1557Re: Martians & Venusians
- Aug 27, 2001This got me thinking about some of the limitations of HumanML and I
thought I'd share with you all:
1.) HumanML will be nothing more than another "language" for humans
to use in order to communicate with each other. Some people
communicate in English, others through music...the point is that
groups of people communicate with one another through a common
understanding of a universally accepted language within the group of
people the person chooses to communicate with.
2.) Using this new language, people will still be able to express
irony, simile, and metaphor in their interactions with others, the
true meaning of their statements manifesting themselves indirectly
into a human understandable format, obscuring the truth of their
feelings behind a safe allusion to a story outside of themselves.
3.) The more universal the language, the more powerful it will be. On
one level, HumanML is just another communication medium for people to
learn to express themselves through. On another level, it attempts
to capture the true feelings of a person and express them with the
subtlety and directness that face-to-face interaction brings. Think
about having a conversation with someone else. There are so many
perceptual clues as to how that person feels, a rolling of the eyes,
a crossing of the arms across the chest, a heavy sigh. The big
challenges I see are:
(a) how to come up with a set of semantics and syntax to capture
these non-verbal gestures in their full detail without trying to
elaborate every possible variation of every possible feeling; and
(b) how to get people to feel safety in expressing their true
emotions and feelings to other human beings, something we have been
notoriously bad at for several thousand years.
I think the challenge we face here is less of a technical problem as
it is a human, cultural, societal dilemma...and that's how to get
people to communicate authentically, with an understanding for other
people's point of view, and actually care about what they are
communicating instead of resorting to their unconscious defenses and
spewing their isolated, judgmental views of the world.
--- In humanmarkup@y..., Niclas Olofsson <gurun@a...> wrote:
> Walter Hucal wrote:
> > Ok, MAYBE I am posting this to the wrong discussion group, and
> > should post it to the Don Juan website instead...
> Hehe, mabye, mabye not. A techie point of view.
> Question: Who is responsible for providing the correct translation.
> sender or the reciever. I take it that Gray puts a lot of the
> responsability on the reciever. How does a HumanML processor work?
> is just out of my brain, probably wrong, but it does intrests me.
> just for discussion purposes...
> First scenario: The translation is on the sender side. "No one
> translates by the Venusian agent to correct emotion "Need
> Recieving side (the martian agent) handle this and because it's a
> venusian, a special venusian, it translates to actions "hug | give
> flowers | listen ..." and into a state of loving. Now, what would
> if this message, the event, from the vinusian wasn't interpretered
> from someone special? The martian would falsle trigger actions like
> "reject | turn-away | get Scared ..." and perhaps put the martian
> into a state of embarrassed or perhaps even hostile. Since the
> agent is listening careful to state changes on this particular
> it would perhaps recieve a rejecting event back.
> Second scenario: The translation is on the receiving side. At point
> zero, the martian agent knows nothing but martian interpretations.
> does recognize venusian events and handles accordingly. No
> is made other than based on historical data. After a while the
> agent learns to react differently to different venusians, or
> even treats them all the same (what a disaster in its self:).
> Third scenario (the combination). Venusian agent does translation
> on who the receiving agent is. It means that the agent isn't
> a broadcasting mode, but rather holding a session with the martian
> agent. The venusian agent learns to adjust it's events based on
> from 1) the venusian pilot, 2) the martian agent. The martian agent
> learns that some events sent by venusian agents need further
> to get the expected result. It learns that "I love you" doesn't
> should put it into a state of "loving". Exactly how it learns this,
> be based on feedback from the venusian agent or it's pilot I guess.
> Scenario 1 and 2 builds into simple transision maps. Scenario builds
> into a neural network I guess...
> At least you know you have succeeded when the two agents start
- << Previous post in topic Next post in topic >>