Above: Meeting BunnyGPT, one of the first NPCs hooked up to OpenAI's LLM
Read Part 1, Will Artificial Intelligence Totally Transform Culture & Intellectual Work?
After discussing how AI might change movies and music, Adam Frisby and I debated how it could alter the future of the Metaverse. It's topic both of us have long been thinking about, Adam as CEO of Sine Wave Entertainment (creator of metaverse platform Sinespace) and myself as the author of The Making of Second Life and Making a Metaverse That Matters.
Wagner James Au: How about metaverse platforms? How much will gen AI transform them? Roblox I know is already experimenting with it for graphics, but how much will it change the overall experience in the next decade?
Adam Frisby: Probably; the general rule of thumb is: do big datasets exist for the job being replaced. In the case of metaverse platforms, you've got primarily the following jobs done by creators:
- 3D modelers
- Animator
- Texture artists
- Riggers
- Scripters
- Environment artists/level designers
Those are the game industry job titles, but it is analogous to virtual worlds. Of those; there are compelling high quality AI tools already for animation.
Here's a tool from 2018 [watch below]:
AF (continued): In terms of the rest; most of it’s not quite there yet - 3D models made by AI are still full of defects. AI texturing produces mostly trash. AI programming/scripting definitely is almost there, but not quite. Lots of investment going into that, and things like Copilot do generate pretty decent results a lot of the time.
So I'd say "don't panic, yet."
WJA: Why panic at all? Wouldn’t metaverse companies not want to hurt their ecosystem of creative users by adding AI programs that replace them?
AF: 97% of users in virtual worlds don't create content. That's an industry statistic from a range of virtual worlds. Of that 3%, 2% remix content -- make new textures for other people's models, etc. 1% do fully original creation. I am not making a value judgment here at all. I need to be very clear about that.
WJA: Do you think AI will replace that 1%?
AF: Probably, eventually. I think there's two reasons why
First, professional creators may end up replacing themselves.
It's a well known secret that a lot of big popular brands in virtual worlds already outsource a lot of their 3D modeling to freelancers and such through sites like Upwork. If the tools are good enough, why even do that?
Second, of that 97% how many will end up joining the 1% once they have the tools and capabilities to do so?
CLO 3D (above) is a sister company to Marvelous Designer, a fashion modeling tool used in Sansar and other metaverse platforms
WJA: So what’s your advice to creators in SL, Sinespace, etc. who now depend on these platforms for their living?
AF: 1) You're not alone - most of the world is going to have to grapple with this. Many, many industries are going to be affected. This is a societal level problem, and needs a societal level solution.
2) You've got some time, we're still a while away from AI being at human level for 3D; and then the tools to be adopted. I'd say, 2-5 years.
3) Think hard about how you can incorporate this; the winners will be the ones who can adapt - use the new workflows to save time, create more, market better. The market isn't going away, but it's changing. The car didn't eliminate the transport industry.
WJA:. My own prediction is gen AI in metaverse platforms will mostly be used for casual user-generated content, i.e. by average users who want to whip up a quick scene, but it won’t replace the monetized ecosystem. 3D creators will use the gen AI tools to save time and start with a base, but then add their unique flourishes to it with their own work.
If companies do try to replace their creative communities in any direct one-to-one way, there will likely be an uprising similar to what we saw with the WGA/SAG strike. Top creators have large passionate user communities who support them, so this would be a platform-wide revolt. And as with Hollywood, companies will have no choice but back down and make agreements on the human creators’ terms.
AF: Where I'd pick that apart would be a few places; First, the idea of needing to add a flourish, or human touch. I think this is a point of difference between us, but I don't see humans being involved beyond the "request". The idea that AI will always produce a lesser product, and need a special human touch, is an assumptive conceit.
I suspect creators will tune their tools to produce a style they like consistently - but the need to tweak the results will go down with every improvement to the underlying tech, until eventually there is no need at all.
You're right on the platforms though -- unless a platform can run with no creators, they won't annoy them. Netflix isn't going to replace film studios. Film studios using AI will replace film studios not using it.
In this case, the creators with AI will eventually win out over the ones without. They can create, faster, more efficiently and with less resources.
Casual users will adopt the tools too -- platforms probably will support that, empowering more creatives is a big part of virtual worlds. Users will appreciate those tools, but I don't think they'll be good enough to replace robust creators for some time. The reality is, some people are naturally more creative, and people flock to those styles and trends.
See: Instagram; also take a quick scroll here (and yes, a lot is junk):
AI image generators have come a long way, and people are finding ways of being really creative with them. The results are almost never tweaked after generation -- the work is on the input, tuning the prompts and weights
WJA: Specifically with avatar fashion, I don’t see the human factor and the brand association ever being replaced by AI. Blueberry for instance is huge because her massive customer base has a personal affinity with her. Look at real life fashion: there are companies in China and elsewhere that knock off the latest high end brands with cheap copies. But the brands remain popular and successful because they still create and market the trends.
AF: Agreed, I think there’s a lot of power in running a good brand.
FINAL THOUGHTS
WJA: Grandiose predictions around AI have consistently been proven wrong and there’s still no evidence that’s going to change any time soon, especially around human expression.
The SAG strike was actually doing the studios a favor, because there’s no proof that consumers actually want to pay for AI-only content.
Gen AI may become an important tech tool like Photoshop and Maya before it, and that’s great, but there’s no reason to think it’ll substantially supplant the culture we cherish.
More key, no technology is inevitable. Humans ultimately control it, and humans can collectively decide to control gen AI’s abuses, as is already starting to happen. And any AI prediction which discounts concerted human action is destined to fail.
AF: To take the side of the robots; every grandiose AI prediction to take, has been premature. Not wrong.
Futurists like Asimov have been speculating for a long time what shape that takes; I'd say their biggest mistake was predicting we'd be replacing physical labor, before intellectual labor.
There is nothing special about the human mind - we're a collection of neurons and chemicals. The idea that algorithms running in squishy meat cannot run in silicon requires a real act of faith. The idea that robot creativity cannot match human creativity requires similar faith.
So, to me, the argument from the position of human supremacy rings hollow -- what matters is the output If you cannot distinguish between work produced by a human, or work produced by a machine - is there a distinction? Does the distinction even matter?
AI knows us better than we know ourselves. TikTok's ability to hyper target videos at you, is better than the best marketing agency in the world. No human understands what you want better. It knows you better than you know yourself, or at least, it's more honest.
I see backlash against TikTok, but I don't see people not using it because of its targeting.
AI holds the potential to create content using the same hyper targeting. The ability to create the perfect piece of content to keep you, or your friends entertained. It can do so far cheaper, far quicker, than the most nimble human production studio; at quality levels that eventually no human will match.
The idea that humans will band together in some kind of boycott, would require us denying the entire nineteenth and twentieth century. Just about every kind of product used to be produced by hand, now we don't give it a second thought.
Once upon a time, clothing was ridiculously expensive (a shirt was estimated to cost $250 - $800), then we automated weaving via the loom - Ned Ludd famously objected, but in the end, we took the cheaper, more accessible clothing - and now we’ve got the problem we produce it so cheaply, we’ve got a surplus. Technology always wins in the long run.
In my mind, the only argument that's sound is "the tech isn't actually there yet"; it's not. But unless the trajectory rapidly and suddenly stalls, it's heading there, fast.
What do you think? Share your own thoughts in Comments, and prepare to take a "Who do you agree with?" survey next week!
Comments
You can follow this conversation by subscribing to the comment feed for this post.