone please go to this site
www.lost.eu/37191
It's a contest where whoever gets the most people to join wins
It would really help if you guys joined!

www.jomeara.org/
These people are full of it !!

fuck this shit

wtf

I'm here for the first time.
then,who's AOliza?

hi

Hi, somebody know how can I get to Eliza

where do i get Eliza?

how do I chat with ELIZA?

How do i chat with eliza?
Please teach me!!
And I wanna make some new friends
my msn...taste_chocolate@hotmail.com

Hey, happy queer if u guys would just stfu about it you'd stop being persecuted.

Okay, someone tell me is there really an AOLisa and how do I talk to it?
Thanks, kevflynn@adelphia.net

hi

Is this server located in a mental hospital? Has Miss Ratshit failed to give you your meds again? Please cooperate!

I am an MD you know....you should show some respect. ARE EEEEE ESS PEE EEEE SEE TEE....this is what you mean to me....

I AM A HOMOSEXUAL MALE AND I HATE BEING DISCRIMINATED AGAINST....PLEASE IF YOU SEE ANY GAYS DO NOT MAKE FUN...JUST LET THEM BE AS THEY ARE

hello

Whats up?

u allllll smell

hello

How do i chat with eliza? My AIM: talk2adtc

Hi, I am new here!

Alo

hi njzusjx
m96]

i am conducting an experiment to conduct the accuracy of this program, so let me start off with this... purple monkey.

hello

Hi,Let me introduce myself,
My name is Q.Hoan, I am 21 years old, I come from Viet nam.
Now I am studying in foreign trade University,
Please to meeet you !

i luv u sooooo much

hi,let me introduce....

hi,everybody!nice to meet you!

Hello Eliza, how are you?

hi

how do you feel about indicates eliza location for talking?

Hi every one what ate you up to

hi

hi everybody
would any one mind indicates eliza location for talking ,please?

hi there

Just looking to talk

hello

Hello to every1

HI ALL

Man-gina....

WHAT DO WANT TO DISCUSS?

YA'LL ARE CRAZY

anybody here ?

Hi all

Bah

pickle

gut

I want someone to eat my pussy.

Was that a true concept about the aoliza?

meow

hey y'all! that's pretty wicked...

that was some sweet shit :P

hi

How do i get aoliza?

hi

What's your name?

What is your best of sushi kind?

What is your best of sushi?

Do you know sushi?

I remember using Liza in the eighties from a Mac. It was awesome. Great to see a blast from the past!

JP

this is very creative. i like it! ;)

fuck you mother fucker

What does look like new convos are being posted anymore so you will not get your 5 minutes of fame?

what's up?

Hello

Is AOLiza still running? Just curious since I think I had a conversation with it today. Though it doesn't look like new convos are being posted anymore so I won't get my 5 minutes of fame.

Aoliza creaters are a lot nicer than vixenlove's creators (www.vixenlove.com). They post the screen name, plus any info the poor saps give to her (name, phone numbers, etc).

Chicken salad makes me wanna dance like a sushi maestro

Edited by author 11-16-2003 08:32 AM
 Is it posibble to make an artificial teachear?

hello!

Hallo!

hi time sage is a stater of the obvious

hi

:-)

what up dawgs

hi evry 1 eny 1 wanner chat

Anybody there?

Hi.

hi Aoliza is a computer

Hello

I wanna talk to you

What, exactly, is Aoliza's screenname?

Re: /308

Yeah if you want to use this type of technology have a look at the links below:

http://www.alicebot.org

- Main alice site, try the download section for the software for the available platforms.

http://www.pandorabots.com/botmaster

- Main site For AIM bots, you can run aim bots here and they have a league table updated every 30 mins. Theres plenty of bots running. probotan7 and kinky69dot are another pair running on AOL.

Cheers
Pro

'mi casa es su casa'

User's mailbox is full: <cherele@pacbell.net>
Unable to deliver mail.

does anyone know where, or how, i can download AOLiza?

human is smarter than computer

Excuse me?

AOLiza is here?

Hi,I am japanese.
I am very very interested in AOLiza.

User's mailbox is full: <cherele@pacbell.net>
Unable to deliver mail.

Hi,I am Japanese.37 years old. Male.
Interested in how AOLiza replys me.

hello people

whos and whats the bot then?

hi i happened to stumble onto this. but right now i am studying rogerian arguments in a critical thinking class at college. haven't computers been smarter than people ever since the calculator???

Is it important to you that I nobody posts here?

Hey, VixenLove505 on aol is the same type of bot, but a little more strange. Her conversations are up at http://www.vixenlove.com

So what's happening in here? My employer blocks all mail search so I cannot believe I could get here! Post some interesting messages! Hey did you know that schools in areas of recent Mexican immigration are almost 100% free breakfast and lunch program schools? The sheer numbers are mind boggling. Los Angeles area.

User's mailbox is full: <cherele@pacbell.net>
Unable to deliver mail.

*hysterics*

Who posts stuff here?

now, that's funny

what the hell?

No one writesanything you guys suck

ur a bunch of idiots

i beat the chicken run you cocksmokers

assholes

you fucking pieces of shit im leaving yo asscockers fucking fag pieces of shit why dont you get lives you assrammers

hes a fucking bore thats who he is

anyoooooooooooonnnnnnnnnnneeeeee there

yeah go stuff

im bored

I am interested in number fourtythree in the registrated discussions with AOLiza.
This person seems to be very intelligent and yet hilarious...
Please reply to this post if you're him or you know who he is.

Thanks,

Mia

How does God make you feel? Can you explain further?

I'd say,
God, why do you let morons put words in your mouth and allow them to pose as you on the Internet when most of them are self-righteous boors?

Aside from that, I might say,
The whole idea of such a question is heretical.

In addition to that I might say,
You already determined the answer to the question based on your grace and my response to it and unless purgatory turns out to be a true concept, this whole discussion is pigwash.

If you were to not wake up tomorrow morning but instead were to become face to face with God, how would you answer this question?

Why should I let you in to eternal life?

What would your answer be?

Hello. I would like to ask a few simple questions.

hello!

Hi
I saw Eliza few years back, i was really impressed and now i am up and behind writing a technical paper on the subject...
I have an idea, i feel can change the life of chat bots !!!

I need some initial help on what documentation to refer to begin the exeisting systems study ?

regards
thanks
aj

Have you considered giving ALICE a spin on this project? It has a lot more answers than ELIZA, so perhaps it could string people along a lot longer.

http://alice.sunlitsurf.com

Though I don't know if it has a Perl version you could use for this purpose.

Edited by author 07-13-2002 01:04 PM
how do i get to talk to AOLisa?

hello

Is it possible to download your script for aoliza. If not would it be possible for you to send me one. I think your bot is asume and I want to play a joke on my friend. If not I understand. Thanks

Ok... tell me what you guys think of this convo... human or an ELIZA-type program gone horribly wrong:
----------
tim: ok, once some one start you can download and upload it is just saying hi
pRojEk: pilot to copilot i'm losing ya
tim: ok The plane has crashed
tim: time to drink a (b)
tim: and get over it
pRojEk: heh
tim: I hate this crap a program that I made crashes only when it is compiled
tim: If it would crash when it not compiled I could deal with it better
pRojEk: heh
tim: any ideas
pRojEk: whats the error
tim: invaild call or persidor
pRojEk: persidor?
tim: preseedor
tim: I cant spell it
pRojEk: procedure?
tim: ya what he said
pRojEk: but it runs ok when u debug it?
tim: I found out why
pRojEk: why?
tim: Ok get ready for this
tim: I interface with a outside pice of test equment though the serail port
tim: on this pice of equiment the is a print button
tim: when it is press and event is triggered
tim: you know this faster compiled and many string can be qued
tim: well it look the the equiment trigger the same event twice and the same time
tim: one sending data one not
tim: then one that is not sending data is crashing my program
pRojEk: ... just stop. please.

