Debunking Art - fraudster with python AI engine?
Mon Aug 23 02:27:41 CEST 2004
"Nick" <primz at bigpond.com> wrote in message
news:4128f60b$1 at news.unimelb.edu.au...
> I recently went to an art exhibition, and one of the artists had an AI
> engine projected onto a screen with a keyboard for visitors to type
> questions in with.
> Curiously, I asked the artist if he had written the engine. He claimed
> to have, but as far as I know, he has done no AI study, and only
> started programming python some months ago.
> Naturally, I find it hard to believe he has picked up on natural
> language processing and the other various skills required to write AI
> engines in that time. My guess is he has a python AIML interpreter,
> and he's found a default "brain" somewhere and tweaked it with some of
> his own stuff in order to pass it off as his own.
> The engine could answer all the usual questions, like "what is the
> meaning of life" (42), "what is your name", "how old are you" etc etc.
> It was just a chatterbot, so ELIZA or a deriviative is marked off the
> list. I am guessing probably an ALICE engine is behind it all. Does
> anyone know any questions or commands one can issue the bot which
> might identify it? It required all questions to start with a capital
> letter, if thats any help.
>From the evidence you present clearly this is a fraud. Mastery of Python in
just a few months with no prior programming experience is definately fishy!
And never having formerly studied AI or read the tomes of literature on the
subject..... why oh why do we tolerate such lies and falsehoods from our
"artists". I for one vote to repress them into just painting what the eye
can really see ;-)'s
Originally I had planned to write a response from this point onwards about
how simple it would be to achieve what you've described. Firstly I'd have
mentioned that a little research would bring you to say something like
Markov chains for emulating natural language structure (a number of the
recent and best chatterboxes have used this technique). I'd have followed up
with how these are crazily simple to implement in Python using just a
dictionary and lists. And I'd have finished up by mentioning that Markov
chains along with a lot of other statistical tools and algorithms originally
developed outside of AI research have over time been brought into this
Then I spoke with my Girlfriend who's an artist and not an IT professional.
Who explained about the artist showing how life is really an
anthropormorphic projection of ourselves onto other objects. How in our
desire to communicate intelligently we project our own intelligence into
everyday associations. How from this we can create our own Deus ex Machina
based merely on responses that seem natural to us but as yet we don't
understand. But I don't believe any of that, hell she thinks clippy is cute
But it explains your question fairly well. Presented with a non comp sci
graduate doing something that appears to be the domain of a lifetimes AI
research, you've projected your inability to do the task onto them and
formulated the response that plagurism is at the heart of the matter. You've
failed to see through the smoke and mirrors and spot just how easy on one
hand Python is to learn, and just how much power on the other hand it allows
you to lever. And you've forgotten that Google is not just an IPO its also
your friend for research.
I wish you luck in your quest to prove this was a fraud. As for me its back
to updating skynet 1.0 to use decorators and some new introspection tricks
More information about the Python-list