From: Richard Loosemore (rpwl@lightlink.com)
Date: Sat Feb 04 2006 - 11:13:40 MST
Rick Geniale wrote:
[snip]
> Try to explain:
>
> Is the concept of "finished" that is wrong.
>
> 1) PIBOT is not complete (we are currently working on it).
> 2) PIBOT don't know everything and cannot do everything.
> 3) We are implementing PIBOT general architecture right now.
> 4) We are working on learning strategies.
> 5) We are mainly working on natural language (for now).
> 6) We have said Turing Test by 2007 (but for us this is only a single
> step in AI).
> 7) Turing Test is a thing (play with the stock market is a completely
> different thing).
> 8) After complete, PIBOT should be trained to do things.
> 9) It will occour several time to train PIBOT on knowledge domains
> (exactly like a human it will have to be assisted by experts of various
> fields).
> 10) Nonetheless PIBOT is already a "singularity" (depending on how you
> define this term).
Rick,
Now you seem to be answering some of the questions I wanted you to
answer a few weeks back.
What you describe sounds like a work in progress, with no proof that it
is currently working any better than other projects, or that it ever
will. Many of us have projects in states that are at least as advanced
as yours, by the sound of it.
By claiming or implying in public that it is already finished, when you
now seem to be saying that it is simply in progress, you are being
somewhat economical with the truth, no? Do you have any reason to
believe that PIBOT already has advanced capabilities that put it well
beyond the capabilities of other AGI projects? If you do, could you
share with us what is the source of this optimism? What has PIBOT done,
so far, that has impressed you most?
And by the way, PIBOT is *not* a "singularity": you cannot redefine a
term that already has a clear meaning.
Richard Loosemore
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:55 MDT