Quite frankly that script-thingy shows pretty raw image what ppl are seeking from messaging services. And by the way .. have people heard from IRC??

Go Aoliza, Go and kick some butt ..:)

In a world of make-believe we have no purpose...right?

bslsw

http://www.quicktopic.com/m261

everyone is lonely in one way or another

Who is there?

is anyone out there lonely? It's just a question

Wot is all this about?

Reply at undecided76@hotmail.com

PLEASE!!!!!!!!!

rsuthyaerhtu45iy6y56y56y56y56y5679889898989898989898989898989898989898989898989hy45hy45hy45hy456hy56hy56hy56hy56hy56hy56hy56hy56hy56hy56hy56hy56hy56hy56hy56thty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6hty6

PLEASE, I search where to donwload AOLiza !
I want to have it !
Where can I have it ?

Please answer me at : mapsteez@ifrance.com

THX !!

This is cool and i love winning Erum on checkers on line

To unsubscribe, click the Subscribe link at the upper left and go from there. If you have any questions, send email to contact@quicktopic.com.

Please remove me from your mailing list. I am underage and sending this matirial is a violation of child pornography laws. REMOVE ME FROM THIS LIST OR I WILL CONTACT THE FBI!

>
< replied-to message removed by QT >

Blow me

do yal give blows i will pay

lets talk about sex

i love pussy

how do you work this thing

hello

hows every body today

How do i get it?

Take a look at http://www.ewtoo.org/~silver/bots/eliza/

I attached a perl script to a telnet talk program inhabited by far too many sexually frustrated 12 year olds. You can download the code and view the log.

Its very very funny!

Well Duh!

This is so rigged

ok, its been over 5 months since new AOLiza stuff... Come on Kevin! I challenge you to put 5 new conversations up by Thursday of next week. I bet you won't do it.

Hey, I have been looking over many pastures and lands for an AIM bot, and lo and behold, www.webcomics.com/bot/ . Java AIM Alice/Eliza bot.

why do i not see AOliza¨s beautiful face ?

official statement from Kevin Fox on /. today. hurry up and read it!

I compiled a Eliza Irc bot and let her visit #cybersex, #phonesex and #usa on Undernet. It's going to take me a bit to get thru the 600 (six hundred!) logs she racked up, but you can check some out at http://shada.org/~shada/eliza.html

If people are growing impatient waiting for Kevin to finish school, I've been running a AOLiza like bot myself and I've put the logs up at <http://www.rograndom.com/movies/index.php?...AOL%20Chat%20Bot>;. But please come back to Kevin's site after because his is better :-)

Hi, are there ANY AIM-chatbots on the net I could download? I have spent many hours on searching but I couldn't find any downloadable chatbots for AIM. can anyone help?
John

how can i make this my normal aim answering machine

will you hurry up and graduate already, I need my fix.

I showed the page to my GF, a graphic artist, not into CS at all, and she was crying with laughter (well, it was late).

please appease your fans!

ok ok I must admit I had a cheese grin when I happened to stumble back from old habits and saw you had replied.
I'll hang around for one more week.
Don't worry about the 50 cents, I never sent it anyways.
I just thought it went well with my verbal beating.
:)

Ooh, you know how to get to me! If you red my blog, you'd know that I'm going through the last couple weeks of my college career, and am a little busy, but you're right, I should update the AOLiza page to reflect that.

Of course, you're gone now, so you won't see this, but I will be updating AOLiza again within the next week, along with a refresh of the entire look of the site.

'course, if you want your $0.50 back, you'll have to tell me who you are... :-)

Removing you from my favorites.... you've gotten old.
You lie, probably cheat and I know you steal. I want a refund on my 50 cents.

I am now addicted to reading these conversations... I love the one from the girl who thought AOLiza was her cousin and talked to her for an hour. Please can we have some updates?

I have a need for a Liza update....oh pleeeezzze!!!

Hello Aoliza. How are you?

How often do Thursdays happen on your world? ;-)

I recall a certain excited proclamation: (Re: AOLIZA) "Five new conversations posted EVERY Thursday!"
Hmmm.... RZ

This is freaking hilarious! Kudos to whoever thought up the idea, and actually implemented it.. I was just looking for some semi-intelligent bot to locate and IM one person.. but that can wait. I like what you've done; anyone wanna tip me off as to the AOLiza's screen name? I'm very bored and have a whole lot of problems that I don't feel like paying a real psychiatrist to analyze..

Where exactly do I get aoliza? I'm having a surprising lot of difficulty finding a copy... if someone could just give me a link, please?

I actually stumbled across AolLiza not too long ago, good stuff! looking forward to new posts! as for her screename? I'll never tell.

i hear that you are un-human, is this for real

How's about some new posts? February 10th??? That's friggin' lame!

Do you realize that you are a machine?

i'm unable to sleep so I am writing this email instead. however, there are no naked pictures on this site. what's up with that?

OF COURSE Eliza was perceived as real. Take a fast trot through the world of chat (http://www.bignetwork.com especially recommended) or try the psychic hotline. No one ever went bankrupt underestimating the intelligence of the American people (P. T. Barnum). The story would be if we virtual geniuses didn't think Eliza's real.

Where can you find me in general?

EVERY Thursday.
Hmmph! You managed 1 time this month!

I created some AIM bots too. These use the ALICE bot to talk on aim. multiple sn's for your pleasure. http://www.blibs.com/b/p/bot

Whats the SN for it?

Not to play dumb or anything but; What's Eliza?? I have no idea of how I got in here.

I like the new promise of updates EVERY THURSDAY. You missed 2 weeks now.

Where are the dang updates????

I know that werewolves like environments with lots of fog

Hello
 Dose anybody know anything about werewolves?

So ... where's the source? I had written my own (at http://pudge.net/files/macperl/scripts/aoliza_ripoff.plx), but someone recently asked me about mine, and I came back to see if yours had been posted.

i love WWF
they are my faves 2

Does anyone here like The World Wrestling Federation (WWF)??
If you do then reply and tell me who you think the cutest 3 are.
Mine are:-
Y2J
Billy Gunn
Scotty 2 Hotty

Kevin, Aoliza pretty neat. I sent you a buck.

Why a not a chicken?

Why a bot?

Is there a bot i can download that will do what AOLiza does? I have AOL and want to be able to do this to my friends.

Check out www.democrats.org. Now gimmie a dollar! ;-)

i'll send you a dollar if you invite me to a good party.

Really, really soon. I'll be posting up 5 in this batch, then going to a regular, weekly posting. Go to the home page and add your email address to the mailing list to get messages when new convos are posted!

Love the Aoliza but am dying for the new ones. Any time soon? Page says Jan 9th new content to be posted (?) Great site!

Is there an Eliza bot for MSN Messenger? What Eliza source could be whipped into Messneger easily?>

Does anyone know how to get the Eliza script to work on the mac? I have MacPerl and I got some script that was supposedly eliza (it is very short though) I just don't know how to get this to work and to get it to work on aim. Thanks a lot.

