[ExI] AI wins at Poker now!
sparge at gmail.com
Mon Jan 23 20:14:46 UTC 2017
On Mon, Jan 23, 2017 at 2:20 PM, William Flynn Wallace <foozler83 at gmail.com>
> Here's what gets me: millions of videos and film depicting just about
> everything. Why not take from those shots the code for what you want;
> that is, if you want some virtual reality robot to walk like a person,
> then why not get a video of a person walking and take the code from there,
> eliminating superfluous actions? I have seen a person hooked up to all
> sorts of sensors that record his movements which I suppose is then turned
> into code. I don't see why this is necessary.
One reason is that video is only 2D. Another is that data captured via
other means can be more precise and at a higher rate. A third is that even
with unlimited data, emulating some things is incredibly difficult. For
example, there are countless videos of people talking. The biomechanics are
well understood. But no visual simulation of a person speaking has ever
done a really convincing job of it.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the extropy-chat