The Conundrum of Building Out Virtual Worlds

Even the most expansive, expensively-developed game world is fairly thin. The barkeep offering to sell you his treasure map has scant back story to share, making your efforts at conversation unrewarding. Storefronts passed while Accomplishing the Mission are largely placeholders. It's never very deep, or broad, or interesting. That's why there always needs to be some exciting Mission.

Yah, obvious connection to the final paragraph here; life in Utopia, with nearly all our actual problems solved, feels thin, leaving us aristocrats eager for some craptastic Mission or other.

In fact, if a game creator tried to fill in every last detail, it would be off-putting, ala uncanny valley, because it couldn't possibly be deep enough to be convincing. There isn't enough money in the world to hire enough artists and writers to build out a really rich virtual world. In fact, there aren't enough artists and writers, period. Quality creative work is painstaking and does not lend itself to output at scale. Creativity is a retail enterprise, not wholesale.

Crowdsourced virtual worlds like Second Life are a bit more enriched, with a degree of sprawl and detail unachievable by a centralized team. But the tools used to construct in-game elements are inherently limited, so it's paper thin despite the diversity. And the mechanics of gameplay are necessarily rigid (we can't need to learn new gameplay methods or buy new hardware whenever we swing open a door or walk around a corner), further narrowing the experience. And there can be few through-stories in such a world. There are hard limits to the ability to align the work of disparate people.

So when cyberpunk authors like William Gibson, Rudy Rucker, and Neal Stephenson conjured up Virtual Reality - an immersive online world that could be dived into at will by a future populace - it was hard to imagine how such a thing would be built, inch by inch. I was dubious, even as someone who's run online communities since their early beginnings.

But it just occurred to me that AI is perfect for this fill-in work. Yes, the result would be chopped up and reassembled from bits and snatches of previous works of imagination, but that's ideal! After all, that's true even of the worldiest world we know - this one. It's why history rhymes. We revel in deja vu.

An AI could build out an immensely large, immensely detailed, immensely diverse world with comparative ease. And it would likely be doable within a couple generations of AI development.

It’s a perfect match for the recycling nature of AI. A given area might be very loosely reminiscent of The Maltese Falcon, another like a scenario from the Illiad (with a bit of hippy-era Berkely). And it wouldn't feel like cumbersome "genre". An AI could apply a far lighter touch than humans are capable of. I.e. you don’t need to register that you’re in the Maltese Falcon. Most earthly experiences are rather thin regenerations of previous contrivances (hence “archetypes”), anyway. So the signature AI mash-up touch would serve to meet people’s expectations. Things would seem just familiar enough.

If the virtual world were generated on the fly, users might even configure it, e.g. order more familiar or less familiar environments and behaviors. That’s easy for AI. 


Hardware aside, the trickiest matter would be wisely setting standards. Neither too chaotic nor too confining. That would need to be done by humans, but it takes a certain kind of mind to envision - someone like Walt Disney or Ben Franklin or Steve Jobs. You need to really understand human behavior, while most of today's tech bros have a naive comic book view of human nature, shmuckily imagineering proposals like Universal Basic Income and Mars Colonization as cool ideas for a happy society.