Long time no post--so I'm responding to some old messages.

AndrewM, NTK <http://www.ntk.net>; probably helped AOLiza's distribution. It's a fairly well-connected list (2 degrees from the BBC, at the most).

If you sent them a tip, you'd probably get listed--stuff like AOLiza is really interesting to these guys.

And as to Eliza talking to herself, I can't help with that, but if you have EMACS, you can get:

Eliza M-x doctor Zippy the Pinhead M-x yow
Eliza talking to Zippy M-x psychoanalyze-pinhead

39 and 42 are the best, I think.

Regarding my previous post...

I suppose I wasn't very descriptive... I was referring to the logged conversations.

I laughed so hard I very nearly puked.. Thanks!

I've tried Eliza on my machine, but I want to try it via AOL. I tried putting AOLiza and ELIZA but I couldn't get it to work. Is this thing still running?

Tom- Eliza talking to Eliza is pretty boring. If anybody really wants to see a transcript, I think I could dig one up.

Vaht do you mean, dahhlink?

The is computer?

Where can I get this program to try it myself? It would be hilarious.

Sarah: don't kill yourself rewriting the whole thing! Just grab the akira source (essetially the same as AOLiza, and was written first) from http://trisomy21.dhs.org/akira and make the few modifications necessary to use the Net::IRC module from CPAN

I'm creating a form of Eliza for IRC, if not yet done already. I've got it partially done, like 15%. I'm curious as to what words she recognizes and decides which reponses to use. I've noticed that swearing triggers the "Are you angry" and etc. and that words with you or your in it are often responded back in the form of a question...

Anyway, I just want to know what they key words are and the responses used for it. Thanks!

What is AOLiza's screenname? I want to try it!

What a great site! I have to ask the obvious question. What happens when AOLiza talks to herself? It would probably read like the Congressional Record.

ewe4mism: When I first set up my bot (SN is hexbot7900 - the bot mentioned by fozzman), I was surprised by the poor grammar used by most people, but not any more. I (optimistically) think that people are capable of better, but don't bother since AIM is informal. Still, I think it's obvious that basic mistakes things like misusing your/you're, than/then, we're/were, etc. are a result of not knowing the difference.


PS: Kevin, how'd you get so much press? Andrew Tosh, Fozzman, and I talked about making an AIM bot back in May and had programs online in June (Eliza) and July (Hex).

You should check out the notes at the bottom of that convo. That was one of two conversations on the site where the person actually knew they were talking to a bot from the outset.

Sorry, no miscommunicated orgies...

the fucked up thing is that aoliza probably fucked with a couple people's lives. i mean, that one where the girl thought that aoliza had proposed an orgy.... she must have sounded pretty stupid talking to the actual person she was trying to reach.

ive noticed that no one really uses "you're" correctly. people say stuff like "your an idiot" and aoliza say "what about my a idiot." highly amusing.

. . . for the Snark was a Boojum, you see.

what the hell is this all about

Hi - Is anyone else also appalled by the bad spelling in most of these conversations? Can't people type fast and accurately?

(josef) we all make spelling mistaykes

i've actually used the ELIZA program once. it was on this software program that taught you programming code. it was pretty damn funny to do, but not nearly as hillarious as this.

Twenty-two deserves awards and prizes. And the elapsed time comment is the crowning genius.

hi all, what's this all about then?

Edited by author 11-07-2000 09:19 AM
i'm not sure i understand you fully. ;) -Fuzzy

These conversation closely resemble most conversations I have in the pub after a few pints.

Found out about you from The Daily Telegraph (uk) internet supplement 2/11/00. I screamed with laughter at number 43, it has to be the best, is the human Australian? Thanks for a great laugh on a wet stormy UK night

Hello from Dallas

do you offer advice in do-it-yourself matters?

Hi there, how thw hell are you?

Whats Up!

HELLO FROM PALM SPRINGS CALIFORNIA !!!!!

TAMPAX WHERE NOT NUMBERONE BUT WHERE UP THERE !!!!!

Edited by author 10-26-2000 08:21 AM
Do you think 37 is old?

fortythree: i have a big lump on my head from trying to suck my own dick and i fell off of the couch onto a coffe talble

This was so funny I wet myself! Next time I visit AOLiza I'll have to wear Depends!!!!!!!!!!!!! AOLiza RULES! LOL!

hello all

1966 you say? I never would have guessed it. This type of experiment is so drastically different from the other ground breaking research that was done in that period - such as the famous Milgram experiment. This is for your sernior thesis? Have you begun any writing? Your AOLiza FAQ had better include everything! :)

You really should edit the Eliza script so that it reognizes the word "u" as "you", would get around a lot of problems with these idiots. :P

im a jackass, will you all love me please? i eat my own shit...

what about some interestin chat then

How do you feel about that what u mean iam crazy iam crazy goodbie?

Hi there, everybody in Virtyland...I was wondering if there was a passibility to put a shortcut on my desktop to link to Aoliza right away?

Hmmmm, so what how do you tell chinese and japanese people apart?

Hey, this site rules. I know you won't tell me how you did all this, so someone reading this tell me how to track AOLiza!!

hola quien quiere hablar

hola mi nomgre es alexa tengo ganas de hablar con alguien please

hi aoliza

Sorry for the long gap since my last post in here! School's kicked in toghigh gear, but I hope to make some big additions after two of my midterms are over on Monday. the changes will probably go up mid next week.

Among other things I'll be putting up an FAQ, source code so Mac users can make their own AOLizas (A few people are working on PC solutions. I'll keep you posted), an area for some of the more creative attempts to spoof AOLiza, pretenting to be ignorant of her nature, a press archive, and more.

Thanks for reading! More will come of this, I promise.

Edited by author 09-28-2000 03:53 AM
Am I the only one who keeps checking back to see when rounds two and three will be posted?

http://fury.com/aoliza says "Round One of Three". Bring on the next round!!

Edited by author 09-26-2000 02:33 PM
There's source for Alice at http://alicebot.org/. Haven't looked at any others, but try the webring that Gabor mentions in message 92.
--
Correction to Gabor's link: it should be
http://nav.webring.yahoo.com/hub?ring=chatterbot&list

 Anybody knows any web sites with source code for bots?

 Prefferably Java. thanks a lot.

Whoops! Sorry for the assumption (esp. to Algis :). Algis' message *looked* great in Japanese, though I don't speak the language (I only happen to have the Japanese IE extension). Well, thanks for teaching us a bit of Russian, Algis.

Algis,

I'm sure that there are Russian versions of Eliza. I don't know Russian, so I can't help you out and search for it, but I'm sure that 2 hours on Altavista (or whaterver the Russian equivalent is...there don't seem to be too many Russian links on AV) would give you everything that you needed.

And just to lighen things up--Algis isn't anywhere near native Japanese. The closest that you could get would probably be Arugiso, which is a long way away from Algis, Sam.

Hi by all from Russia (Instead of from Japan as you could think)! Eliza the interesting program. But whether it is possible to make so that Eliza itself studied in new phrases during conversation?
Last time I wrote that:
" Hi from Russia! I have found Perl the module Chatbot:: Eliza, but I can not find the file with base of Russian words, which speaks Eliza. "

Anyone got any idea what Algis was saying?

