[ExI] revive thread with new and improved content,

[Anders Sandberg]

On 2010-12-28 22:17, spike wrote:
> It occurred to me that the process of copulation is a real time
> simultaneous positive and negative feedback control system, where the
> control law is derived in real time by trial and error. Is not this what
> one does (or rather two do) when engaged in amorous activity? We watch
> and listen carefully, adjust accordingly: if she, he or they do this
> then we touch here, if she does that then caress over there, if she
> sprays us with mace, we shriek in agony and attempt some other approach,
> such as feigning a seizure.

Teledildonics has been around for a while, but I think you are right 
that there is a need to go beyond the (likely tricky) mechanical 
elements. Signal processing and the right user interface designs are 
going to be the real selling points once the kinks of the hardware are 
straightened out.

I chatted with an entrepreneur in the business, who had subscribing 
customers who "networked" with each other. I really think those data 
streams contain gold: run Kalman filters on them to make them predictive 
and smooth out network lags. Calculate eigensmooches that are *just 
right*, and can be both used to reduce bandwidth, might have the same 
appealing effects as averaging faces and be invoked as macros. In fact, 
with macros one can do erotic things one's normal body cannot - rapid 
repeats, sampled motions, extra limbs etc.

Hidden Markov models or reinforcement models linked to GSC or vocal 
sentiment analysis to learn what behaviors are appropriate in what 
situations, and slowly build a proper sexbot behavioral repertoire. 
Generally, the more data to play with, the better chances it can be 
mined for erotic benefit - especially if one can figure out how to 
estimate hedonic state from various cues.

My iphone has a simple app that uses the accelerometer to estimate sleep 
phases from body movements (no idea how it handles two people in the 
bed) and then uses that to try to wake me up at the optimal time. It is 
a neat demonstration of how a simple data channel can detect a fairly 
complex biological state. I wonder what other simple channels found in 
current computers and sex toys could be used to figure out hedonic 
information?



-- 
Anders Sandberg
Future of Humanity Institute
Oxford University