1  |   Order by default

Feb 9, 2000, 3:37pm

I've just finished a C implementation of Alice in AW. I'll post it in a
few more
days if anyone is interested.

[View Quote] > Concept,
>
> I have never pretended that conversationnal bot are very smart, and can
> now be compared to human.
> Actually, I personally do not very like the interaction that are only
> conversational. (I prefere that a bot do something rather than talk).
>
> However, you should know, if you were aware of what is being done in this
> domain, that one of the first use of conversational bot is to provide
> a more anthropomorphist interface to information (such as a FAQ, etc.).
> So they can bring some value. Indeed you can not expect a real human being
> be 24 hours a day and 7 days a week present to answer to your questions !
>
> say, and changes it to first or 3rd person most of the time.
>
> That is not true.
> This kind of (Eliza type of) behaviour has appeared to you because your
> where expecting too much from this bot.
> You have a lot of "rules" coded in the Alice "knowledge base".
>
> Actually, Alice is a technology (plus a basic knowledge base), and it is this is
> up
> to the designer to enter the "knowlegde" he/she wants Alice to answer.
>
>
> Because they can spend more time than if they had to reinvent the wheel,
> and besides they can expect to create in the future "knowledge bases" that
> can be reuse by different bots. (which is something almost impossible if you
> hard code your behaviour).
>
> Alice bot is very far from being perfect, and in particular the "pattern
> matching"
> mechanism is very limited when compared to other systems.
> However provides you a reasonably working knowledge base as a starting point,
> and all the code is open. (so if you are not satisfied you can change it).
>
> So, I believe that some people (of course not everybody), and in particular the
> non-programmers
> can be interested by this system.
>
> You should forget a little bit the dreams that artificial intelligence has
> promissed you,
> and just ask the question if this system can be usefull (create value).
> For me, the answer is "potentially yet", even if on the long term I believe in
> rather different,
> but much more complex solution, (such as evolutionary psychology and the likes).
>
> Thierry
>
> Thierry Nabeth
> INSEAD CALT
> http://www.insead.fr/CALT/
>
[View Quote]

1  |   Order by default