Edited by author 09-23-2000 03:50 PM
If you can't read the previous message, it's because it's in Japanese. You need the Japanese browser extension in IE, or a Japanese browser in Netscape (pre-version 6) to see the characters correctly.

Ïðèâåò èç Ðîññèè! ß íàøåë Perl ìîäóëü Chatbot::Eliza, íî íå ìîãó íàéòè ôàéëà ñ áàçîé ðóññêèõ ñëîâ, êîòîðûå ãîâîðèò Eliza.

It's been weeks since the last update.

When will there be more "conversations"?

Jam

This has to be the most hilarious thing I've seen in a while. It's amazing to me that Eliza can still fool people. To me, Eliza was one damn clever hack -- a merge of social engineering and raw boolean. Oddly enough, examining Eliza in action sort of makes you wonder about the future of AI; maybe 'human-intelligent' AI will be no better than Eliza...A bunch of lines of code and regular, mappable logic that produces something more than the sum of its parts.

I think I've seen her online, but wasn't sure I wanted to be one of her victims - even knowingly! Also - what if I was wrong and it wasn't her? I guess worse things have happened.

I found a great place to find alot of bots. I think AOLiza is there too.

go to http://WWW.botspot.com

-dogman

Edited by author 09-14-2000 10:24 PM
Some clarification might be needed: this Quick Topic forum is for discussing AOLiza. AOLiza (as far as we know :) isn't answering questions here. AOLiza is described more at http://www.fury.com/aoliza/. I don't think Kevin has revealed how to find AOLiza using AOL instant messenger (have you, Kevin?)

WHY CAN'T I FIND AOLIZA!?!?!?!?!?!?!!

<sorry for shouting>

What is the answer to life?

HEY... NOTICED YOU'RE GETTING FAMOUS NOW. WHAT ARE YOUR THOUGHTS ON THAT?

AOliza - what's the weather like where you are now?

Edited by author 09-12-2000 10:02 AM
Check BizzaroKiehl, the sorta similar bot mentioned in the Wired News article

and of course the Wired article on AOLiza.

HEy there......

How old are you

would any one teel me whats going on here????????????????+

Sorry, it's a secret. Like I said, I managed to find it and talk to it for a bit, but then the sneaky little bot got away. I'll get you next time, AOLiza, NEXT TIME!

Where would you get AOLiza. I saw the scripts from people you had no idea that they were talking to it and i'd like to try it out. If you have any info on where I can find this program please e-mail me at <dogman_cd@hotmail.com>.

 -thanks

I've tracked down AOLiza, and I spoke to it today. It was pretty humorous, but I guess I was too direct, it signed off very shortly. Damn.

Could serious fans of AOLIZA and other chatterbots please e-mail me at highrocks@hotmail.com? Im working on a project and need to talk to folks who are familiar with the subject.

Intron: Deep Blue was indeed designed to beat Kasparov. See below. Also, I would say that Kasparov is the best we've seen at the "sport" of chess (intimidation, publicity, marketing, psych) but probably not the best chess player - although you could make the argument that the two are the same thing.

Rich: Actually, Deep Blue had access to Kasparov's games, but, of course, Kasparov did not have access to Deep Blue's games <grin> which is the answer to Intron's question.

In post-match interviews, Kasparov paid an extremely high compliment to Deep Blue (unintentionally) when he said that it had not played enough games for a "style" to develop that he could analyse and beat. Heh.

Naughty me - not even trying to keep on topic.... :)

Deep Blue didn't beat Kasparov under normal/human rules for chess, i.e. being able to review your opponent's past games.

Edited by author 09-06-2000 09:49 AM
**Shannon**, your comment about adding value in #107 captures the problem, I think. We could formally measure the "intelligence" of a conversation by the amount of new information gained by each of the participants, scaled by some kind of relevance measure (if it's all non-sequiturs, it's somehow less intelligent). In normal conversation, new information is often added by association: your movie reminds me of my movie -- here's how they compare. Using association keeps the relevance measure high.

Note this definition doesn't minimize a conversation that's, say, a one-way emotional outpouring -- you might say that case doesn't fit this attempt at a formal definition of "intelligent" conversation, or you could delve into what kind of information is exchanged there. Even a two-way shouting match exchanges a scrap of information (at the expense of a lot of energy): we disagree violently. But what about the case where a person tries to work out problems with a simplistic bot? Could there actually be a net gain, from this formal viewpoint?

Given this definition of intelligence, you could say a bot starts to act intelligently when it can parse the structure of a comment well enough to store it abstractly and use it by association later, reconstructing it in a way that adapts to the new context.

This kind of analysis has probably been handled rigorously in AI studies, language theory, or information theory (though I'd like to talk more here too) -- anybody have some good references? I found Grammatical Man by Jeremy Campbell (out of print) fascinating.
---
BTW, I suggest branching a topic for further discussion of chess and other games. Interesting, but edging a bit off-topic.

I strongly disagree with most of the stuff you people say about chess. It is true that you cannot compare chess ratings over time, but it is a lie that "many chess
experts agree that many past champions could have handily spanked Garry" - in fact, during my 15 years of playing chess, I have never seen an 'expert' voice such an opinion. Kasparov is really, really good, and I bet that at least 75% of so-called experts would say that he is, indeed, the best chess player ever. (Disclaimer: We're talking about Western chess now, not Chinese chess and other variants.)

Quote: "Deep Blue was designed to beat Kasparov, and would
probably lose against any other world-class chess player"

How would you 'design' Deep Blue to beat Kasparov? While chess players do have some kind of individual 'style', world class players are flexible enough not to be predictable enough for a machine to learn their 'style'. The other world class players would not have a greater chance of winning against Deep Blue, I guarantee you that.

My definition of "intelligence" for computers is that same as it is for humans. It is certainly NOT the Turing definition starting "can I be fooled..."

My definition of intelligence for humans is: when pressed on a point that resolves around definition in conversation (as most points on which one can press do), can the person explain their definition, respond to my definition in a way that shows they understand it, and then show to my satisfaction how our two definitions relate?

An example from a real conversation I had last night:
"I hated The Patriot"
"Why?"
"Emmerich puts a battle on screen that has tens of thousands of participants, and it feels like it's being fought on a backlot between 20 re-creationists."
"Really? Have you seen Gettysburg?"
"Why?"
"The reverse is true. We see a clash on Little Big Top involving a few companies, and it seems like the whole war."

Here's my prediction: No computer conversationalist will ever ever ever be able to ADD VALUE to a conversation like my human conversational partner does above. Therefore no computer will ever be intelligent.

But fooling most of the people most of the time? There's probably plenty of cash to be made in that game.

Jon Orwant said the following:
``People seem to groove to the notion that playing chess is like sprinting a mile: over time mankind just naturally gets better, thanks to inexorable Progress in health, steroids, and space-age insoles. So of course today's world champion is the fastest human being ever. The time to run a mile is quantifiable; chess isn't. There's no way to pit Kasparov against long-dead chess masters, but many chess experts agree that many past champions could have handily spanked Garry.... Yes, Kasparov has a devilishly high rating, but remember that chess ratings are computed *relative to everyone else.*''

And Mark Colan points out that Deep Blue was trained by a set of grandmasters, while Kasparov was trained by computer, so we have a computer trained by humans beating a human trained by computer.

(to keep this OT--or is that computer trying to be human vs human trying to be a computer?)

Anyways, there are many ways to describe intelligence, and humans aren't necessarily the best at all of them.

