Slashdot Mirror
Infinite Games?
Anonymous Coward writes "BBC is running a story on how US scientists are working on improving AI - with potential benefits for coming games.
The system, called Liquid Narrative allows to avoid scripted storylines, and finally gives us, the gamers, full freedom to do whatever we want to do. R. Michael Young, the project coordinator, says:
'Game companies are realising that story telling has a lot of potential that has not been tapped yet.'"
When technological innovation was driven by war and/or exploration? Now, it's driven by games.
... and skynet was born!
Say it ain't so! What happened? Did someone actually play a game that existed before the 3D X-TREME era and realize that games with story and gameplay emphasised over flashy graphics, T&A, and worn out franshises can be actually be good?
Quick, someone call Sony and tell them they're fucked!.
Finally computer game story lines are catching up with pen-and-paper RPGs.
Now if the graphics and audio could only improve on my imagination...
Unfortunately, I am not Wil Wheaton
... The Mimesis Project might be interesting as well. Apparently, they are using Unreal Tournament as a test-bed for the AI discussed in the article.
:-)
But I'm still at a loss why they chose UT, of all games, as a "story-telling" AI test-bed.
Beware: In C++, your friends can see your privates!
This technology sounds like adaptation to certain, shall we say, "naughtier" activities than gaming could be a possibility. ;)
They simulated a visit to the Monterey Aquarium, why not simulate, say, a visit to a secluded hamlet in Soviet Russia with Natalie Portman? Sign me up.
Even the best RPGs I've played for the PC have always felt scripted to me. You're limited in the actions you can take or the things you can say. I suppose this is a constraint of dealing with computers. . . but it's also why old-fashioned pencil-and-paper RPGs are still my favorite. You can come up with something the GM/Storyteller never thought of, pull off your idea, and see the results. Most computer RPGs stifle you at step 2.
!#@%*)anks for hanging up the phone, dear.
but that was before people had run out of ideas pertaining to AI. Today the only problems AI can solve are uninteresting ones.
Not even remotely true. AI faded from the public eye maybe, but there are literally hundreds of interesting projects that are being researched. The field never went away, it just doesn't make it into the mass media.
I'm the big fish in the big pond bitch.
I have been playing Elder Scrolls III: Morrowind since December 31, 2002 and I still can't put it down. It is really neat to have this gigantc scenario to explore and there's always many things around to do that have nothing to do with the main quest. I am positive that after a whole month playing that game I have yet to uncover 25% of the map.
If these people could expand on this concept and come up with a Morrowind model that spans across a few continents instead of one, and with maybe 3-5 main quests that are dynamically generated then it would take months to finish it. The problem is that if it takes so long to finish one game, people will buy less games. Same thing as building a car that runs like new for 10 years. The car company wants you to buy a new car every 5.
Pedro
----
The Insomniac Coder
Yep, soon the researchers will have made up enough ground to come up with *ELITE*. Greatest game ever, and totally open ended.
We ARE the peat bog soldiers.
...there's a very interesting game out there called AISLE. It's interactive fiction, and, while you only get one move per game, you can do pretty much anything that you want in that one move. While it certainly isn't infinitely playable, there's feedback for many inputs that you'd never expect.
-R
By developing this, wouldn't the game developers be shooting themselves in the foot? If the game industry is dependant on people buying newer and better games (and keeping the money flowing into their pockets), by developing a game that is "infinite" (different every time, with no end), wouldn't people just buy that one game, and stop buying others?
Ed Wedig
Graphic design services
docbrown.net
A good few years ago, I wrote an MA Thesis on videogame culture. One of the areas that I looked at was the striving towards psychological mastery in videogames through a striving for the end - telos. In psychological terms, videogames insert "the subject into a narrative in which she or he sees herself or himself projected as the hero and potential master" (Peter Buse, 'Nintendo and Telos: Will You Ever Reach the End?' Cultural Critique 34 (1996) 163-84 (p.169))
The ideas that Liquid Narrative are developing - realtime self-evolving narrative strands, reactive storytelling etc, seem to play interestingly into this notion of psychic development.
However, one question I ask is: do games need narrative at all? Games are about play - we are all home ludens. Do basketball games need narrative? The most interesting, successful and universally appealing games are those such as Tetris, where there is no end, but no story to get there either.
The Japanese style of RPG (e.g. the Final Fantasy series) treats role-playing in an entirely different way. Rather than creating your own character and playing out that role, you play the role of a predetermined character. For such games, scripted stories are very important. The whole point of the game is really to enjoy the story. Japanese RPGs boil down to basically being interactive stories.
As you say, improved AI and non-scripted stories will advance the Western style of role-playing game. However, I don't think it'll they'll have much of an impact on the traditional Japanese story-driven RPGs.
-Stephen
So right now, some people are a bit squeamish about a game like GTA:VC, because of how it sort of encourages killing lots of innocent virtual people.
But I don't think people are worried about killing the "AIs" for their own sake--the civillians are dumber than ants--but because they remind us of "real humans", and we don't want people to become casual about the lives of those.
