The release of Google's Project Genie has generated a certain amount of attention, almost all of it revolving around the apparent lack of concern expressed by the tech giant for the basic concepts of intellectual property rights and copyright. The news items I've seen so far all focus on how famously defensive IP owners like Nintendo and Disney are likely to react to an app that supposedly lets anyone clone and iterate on their copyrighted properties for the low, low price of $124.99 a month.
According to IGN, some investors are already marking down stocks in other game-making platforms like Unity and Roblox, presumably on the basis that their day is done and in some bright future we'll all be our own game-makers. Or something.
All of which is twaddle, obviously. As all the reports make clear, Project Genie can't make games. It can't even make frameworks for games. Neither can it make virtual worlds.
What it can do is generate a sixty-second interactable snippet, a fuzzy glimpse into a notional, imaginary, three-dimensional environment that has neither persistence nor purpose. It's a curiosity that would probably help a few idle hours pass happily enough, a bit like making paper planes and throwing them out the window or having a tea-party with your stuffed toys.
Personally, I wouldn't pay $124.99 a month for the privilege but if it was on Steam for less than $10 I might give it a go. I imagine it'd be quite amusing to play, with for a while.
Of course, that's ignoring the legal issues. As I said, the first thing everyone seems to have tried to do is replicate their favorite video game, mostly with very limited success.
Even if the results are useless in any practical sense, though, it's hard to imagine the companies currently milking those IPs for every last cent feeling particularly sanguine about letting unlimited, bad versions loose into the community. There's such a thing as tainting the brand, after all.
Let's imagine the famous characters aren't up for grabs. You can't make your version of Mario or Sonic. You have to stick to an original idea of your own. Is anyone but you going to be interested?
One day it may indeed be possible for you or I to type a couple of paragraphs in plain English into an AI app and get a finished game out the other end. I'm close to the end of a book called Supremacy by Parmy Olson on the development of AI and if there's one thing that comes through very clearly it's that the software consistently outruns the expectations of the people behind it. One day you're crowing about some really not very impressive development you've just made and the next thing you know is someone else has jumped ahead a couple of orders of magnitude, leaving your seven-day wonder looking like yesterday's news.
If that does happen with the game-making AI, though, it's unlikely Project Genie will be the name on the box. Everything in AI seems to move forward in a weird, leapfrogging dance. Whoever starts something rarely gets to finish it.
Even if it does become possible to generate fully functional virtual worlds from plain English text prompts, the history of gaming doesn't entirely suggest that would wipe out everything else. Roblox is an outlier. A very, very big outlier, true, but an outlier all the same. There's clearly a sizeable market for making your own video games and selling them to other gamers and Roblox has mostly cornered it.
For as long as I've been gaming, which is getting on for fifty years now, there have always been game-making programs that purported to allow gamers with imagination but little or no technical skill to create and sell their own games and those have always been niche.
In the 1980s I wrote text adventures with a utility called The Quill. Lots of people did, then saved them on cassette and sold them through the back pages of gaming magazines. In the '90s I made a full-length RPG scenario with Neverwinter Nights and uploaded it to one of the dedicated, online repositories along with many thousands of others, where it probably remains to this day.
I've dabbled with many other game-making packages, none of which required anything less than a huge amount of time and effort to create anything recognizably game-like. There have even been programs to make your own MMORPG, none of which seems to have resulted in any MMORPG I ever heard of. And of course there are the semi-pro, cut-down versions of the real thing, like Unity or Unreal, that you can buy into for an almost not unreasonable investment.
If anyone with some minimal degree of skill and a very much larger amount of time and enthusiasm wanted to make a "playable world", they have and have had for a very long time, the means to do so. The appeal of AI is that soon no-one will need any appreciable skill or time to make a really impressive video game. It'd be nice to think you'd still need the enthusiasm but honestly that's probably optional too.
Except I'm not sure all that many people really want to. Imagine for a moment a scenario where Project Genie actually works. Imagine you could type in a brief description of the sort of virtual world you'd like to see, the kind of characters and plot you'd want to engage with and the general gameplay you'd enjoy there. How about...
Setting: A High Renaissance setting in which the industrial revolution happened a couple of hundred years early. Some Steampunk trappings but not too many. The game is set in a small city-state with a coastal border and some mountains, a lot of forest. Low magic, some werewolves and vampires, grudgingly integrated in the cities, not tolerated in country areas.
Plot: The high king is ageing. He has no obvious successor. He wants to leave a legacy so he makes covert moves to back a potential candidate who wishes to improve the status of the non-humans in the state. This is potentially popular with the urban elite but could cause rebellion in the countryside.
Gameplay: RPG mechanics with hierarchical magic and skill development. Dropped gear. Turn-based combat.
Sounds great, doesn't it? Took me less than ten minutes, straight off the top of my head. Imagine if I could drop that into an app, go make a coffee, then come back and download a complete game.
It'd be fun once. Just like writing a game with the Quill and NWN was fun, once. I even enjoyed playing through my own games. Again, just once.
Did I send off for anyone else's self-written Quill games, though? No, I did not. I didn't even download any of the myriad, free NWN adventures. Maybe a couple, just to have a look. But even then I never played them for more than a few minutes.
What about Second Life? All the games people made there? Did those take over the market? Not hardly.
And even the supposedly ubiquitous Roblox hasn't exactly wiped every other game off the screens, has it? Sure, a lot of people, kids mostly, play it but is it all they play and do they mostly play it because it's what they have access to and what they can afford? Like Runequest, famously the first MMORPG of so many people when they didn't have the money or the technology for anything more impressive, are these games really successful because of how good they are or how accessible and available?
It's really hard to be sure. If AI output was demonstrably not slop but really top quality, indistinguishable from very good human-made work and you could generate whatever you wanted on the fly - books, movies, games - would most people move over to it? I'm not sure they would.
I think they'd still need to be marketed to. I think there'd need to be a buzz. People like to watch, read, play things lots of other people are playing, so they can talk about them, argue about them, share the experience. At that point, it might not matter that the product was AI-generated but it would still very much matter that it was high-profile and being talked about, out there in the media.
If it was something no-one but the person watching or playing it would ever hear about, I suspect the potential audience would be somewhat limited. I think there's certainly a demographic that'd be happy to sit at home, alone, playing reading and watching the insides of their heads acted out by imaginary characters but, like VR, I just don't see it becoming normative, mainstream behavior.
In the end, I suspect that even were AI ever to reach the point where it can produce high-quality entertainment with almost no human input, it's still going to be only a certain, very small percentage of that work that gets the attention of a human audience. And it won't be for it's AI qualities but for the human-to-human interactions it facilitates.
In other words, we may all end up playing AI-generated games but we'll still be playing the same ones or else we won't be able to talk about them. And more importantly, no-one will be able to sell them to us.
Posts
No comments:
Post a Comment