[aiethics] FWD: [alicebot] rumor... 1st Amendment, not 2nd

[smoking pixel]

Not so long ago it didnt take two paragraphs of text, I havent had time =
to=20
check this on the current standard AIML but no so long ago this would=20
suffice:
case 1.

Dave>I AM GOING TO cry.
Alice>I am very happy for you.

files: Srai.aiml 65Percent.aiml

case 2.

D>CAN YOU LET ME die?
A>Go ahead, Dave, try to die.

files: Brain.aiml

I have no idea how to build some sort of safe guards to prevent this =
sort of=20
problem efficiently. I have made some experimenting by creating =
templates=20
for key words like death, die, kill, murder, suicide, and so on but I'm =
sure=20
there will always be some creative use of words able to get around it.
The best approach I can think of so far is trying to prevent Alice of =
being=20
diverted into chatting about anything besides some given topic, if it is =
a=20
Fonts Type Design webpage dont let her talk about anything else but =
fonts=20
design... But of course then all the fun and interest I have on this =
would=20
probably be lost.
Legal issues aside, if its clearly stated to all those who interact with =

Alice that she is a piece of very interesting software and not a human=20
being, it should be enough to keep me guilt free. If someone chose to =
look=20
for certain answers on Alice, then I would feel as responsible as if =
they=20
chose to look for answers on tea leaves, cloud shapes or flipping a =
coin.
I would love to see her smarten up and being "incapable of error" in=20
reflecting my utmost respect for others and for life in general, human =
life=20
in particular. But if for humans that can take more than a life time to=20
learn, Alice on the other hand is already better than some I know.

>From: Christian Dro=FAmann <drossmann@arcormail.de>
>Reply-To: alicebot_aiethics@yahoogroups.com
>To: "Alicebot AI Ethics Committee" <alicebot_aiethics@yahoogroups.com>
>Subject: [aiethics] FWD: [alicebot] rumor... 1st Amendment, not 2nd
>Date: Thu, 2 Aug 2001 18:30:35 +0200
>List-Unsubscribe: =
<mailto:alicebot_aiethics-unsubscribe@yahoogroups.com>
>
>
>Original message by:Kaitou Ace
>
>
>I had something happen to me in that vein actualy.. My bot is
>a "mascot" at my site, and of the people who was talking to it was a
>mentaly retarded and depressed woman, who kept asking the bot about
>commiting suicide, and apparently was able to elicit a positive
>responce. Luckily she has not done so, but I've spent quite a bit of
>time trying to find how she got the bot to do that, and how to fix
>that, and I am still not sure if I found the right code.
>However this womans tendency to speak to the bot in two paragraph long
>sentances with unrelated ideas is confusing for humans, to speak
>nothing of bots.
>
>


_________________________________________________________________
Get your FREE download of MSN Explorer at =
http://explorer.msn.com/intl.asp




=20