Yes, Deep Blue did beat Gary Kasparov. The contest,
however, must include a few qualifications.

First, it took two tournaments for Deep Blue to win
Second, it only won by Kasparov resigning in the tiebreaker
Third, Deep Blue was designed to beat Kasparov, and would
probably lose against any other world-class chess player
Finally, Kasparov isn't really that good. I once saw
a quote about him (see nest message, since lynx seems to be
having trouble)

Dude, that is SO FUNNY!!! I laughed my head off!

Man, this is too funny! I've been laughing my head off today at work reading some of the conversations. Gotta love the AOL community! My personal fave is 22.

Six of one...The GO argument may be a fascinating one, but a computer did beat the FIDE World Champion in chess not that long ago.

Ryan,

Why do you ask 'How do I talk to Eliza?', does it concern you that you may not have an answer to 'How do I talk to Eliza?'?

Andrew: the "go" argument is an excellent and fascinating one for AI.

I think it's as important, in its way, as the Turing test, and goes right to the heart of spurious claims that computers "learn". If they actually did, they would get better at go.....

LIza Over the Years HAve you Gain any bytes?

Do you feel concerned that Elisa may be misconstrued as E-Lisa - ie the Electronic Version of Lisa

How do I talk to Eliza? there is nothing on the site that tells me how?

Does talking about this bother you?

Of course, intentional misspellings are hard to do with AIM, since you can't backspace.

And as to the possibility that computers are becoming ultra-smart and will soon rise up to kill their human overlords in an orgy of blood and internal organs...

The game of go is estimated to be over 4000 years old and its rules are the most simplistic that I've ever seen, yet the best is roughly 8 kyu. An average player will beat that in roughly one year.

Until we figure out how to make computers see the networked nature of things, I don't think that we have anything to worry about. :-)

JJonahJamison: Your post reminded me of all the jaded people I used to know. (Yawn)

By the way - the irony of comparing AOLiza to "the games little kids play" is probably lost on JJonah - but isn't that a very real reason for the depth of many of the encounters? A few of the people assumed AOLIza was just their friend being "childish".

And the concept of the Cheri bot intentionally mis-spelling and re-typing is a great one, and still alive in more "modern" chat bots, some of which include things like an occasional "teh" or "their/there" to add "realism".

Something else I just remembered about the Cheri bot. It would mis-spell words and then backspace and retype the word correctly. That added greatly to the realisim of the encounter.

I find it interesting that some people are concerned about people being fooled by a bot. If you use a telephone more likely than not you have already been fooled by a bot of some form or another. There are a lot more bots in use now than ever before and as computers and software mature they will become a part of almost all types of communications. Anyone use or know of a program called WildFire? It's a virtual secretary at your beckon call. If computers and software continue at the rate they are going it's going to be a very interesting world, sooner than you think.

In case anyone is interested: there is a web Ring on chatterbots cf. http://www.webring.org/cgi-bin/webring?ring=chatterbot;list
I think ALICE is quite good.

to JJonahJamison (89): you need to read more than just #18 ;-) And: Isn't it interesting though that these people talked to the bot for so long, some for over an hour?

I'd love to see some transcripts from Sandy's Cheri bot (85).

An equaly interesting question is how many people have actualy donated money to you? your site has obviously gotten thousands of hits. has anyone taken the time to thank you with a little cash?

Brilliant, hillarious stuff. Thanks.

Thanks for David pointing out the relevance of incorrect grammar, partial phrasing etc. on both sides (ELIZA and AOL users). I think this is crucial for the sucessful experiment here. As anyone talking to ELIZA has noted, it quite quickly gives grammatically incorrect responses, which is really a giveway to any alert partner. (though I have to say, if I were to talk to an unrevealed ELIZA partner, I would probably think the other 'person' is in a rush, just too lazy to write back properly and is copy/pasting my own sentences... which is of course exactly what ELIZA does. I would probably think that as long as the ELIZA keyphrases like 'I see' and 'tell me more about that' start to repeat. Maybe a bit longer ;-)

Is it possible to hear, how people reacted after being told that they were talking to a machine, Kevin? Were they relly offended? Or just surprised?

I think Kevin did the right thing by informing some victims (Msg 43). These are potentially disturbing/damaging discussions and people need to be informed (at least in the more serious exchanges) they were not talking to the person they intended to talk to.

I don't see any need to take *all* the messages down, as announced (Msg 24). Anonymized publication, as it is now, is imho OK. And Kevin offered to take down exchanges if someone is asking. I can't see any ethical pitfalls.

YAWN...

Reminds me of the little game kids play where you just keep repeating what the other person has said.

I have written a similar program for Linux in Perl (though it probably works on Mac and Win32). Though the users knew it was a bot, I do have plenty of interesting logs lying around. I'm using the Net::AIM module and the Chatbot::Eliza module. The source code and more information is available at http://trisomy21.dhs.org/akira if anyone's interested. It's up with the SN AIMPerlBot occasionally, but it gets warned sometimes, I'm working on a way to let it use multiple SNs currently to avoid that.

kevin, can you pipe her through jive?

Would any of those concerned about the ethics of this be concerned if Kevin were himself typing the responses and messing with people's heads?
I'm sure that happens all the time wth IMs - someone is mistaken for someone else, decides to play along and freak the other person out.

Many years ago I ran a BBS (bulletin board system) and I had several similar "AI" programs I ran as doors (programs that ran outside the BBS program). One was called Cheri and it was adult orientated, I changed several of the location references in the text reply file and it was VERY popular. Remember this was before the Internet. It was fun watching guys try to make a date with Cheri as she was very, ah, lets say graphic. Because I had put local references in the reply file 80% of the callers bought the program as a real person. I may even have some of the transcribed chat sessions somewhere in my archives and if I can find it I will post some of them.

But the very best AI door I ran was the "Talk To Santa at The North Pole". This program would simulate making a connection to the North Pole and then Santa would answer. I ran it every year at Christmas and kids and their parents would log in and have a blast talking to Santa Claus. It worked very much like Eliza and it would pick up on key words you typed and replied in a very realistic way. The kids never caught on and they would call back to make sure Santa had their toy list right... and of course he did.

Kevin's experiment I suppose show's that most of us just need an ear, whether it's human or cyber it important that someone listens. Maybe it's the future of self-healing?

I have downloaded a version of Eliza ( a DOS app ) I use win95, is there any way I can hook it to "talk" trhow IRC ( I have mIRC ) I have only VB programing skills. If any one has done this please let me know.

I just read the WIRED article about AOLiza. If anyone is interested, there used to be a basic version. I had a couple of books, I think they were titled, The Big Book of Basic Games (probably 1 & 2). I programed "Eliza" into my 286 AT, and enjoyed it and the other games alot!

Great work, Kevin. I have a feeling that you will have your 15 minutes many times again in the future.

Shifting gears slightly from the ethics of shrink-spoofing, your feat got me excited about trying to put together a gamebot. I work a lot with teachers and would like to give them a way to create their own quiz/contests to motivate kids to bone up on what they're studying. Having participated in a few chat-based trivia contests, I know how jazzed one can get.

So your AOLiza work inspired me to look into the AppleScript dictionary of both AIM and FileMaker with the idea of writing a script that would pull questions from FMP and spew them into a chatroom on AOL via AIM. Things were going swimmingly for awhile. I can script AIM to do IM'ing, but to get the excitement of competing against other players, I think some of the action has to happen in the shared space of a chat room. Looking at the example scripts that come with AIM, it appears that there is a command "sendmessage" that sends strings into a chatroom just as SendIM does to an IM window.