But what if AI advances to the point where the enemy in the game is effectively self-aware? Works to defend its self-interest, understands the situation and its place in that, has an idea of the motivation of the human player and other ingame entities, etc etc....it's a long way off, but should we ever feel bad about killing 'em?
And if not, why not? Does the fact that these virtual people are likely to be trivially duplicatable inherently diminish their value as entities? (And if so, if someone could make a perfect copy of you right now, would you be more willing to get killed?)
(I think all these thought experiments are interesting, though less so if consciousness (as we commonly think of it) ends up being more or less the "benign user illusion" some materialist philosophers describe it as. But if we take that full viewpoint, we need new standards to base some of our concepts of right and wrong on.)
SO YOU'RE GOING TO DIE: The Comic for Dealing with Death
If the game lasts forever then it will probably be sold on a subscription basis.
I had an AI class with him. In one of those classes, he demoed this stuff. It was using the Unreal Tournament engine. Two demos he gave. One, two characters were put in a maze, without scripted moves and only knowledge of their immediate surroundings, and the knowledge of where they want to be and how to open doors (seemed like they had to go press and hold triggers or something, it's been awhile. Wasn't too terribly exciting by itself. This is along the lines I think when I think about traditional AI research, but doesn't strike me as very useful to a game..
The other was the user walked around an aquarium, and fish swam however they saw fit. The interesting part was the plaques that gave information about the animals. There was a database of factoids, and some rules about grammar and various languages, but no pre-written plaques. When viewed, the plaques contained a generated paragraph which presented some of the facts. The paragraph was always different every time you looked, and it could do it in several languages. This demonstrated how it could be used in an educational application, but also how it could be used to make NPC dialog more dynamic and realistic ('Times are Tough...').
The ultimate goal was to have a few stated conditions, and maybe end conditions, and allow the gamer full control over the environment, and have the story adapt to the conditions the player causes, if the story as planned to that point becomes impossible due to a players actions (say player is on an island with only one boat around, and he is expected to go to another island, but destroys the boat instead), a new story is generated on the fly. The computer adlibs. Also, if the game absolutely, positively requires that the player go to another island, some mechanisms can be put in, such as if the boat is not there, helicopter or another boat comes in and the occupants conveniently walk away from it.
He described the goal to be a fully interactive story, that is never the same twice through. A very interesting boon to RPGs as we know it. The aquarium demo at least showed promise for better NPC dialog. I don't know if they have anything to show the evolving story yet though...
XML is like violence. If it doesn't solve the problem, use more.
No AI is ever going to be a substitute for good game design and a good story. The article talks about rebuilding the game world if you kill some important ally or destroy an important object, but that's really only applicable for games like Unreal (that they showcase in the article). Unreal isn't a story; sure, it may have some story cobbled together, but Unreal and most games like it are only good for playing against other people and showcasing graphics cards. Something like Deus Ex is infinitely harder to design, because not only do you have to write a compelling story, but you also have to implement it.
Games like the Quest for Glory series were built around the theory that the player will want to be able to do just about anything, to break just about anything, and to be just about anything. They did this very well. It's not about scripting or AI that can allow the player to do anything; it's about using the story and scripting to guide the player without making them feel like they're being guided. Deus Ex is a good example. There are levels that you have to finish, so it's static in that respect, but the manner in which you finish them is completely up to you, and so you feel like you are in control, even though you're doing exactly what the designers wanted you to for most of the game.
People are easily wowed by the next generation of Unreal, and they certainly are quite impressive and expertly done. But they are also quite forgettable. When the last Quest for Glory game came out in '98, I'll bet you that most people pulled out the first four and re-did them (games from the 80's!) just so they could keep their character. Or if they didn't redo them, they had a dusty old floppy somewhere that had it.
Even if we had an AI smart enough to behave like a human, we will never have an AI smart enough to be as creative as humans can.
He plays for the users.
Ah crap. Does this mean the next version of Neverwinter Nights won't let me hide behind a rock formation and waste that dragon with my arrows? The bloody thing will actually find a way around? Damn scientists..
Trolling is a art,
The problem that 'Liquid Narrative' is addressing goes back at least to George Polti's "36 Dramatic Situations" in the year 1900. My AI faq gives infinitely more perspective than this BBC pap, on the important questions. (It's getting a little stale, but I'm currently revising the timeline with lots of rich resources.)
I dunno about you, but I get a sense of satisfaction when I finally BEAT a game. You know, complete all the missions, quests, whatever, resolve the story. I want there to be an end boss. I want to kill that boss, save the princess, save the world, whatever. I want to soak up the story that went along with it and remember it fondly, like a good movie. And then I want to get another game and experience the same. For me, gaming is like playing a good novel. Just because it's "open ended" doesn't mean it'll be good. Part of the fun (and frustration) in many games is the limitations and learning to work around them.
If you were me, you'd be good lookin'. - six string samurai
AI is most definitely not nowhere. What is true is that the expectations of most people outside the AI scene were highly unrealistic, and AI had to go through a massive period of readjustment to determine the direction to head in...
Just a recent example - Boeing just nailed the design specs of the new 777 using genetic algorithms to determine the most efficient dimensions.
AI is not what most people imagine it to be.