These are low hanging fruit or "Duh" observations.
> -----Original Message-----
> From: [EMAIL PROTECTED] [SMTP:[EMAIL PROTECTED]]
>
> 1. Practical side: given the enormous limitations of current Natural
> Language technology, how can we work around those limitations to produce
> useful and interesting effects? (Lessons may be drawn here from existing
> games, notably mystery-solving adventure games; perhaps the best solution
> I've yet seen is that employed in the _Sam & Max_ game from LucasArts.)
[Bullard, Claude L (Len)]
First I need to know what is bot and what is notBot. OTW, what kind
of effects are employed in the Sam & Max games? I suspect that
a very interactive 'bot is an expensive prop to use.
> 2. Visionary side: how long will it be before we have decent NL
> processing? (Someone on www-vrml -- P. Bralich, maybe? -- has been touting
> his company's NL work; I tried the public version available on the Web and
> was fairly impressed by it but it too breaks down pretty quickly.)
[Bullard, Claude L (Len)]
No clue. If one drives the interactor through an interaction quick
enough,
they may not notice the botNess. Like magical sleight of hand, much
depends on misdirecting their attention, so pacing counts.
> 3. Philosophical side: how much AI do we need, and how much are we likely
> to get, to produce convincing enough illusions of character to let users
> suspend disbelief?
[Bullard, Claude L (Len)]
See above. How much we need depends on the qualities of the
interactor
and the interaction. Kids are usually non-plussed by the deus ex
machina.
So this can vary a lot by person, age, cynical persuasion. Ever
watch the
Mystery Science Theatre bits? Here is a guy with 'bots doing
exactly
the kind of critique that we did as kids to bad movies, but also
precisely
the critique one might do of the station 'bots themselves. It is a
conceit
inside a conceit and effective for that reason.
> 4. Lateral thinking side: can we entirely bypass the need for interactive
> characters controlled by the computer? (IrishSpace did it by
> pre-recording; VRML Dream will do it by having live actors; MUDs and
> MUSHes
> do it (for the most part) by having a large enough set of human players to
> flesh out the world.)
>
[Bullard, Claude L (Len)]
Sure. But if they are practical in the story, we can also take
advantage
of botNess. IOW, Robby The Robot acts like a robot; Commander Data
only does when the story has to take advantage of his botNess for
plot point. So much depends on the role the bot plays.
len