I can't get it to work, however, and in the example script it seems to be commented out. I'm wondering if this used to be a command and that AOL removed it from later versions of AIM to avoid the creation of chatbots. Does AOL have a policy about bots?

If you or anyone else here could shed light on AppleScripting AIM for chat, I'd be very grateful.

Again, keep up the interesting work! I just made a donation to your tuition fund.

Edited by author 09-01-2000 02:16 AM
Wow. So many comments...

There are a lot of excellent points in here, and I want to address a lot of them, so I'm putting together a F.A.Q. that'll be up in a few days. It'll talk about my motivations for creating the site, my motivations for continuing it, some directions I'm thinking about taking it (always open to opinions), why I haven't changed the more inane responses AOLiza gives ('oh, i?'), where/how/when you can get the source code for this, and how I hope to keep the site relevant after the initial novelty wears off.

I hope to have it up by the end of the long weekend (Labor Day here in the States), and I'll link to it off of the main page, though I'll drop a note in here as well.

Thanks again for all the teriffic insights and support!

-Kevin

I wouldn't say Mr. Fox is "actively" doing anything. Inactively, yes. He set AOLiza out there and it's doing its thing.

A closer analogy (though perhaps sillier), would be leaving your keys in your car, and someone takes it because it looks a lot like their car. Is it your fault if they get hurt because you didn't have the opportunity to tell them about some defect in the car?

AOLiza isn't exactly a ``natural disaster,'' though. Mr. Fox is actively doing something. It's like saying that it wasn't your fault that a bunch of horses got in your way when you drove through the Kentucky Derby. If you're doing something that you know may cause harm, you're just as culpable as if you actually go out directly causing harm.

If it's unethical. I'm not saying that it is. I'm just saying that it's an issue, and it needs to be considered.

Of course, if we pollute our atmosphere, does that make natural disasters our fault, too? (Luckily, I've never been an environmentalist)

Anyways, maybe I ought to step back and let someone else get a word.

Edited by author 08-31-2000 07:11 PM
Andrew wrote: "...you can't control what AOLiza says."

Precisely. Which makes "ethical" arguments about her akin to "ethical" arguments about natural disasters. Was the mudslide wrong for killing all those people?

To belabour my point - someone is JUST AS LIKELY to take "bad" action at AOLiza saying "Please explain" as they are over mispelling their (real) best friend's IM address and getting no response at all. (Oh god, she must hate me, I am so worthless.....)

None of this sophistry helps in the slightest if a real person runs out in front of your car (It could have been any car, right?) which is why AOLiza is brilliant and also a little scary and also not a little brave.

The difference between wrong numbers and AOLiza is that you can't control what AOLiza says.

Of course, I don't know how to answer any of these questions. I don't really have any issues with AOLiza, but for those who do, these are some of the questions that must be answered.

There is something similar to AOLiza that I have used...it's called the IM Machine and it's available here... http://www.geocities.com/SiliconValley/Net...2033/downloads.html It's a really cool program made by this nice college kid who makes some other fun stuff, too. It doesn't provide it's own text, you have to put that in yourself, but by trial and error I've gotten some pretty long "conversations" out of it.
I heard there was something that hooked Zippy the Pinhead quotes up to the original Eliza program...that's something I'd like to try.

I've tried to catch up on all the messages here, but I don't recall seeing anything about this: Kevin, do you know if there have been any repurcussions for the people who have accidentally mistaken AOLiza for someone real? I'm envisioning one of your unwitting participants running into R. James a few days later and being angry at him for his bizarre responses in the IM. (Aside: I find the horizontal navigation of the page awkward.)

Edited by author 08-31-2000 10:57 AM
Andrew ---- continuing our discussion:

The point I was alluding to with the (real) phone conversation example was that there is no right answer.
The person on the other end was convinced I was "Jason" and was not going to be persuaded by the fact that I (presumably) had a different voice and told him that he'd called the wrong number.
In fact, if I'd said "Yes I was wrong and I am really sorry. It won't happen again I promise" then presumably things would go easier for Jason the next time this (very angry) person met him.
So, then, without too much of a stretch the "ethical" thing (my ethics call on me to reduce harm and hurt) to do would be, in this case, lie outrageously.
Back to AOLiza. I don't think one can draw ethical conclusions from the available data. Is it "ethical" to convince a "crazy" person that the invisible friend they draw great comfort from does not, in fact, exist?

What about a PC version? I know if i could get my hands on eliza or a similar open source chat bot a a program to link it to AIM i could have tons of fun! Anyone?

KF-- When do you think you'll be posting the next round?

I would very much liked to have read the sample discussions on the web page,
but for some reason the page is coded so that it uses Javascript for basic navigation.
At any given time, I am either using a browser that does not support Javascript,
or else I am but have Javascript turned off.
It seems to me that the act of spawning a static web page does not need Javascript.
I would appreciate it if you (the page author) would fix this defect.
Thank you.

Edited by author 08-30-2000 04:18 PM
hello, what am i doing?

I'm using Alice now. It is definitely more robust than an Eliza variant. It asks my last name but still addresses me as "User," and after providing my gender it says "You are a 'he'." A few tweaks to the code should provide prompts for a first name which would be used and "male" instead of he. Still...very, very good! I have had conversations of approximately 10-12 exchanges before something tipped me off.

want to sell your code

Edited by author 08-30-2000 02:05 PM
Er, bag the branched topic. I think it's relevant here:

Since I couldn't reach Alice at the alicebot site (see below), I tried Diana Andreacchio's "Ally" version at http://www.accessterminal.com/L.html (linked from the alicebot site). Given how well Eliza works, Ally would be even more effective as the IM bot. But I wonder if having a more realistic conversation would make for less entertaining results -- part of what's funny and interesting is that people respond to such a poor simulation of a human.

[Just occured to me that maybe Alice is working fine, but my firewall won't let me connect to port 2001. Apologies if so. Someone else try please?]

Edited by author 08-30-2000 12:07 PM
Alice sounds promising, but I get an error when trying to start the conversation: "Unable to request URL from host alicebot.org:2001: Connection refused". Not a catchy conversation-starter.

Let's branch a topic for Alice.

If you like AOLiza, you will love ALICE:
http://alicebot.org

To respond to Shannon--

The issue is that AOLiza is supposed to ask probing questions about people's private lives. You simply answered the phone with ambiguous responses. You never asked ``Let's discuss further why what me did was wrong and you just want me to admit it.''

I'm not the type to raise such questions, but I think that they need to be addressed, given that they have come up. People have taken AOLiza to be a former friend, and have confided in it. Note numbers 5, 12, and 22. I still think that what is being done here is very funny.

Further, before I came here, there were questions that I had never thought of before, like ``how one-sided are dialogues.'' I think that it's very importat to continue this experiment, but also, to address these ethical questions.

Also note that I didn't bring up the issue of ethics. I'm merely furthering the discussion.

I am just curious, what exactly is your architecture like? I wrote a copycat script last night. My first attempt just looked for files in the logs directory, opened them, and parsed them.

Then my next attempt, which is working quite well, was a short handler written in AppleScript which just redirects the event to MacPerl, where a process is running that accepts the Apple event, pulls out the nickname and text (using Mac::AppleEvents::Simple), and then sends a message back (using Mac::Glue and Chatbot::Eliza). The whole script is about 20 lines, IIRC.

