At the recent VW conference in LA I heard Dr. Ben Goertzel speak about his AI project at Novamente: virtual pets. This was within the context of ownership, content protection, and how to move items from world to world. He made the rather intriguing comment that it’s not just about objects but about “cognitive agents”. He explained the challenge through the Novamenta lens: if I buy a virtual dog and then train the dog, who owns the rights to that ‘cognitive agent’? Afterall, I’ve invested time in training the dog, so it now “houses” more than just the object (the dog) or the program (the scripts that run its AI) it’s also now “home” to everything I taught it.

Ben hypothesized that the business models for virtual worlds will become increasingly complex. It’s not just about the objects or their scripts, it’s also everything I’ve taught those objects.

I was having a hard time understanding cognitive agents – somehow worrying about whether I can take my dog with me to openSim didn’t strike me as a high priority, but it was an interesting intellectual challenge.

But the AI people are doing more than teaching virtual pets: next up, virtual babies. And if you’ve ever seen pregnant women walking around the Grid with their stomachs making noises as you attend a conference or whatever, you’ll know what I’m talking about. Odd, maybe, but no more odd really than trying to make your Sims happy by plopping them down in front of a big screen TV.

Photo: Jon Birke’s Photostream

And I’ve posted before about how the issues around avatar identity will keep getting more complex: as our avatars become intelligent in our absence, or as we have multiple avatars, some of which are agents and some of which are controlled, this stuff is just going to feel more and more like a rabbit hole.

The Singularity Institute (I had no idea there was an institute – shockingly, I couldn’t find anything about vitamin packs on the site) highlights developments on the AI front (emphasis added):

“With luck, during 2009 we will make serious progress toward creating a “virtually embodied artificial infant” (which may take humanoid, animal or virtual-robot form: that’s not the point) based on the OpenCogPrime design, and will also see a variety of other AGi approaches implemented within OpenCog by a diversity of researchers.

This year the focus is on getting the basic AI mechanisms in place … next year (though for sure more work on AI mechanisms will continue!) we hope to segue into more of a focus on artificial baby-building.“

I’m, um, looking forward to it?