How my program passed the Turing Test

by Mark Humphrys

A few years ago, I hooked up an Eliza program  to  the  net  with
some  interesting results. Now I've finally got around to setting
up a page to tell everyone about it ..


This goes back to an Eliza program I wrote in 2nd year of my  un-
dergrad at University College Dublin, Ireland, in early 1987.

In 4th year (early 1989),  I  hooked  it  up  to  my  account  on
BITNET/EARN  so  that,  when  I was logged off (disconnected on a
VM/CMS system), it would process all talk messages  sent  to  me.
There  were  numerous  hilarious  incidents, but this is the best
one.

So, it's Tuesday 2nd May 1989, 8.12pm in Dublin. I've logged  out
and  gone off to see my girl, leaving Eliza (or MGonz as I called
it) to mind the fort.

So some guy on the net decides to start talking  to  my  machine.
Someone  from  Drake  University,  Iowa,  USA,  where it is early
afternoon. He stays talking until  9.39pm  Dublin  time,  unaware
that  no  one  is  at home. During this time, my machine's brutal
cross-examination forces a remarkable admission:

*    when    was    the     last     time     you     had     sex
05/02/89    20:25:50    From   SOMEONE   at   DRAKE:   yesterday.
05/02/89 20:26:2

*  ok  honestly  when  was   the   last   time   you   got   laid
05/02/89  20:31:20  From SOMEONE at DRAKE: i told you it was yes-
terday.                              05/02/89 20:31:5

*  ok  honestly  when  was   the   last   time   you   got   laid
05/02/89  20:35:48  From  SOMEONE  at DRAKE: ok ok it was over 24
hours ago.  for you it must have     05/02/89 20:36:5
                       been               20                years
05/02/89 20:36:56

Police interrogation teams will be familiar with this  method  of
`keep  asking  the  same question, ignoring all answers, until he
cracks'. Note the timestamps on the rhs.

The next day, I logged in, and was amazed to  find  out  what  my
machine  had been up to in my absence.  Here  is the full conver-
sation. A couple of notes:

7 You are warned that there is a lot of bad  language  and  crude
humour  here  (I  always  wanted  to  send this transcript to Joe
Weizenbaum, but this  factor  inhibited  me  from  doing  so).  7
MGonz's  statements  (which are pure stimulus-response replies to
the last message) are all preceded by an asterisk. 7  Because  of
net  delays,  the  stimulus and the response do not always appear
side by side.

 Here  is the LISP source code. Don't ask me to explain it.



Eliza and Turing Test related links

7 biographies of Alan Turing  here   and   here  7   The  Loebner
Prize Competition  (and a Wired  article ) 7 Karl MacDorman's pa-
per on the Turing Test  (abstract)  7  Aaron  Sloman's   Eliza  7
Eliza's  etc.  that you can talk to: 7  Julia 7 one on  IRC 7 You
can also talk to a Markov chain on  the   web  7  Eliza  programs
available:  7   CMU AI Repository  (and  here ) 7  Fredtrek 7  PC
Therapist

Incidentally, I think that trying to pass the Turing Test is com-
pletely the wrong way to think about doing AI (see  here ).


Return to my  home  page.

 Mark.Humphrys@cl.cam.ac.uk