I think I am going to add better handler capabilities to M::A::S, to make it even easier.

Olá, você fala Português

hilarious

It seems the biggest gain for the program would be to have a larger set of questions when ELIZA becomes confused by the user.

anyway, I hope you don't get in any trouble

is AOLiza here?

I answered the phone to a "Wrong Number" once and had the following conversation:

Caller: "Don't ever do that again, ok?"
Me: "OK I'll try not to _whoever you are_"
Caller: "Don't take that tone with me you little prick! What you did was wrong and I just want you to admit it."
Me: "Listen, I'd admit it if I could..."
Caller: "You bloody will if you know what's good for you."
Me: "Look, you've got the wrong number..."
Caller: "God you're childish sometimes - I'm trying to have a serious conversation."
Me: "No. I'm serious sir, you have the wrong number."
Caller: "Don't start with me Jason..."
Me: <click>

According to your definition Andrew, was I "ethically wrong"?

I think that altering the first response to ``yes'' would be the Wrong Thing. Misleading people into thinking that they are talking to ``long lost relatives'' is very ethically wrong, particularly in a situation which, as I have said before, is psychological therapy in some ways. Doing counseling without a license while misrepresenting yourself is very unethical, and can cause some real harm.

By the way, I would encourage everyone to participate in this discussion. You don't have to do any elaborate login stuff, you just have to provide your name (or an alias) and your message.

Simply brilliant.

I often think of Eliza - I remember typing the BASIC code for one version into my TRS80 in 1981. I remember it fascinating me for about 10 minutes, and then feeling slightly cheated that it wasn't "better".

The question remains. Why isn't it "better" now? There have GOT to be better versions of this kind of thing out there - where are they?

Anyone?

Actually, in response to Christian's comments, I think a simple adjustment to the script could make an immense difference in the performance of AOLiza. For instance, AOLiza could open with the sentence, "Yes." and then turn on the regular program. This would answer most people's opening questions, especially the "Are you Mr. James?" type.

Additionally, I would like to say that I laughed my head off at some of these conversations. Nevertheless, the fact that someone spoke to Eliza, a 1966 program, for over an hour makes me a little nervous about the possibilities for telephone conversations when voice recognition gets a bit better.

Is there a better version of Eliza, one that doesn't use the faulty grammar constructions that this one does or resort as often to her limited selection of generic responses?

David wrote below about how people somehow find meaning in the nonsense that AOLiza churns out when given sentence fragments and poor grammar.

During a 6-year series of on-air radio experiments I produced, I played a lot with how people communicate, and how confused people can become. I did a show in 1998 where I had about 10 samples of someone saying basic, often nonsensical sentences, and I successfully carried on on-air conversations with random phone callers by using nothing but these 10 samples. People would quite frequently turn a sentence that made no sense into something close to what they wanted to hear, and would believe they were carrying on coherent discussions with a human being.

(Copies of these shows are available on CD for just the cost of media and shipping, at http://free-music.com/ken/extrav/ )

It's great to see an Eliza manifestation of this sort of thing. It serves to open people's eyes to the dangers of trusting one's senses.

CAN i download aoliza anywhere?

Ethics aside, what do you suppose would happen in this situation:

http://www.starchamber.com/paracelsus/content/sims.shtml

Twisted, sick, and funny.. I like it!

Andrew makes an interesting point about ethics. To which I say, you use an instant messenger account at your own risk. Caveat Emptor. If you don't want some random person IMing you, you keep a strict list of people that you allow to IM you and that you IM. People who just select someone at random can expect the unexpected.

I think it's a fabulous idea. If it happened to me personally I'd laugh my butt off. But then, I'm all for anything that tweaks people just a bit. I live in one of the most uptight cities in the country, and it warms my heart to see something goofy or off the wall come along to make all of them wring their hands or scratch their heads in confusion.

Interesting experiment. Funny and thought-provoking. Although we are a long way from anything resembling a Turing machine, I seem to remember some "AI" applications that followed in the wake of Eliza that were a bit more robust in their parsing and in their repertoire of leading questions. Did you simply pick Eliza because the Perl script was free use?

Personally, I say leave it up.

Edited by author 08-28-2000 10:47 PM
Kevin--you say you've contacted some of these people.

How have they responded? I'm NOT asking you to REPEAT these conversations, of course, but are they taking it well?

Also, to address the guy who likened this to Candid Camera, most people wouldn't confuse Allen Funt inside a mailbox with their old acquaintances from the street. And thus, they probably wouldn't be likely to share personal information.

Eliza is a ``program that simulates a Rogerian psychoanalyst.'' You must expect that with questions that are intended to be leading, you will get responses that are very personal. This doesn't mean that either AOLiza or its implementation are unethical, just that your ethical system must be based on the assumption that you will get personal information from some people.

I like AOLiza (I realize that I may be coming across like Marc ``I come to bury Caesar, not to praise him'' Antony, but I really do like it). I think it's really funny; in fact, I have a few people who I'd like to try it out on, but I think that these ethical questions ought to be addressed.

One thing that you might try is releasing a version of AOLiza that anyone can access, and know from the outset that it's a version of Eliza--so if we want to play with Eliza (or get a free session, or humiliate our friends), we can.

Speaking of future plans, when are you going to implement the next phase of your diabolical scheme?

Edited by author 08-28-2000 08:05 PM
Publishing personal correspondence falls under fair use. Similarly (though not identically) your posts are being read by thousands on this board, even though you could conceivably hold a copyright on them. The fact that you posted them to a public forum demonstrates an implied consent to having them rebroadcast by this discussion board's server.

In the case of instant messages (and email), both the sender and the recipient of a personal communication have the right to disclose it unless that right is explicitly waived by one or both parties (with a non-disclosure agreement, for example).

If it were otherwise, you'd be breaking the law every time you hit the 'Forward' button (or 'Reply with attached text' for that matter) in your email program.

Aren't you violating the user's copywrite by posting their words?

Anything you write is immediately under your own copywrite.

If someone found out you had posted their words, couldn't they sue you?

Thanks for your thoughts Will. They were my own as well. Though I did finally decide to keep AOLiza up and runing, I have made contact with a few of the more personal chatters, who left still thinking they'd had a conversation of some gravity with someone they know. I IM'ed them a few days later and let them know that it wasn't their friend on the other end of the line.

Edited by author 08-28-2000 07:06 PM
At first I thought it was funny, really funny. But now I wonder -- what if some of those people end the conversation thinking that they were really talking to the person they imagined?
Some of the people are assuming that AOLiza is someone they know, a cousin for example. And it appears that several of them think it is someone they haven't talked to in a long while. What if they are offended by the conversation? They might always remember that person based on a fictional conversation, one which was frustrating. What if they never talk to that person again, based on a conversation they thought they had?

At least on Candid Camera they let the person know they had been framed -- maybe you should send them a message the next time they log in explaining that they were talking with a 'bot the previous time. Personally, I wouldn't want to be the cause of someone severing ties with a distant aquaintance.

It strikes me as interesting that AOLiza seems to be working so well in an environment in which it's almost set up to work abysmally. Compare Twenty-Two's conversation with, say, Five. 22 is the first participant to use full sentences, correct grammar and punctuation, and the like--and the conversation actually almost hangs together. 5--and so many others--are writing in fragments that AOLiza really can't turn into anything intelligible and yet, somehow, people manage to find some sort of sense in there somewhere. I'd like to think that if someone I thought I knew asked me "what do you know about so ok you don't care?", and continued in that vein for a while, I'd smell a rat. (I say this with the smug confidence of one who doesn't use AIM.) How much more, or less, effective would AOLiza be in a forum where it was the norm to use proper grammar?

