RE: Sexbots??

Billy Brown (bbrown@conemsco.com)
Thu, 11 Mar 1999 12:12:36 -0600

Terry Donaghe wrote:
> Wouldn't virtual-reality sex partners be better? Most of today's huge
> porn industry (it's in the billions of dollars I believe) is
> completely non-tactile. You've already got millions of horny guys
> (and girls) shelling out big bucks for magazines, videos, computer
> games and websites - think of how much more money you could make with
> a simulated virtual sex-partner?

This is certainly going to be a big money-maker for some people, but it doesn't really give the customer what he wants. Its just the best substitute our current technology can manage for a reasonable cost.

> The first phase could kick in as soon as we have simple audio-video
> surround VR (like with glasses and headphones). Instead of watch 2
> people go at on a flat tv screen, why not watch 'em go at it on your
> own floor! Or in your bed right next to you! Or in the tub, the
> neighbor's yard, wherever! The simple availability of such software
> could possibly be the largest driving force for moving this type of VR
> into the mainstream..

If you want to watch a conventional movie on a heads-up display, that's no big deal (maybe 2-5 years out). If you want real 3D VR, where you can walk around and watch from different angles, that means you need a computer capable of rendering complex models in real time (which means around 10^6 MIPS, which you might have in a home computer around 2012). If you want the sim to be controllable (instead of just following a set script), or to interact with the real environment at all, it also needs a decent physics simulation to generate dynamic models for things like hair movement, flexing muscles, and folds in clothing (multiple that MIPS figure by at least 100, which puts us close to 2020).

> Phase two could include tactile simulation. Wear this suit (fitted
> with a "special" cod-piece) and "get some" yourself while you watch.
> Hell, participate in multiple-partner sex orgies! Have sex with real
> (simulated) partners from around the world.
>
> Phase three is the full body immersion VR tanks - as far as you can
> tell you're REALLY having sex.

Cool. The tactile interface doubles the processing load (at least), plus we now need a good, detailed physics simulation (its also not at all clear that such a thing is possible without smart matter or real neural interfaces). We also need a complete sexbot OS & control program for your simulated partner, so it can do a decent job of pretending to be a person, and we need to simulate an entire virtual environment instead of just a couple of human bodies. Now we're talking 2030 before we can run the software, even on an optimistic projection.

> Sexbots? Bah. By the time we can build them VR will be far enough
> along to make "real world" sex seem quaint and silly.

High-fidelity VR is much, much more difficult than is commonly believed. It demands programs with hundreds of billions of lines of code, simulations that will eat up TeraFLOPS of processing power, animal-level AI programs, and extremely complex interface hardware.

If you are satisfied with a cartoon-quality world and a very rudimentary tactile interface, you'll probably have your wish within a decade. If you want something that is hard to distinguish from reality, you're going to have a long wait. I'm sure it will be implemented eventually, but humanoid robots (or androids) look far simpler to me.

Billy Brown, MCSE+I
bbrown@conemsco.com