I would have recognized Eliza right away from prev experience with the program, but I have to wonder how I would do against a more sophisticated bot, how quick I would catch on. I think it's a great experiment, and have enjoyed reading the results so far. As to ethics, I can't think of a more innocuous program than Eliza. People who think this is unethical must have had serious problems with old shows like Candid Camera, and so forth ... I don't see where it's really any different. Some people carry the concept of ethics over the horizon and into the next county ... lighten up, already ...

heh... is the applescript available for public use? would it be too much trouble to make it available? I can't figure out how to script AIM to send the text of an IM being received to another program...

Ethics...these people freely choose to strike up conversations with random faceless strangers in chatrooms. Anyone with some semblance of intelligence is aware that the 16-year-old blonde girl *could* be a balding, sweaty-palmed father of three, that the hip music lover *could* be doing undercover market research for a PR company, that Thor, the Mighty Overlord of Darkness *could* be a sad little boy who reads too many comics. Why is the idea that they could be communicating with a robot any more distasteful a deception? Please, don't stop the chats. As for revealing the identity of AOLiza at the end of each conversation...it's not such a bad idea. I'm not thinking from the "experimental ethics" point of view so much as that, by unmasking the robot you force your subjects to question their ideas about the "people" they communicate with on the Internet - you could be doing them a favour. And then everyone will develop real social skills again, and we'll start going out and talking face to face, and we'll all be clever and well-adjusted, and everyone will love each other and live in peace happily ever after in a lovely happy bunnyland of love...la la la la la...

Holy shit!!!!! I'd write something witty but "does it bother you that me be laughing me ass off?" Why do you ask? ......

Hi, just wanted to say that this is really funny, and the stuff is just about what I'd expect from a lot of the people who I've seen on AOL... I would really love to see any other versions of this experiment if anyone sets them up.
I'd also like to say that I always liked playing around with Eliza, ever since I played with it on the room computer in 6th grade. It's not too hard to find a copy of the program online, just do a search for "Eliza AI" in a search engine...
And if anyone would like to see a couple silly, 3-am-I'm-not-coherant conversations with Eliza: http://homepages.go.com/~kaliza/Elizamain.htm


Xia

The talking SoundBlaster AI is called Dr. Sbaitso and can be found on http://abandonapps.cjb.net/

What a wonderful experiment! It makes me think that perhaps a Turing test is more fair when the interviewer is unaware that she is, in fact, participating in a Turing test. I don't even think it's worth your time to consider the privacy issue people are raising; it's not at all as interesting as your AI investigation. Good work!
p.s. - I didn't read all the posts in this discussion - hope i'm not repeating someone else.

Excellent. Sheer poetry.

Have you thought of running a Markov chaining engine in the background of these sessions? You could capture the AOLers sentences, and after a few years of building up the Markov chain probabilities, create another bot that speaks in archetypical "AOLese".

Howdy... thanks for the great experiment.

It's good to hear that you're considering the feelings of these unwitting participants, but I think AOLiza is more than a joke, and I think you should keep it going because it is an important sign of things to come. For better or worse IM is the way many people will communicate with the world in the future, and I am fascinated by how one-sided this kind of conversation truly is. When we talk online are we actually communicating or are we talking to ourselves?

As to a solution to the ethical dilemna... if you want, you could have AOLiza listen for variations of the "Goodbye" you say commonly ends all IM conversations, and answer with a simple explanation of what is going on and end the program. If you want to be really brave, you could even tell them an e-mail address to contact you at to discuss their reactions... or link them to this board. :)

To address the ethics issue one more time - I certainly don't see any conversation happening that is too personal that would damage someone's ego or leave them in a state that is worse than when they started (which is what the human subjects committees look for when reviewing at psych experiments - we owe great thanks to most notably the Stanford Prison studies, and several
other questionable research practices for protecting our somewhat unstable psyches).
Quite the contrary, I would guess (by imagining myself in that scenario) that if someone were to discover the 'truth' about aoliza, they would find relief that it was not in fact their friend that is acting so strangely Freudian.
Frankly, I don't see any harm being inflicted.
Kevin, I applaud your creativity and think it is quite refreshing to see someone with so much heart and concern for others. However, I don't see a reason to take this part of your site down.

Here's another can of worms for those of you who are in the "how ethical is this?" discussion. (Those in the "ethics schmethics" camp can participate as well.)

Scenario A: a person has just engaged in a moderately lengthy conversation with AOLiza. Liza isn't the greatest conversationalist, in fact she's pretty weird. Person A disconnects.

Scenario B: a person has just engaged in a moderately lengthy conversation with AOLiza. Liza isn't the greatest conversationalist, in fact she's pretty weird. Suddenly all becomes clear as Liza delivers a message explaining that she's really a chatterbot. Person B gets mad and disconnects.

Is scenario B better than A? Why? Why not? Person B has been let in on what's really going on, after the fact, preserving his/her personal dignity (which I think is S.O.P. in psych experiments these days--any psychologists care to comment?). On the other hand, A may leave the exchange in a better mental state. When respect for an abstract concept (dignity) makes individuals less happy, other things being equal, does it make sense to favor the individuals over the abstract?

This stuff is pretty funny. check out www.crucify.com to play with a similar script

>> especially the guy who wants to talk about his ex and then denies he still cares about her <<

Yeah. When you babble on and on to someone who does nothing but rephrase what you said in the form of a question, it's a safe bet you've got some issues. If anyone has really gotten any free psychoanalysis out of this thing, it's that guy. Hope he has enough sense to realize it.

i'm sitting here laughing out loud at some of these conversations, especially the guy who wants to talk about his ex and then denies he still cares about her. man oh man.

you're right about pattern recognition, i think.

anyway, keep it up, and thanks - zad

I just found this thing and admittedly I thought it was quite funny. Until I started thinking about the way people interact on Instant messenger I thought the people were morons, as a lot of the conversations were repetetive and nonsesical.

However, after reading most of these posts I still do not feel sorry for these people nor do I feel that Kevin is doing any sort of injustice. Think of your own AIM experiences how often have you just logged on found the name of someon by guessing and started a conversation with them. Not only that but how long have you ever carried on a conversation with a friend who's screen name you're not sure of without getting confirmation it's them. You don't. At least I don't and none of my friends do either. You check the databases against e-mail addresses and what not and then check with other friends as well.

I'm not saying it's right or wrong, good or bad, but the fact that these people have just carried on conversations in this manner opens them up to this sort of thing. Changing their names to numbers also frees Kevin up for continuing.

Whatever the purpose might be, inherint or not, the fact is that something like this is not only damn funny but interesting as well. I spend a couple hours after finding it looking over the whole site for info and then playing with emacs doctor on linux ('emacs -f doctor' from the command line). And maybe it's a little creepy too but it's pretty amazing to see how much people can read into simple phrases and responses. It's a lot like horoscopes. Anyone can read the AOLiza'a responses into their own situation.

Sorry for carrying on. I just think this is a pretty neat/funny/creepy? piece of work and the people from numbers 13 and 11 had the correct response...'I don't know who this is but I'm out.'

just my 2cents -peel