Slashdot Mirror
Computer Scientists Invents Game-Developing Computer AI
MojoKid writes "Over the past few years, short game writing 'jams' have become a popular way to bring developers together in a conference with a single overarching theme. These competitions are typically 24-48 hours long and involve a great deal of caffeine, frantic coding, and creative design. The 28th Ludum Dare conference held from December 13 — 16 of this past year was one such game jam — but in this case, it had an unusual participant: Angelina. Angelina is a computer AI designed by Mike Cook of Goldsmiths, London University. His long-term goal is to discover whether an AI can complete tasks that are generally perceived as creative. The long-term goal is to create an AI that can 'design meaningful, intelligent and enjoyable games completely autonomously.' Angelina's entry into Ludum Dare, dubbed 'To That Sect'" is a simple 3D title that looks like it hails from the Wolfenstein era. Angelina's initial game is simple, but in reality Angelina is an AI that can understand the use of metaphor and build thematically appropriate content, which is pretty impressive. As future versions of the AI improve, the end result could be an artificial intelligence that 'understands' human storytelling in a way no species on Earth can match."
Smells like bullshit to me. What do you think?
Required reading for internet skeptics
Today, "We are gunna do this it's gunna be rad." Tomorrow, "Shit, making AIs is really fucked up."
I'd just like to see more offerings from the engine. It seems a bit similar to procedurally generated dungeons that have been around since Rogue, but with an interesting twist. Perhaps nothing groundbreaking, but kind of weird and interesting.
As Ludum Dare entries are all freely-downloadable: http://www.ludumdare.com/compo/ludum-dare-28/?action=preview&uid=29184
it's 2014 you idiot
Yeah, I was expecting something more than a level generator for a simple 3D engine.
But can an AI made today be creative? Creativity is often just a mesh of existing ideas, but how hard is that to replicate in code?
Cloudberry Kingdom, Spelunky, and many rogue-likes all do this on a smaller level, but are always constrained by parameters. While they seek to create an AI that will take on more of the tasks, it will still have to be fed parameters created by an author, so unless this AI can create itself, how can it be called truly creative? Rather it is just procedural generation. It may be worth doing but calling it creative is hyperbolic.
Twinstiq, game news
Do we have a rigorous definition of creativity?
A computer can very easily do randomness, and using fairly simple AI methods we can introduce themes and consistency.
The hard part is making something that humans will like.
Angelina's first game may be very simple, but it is conceptually quite something. For example it is interesting to read in the words of Angelina how she picked the music. So, let us not focus on how "meh" this first outing is but rather think of it as an interesting first step. By 2020 a hypotetical Angelina v6 should already be much farther evolved.
Computer Scientists Invents? Really? Nobody caught that?
Someone needs to establish a proper global AI committee to assess all these silly attempts and classify their relevance in the field. This is clearly an attempt to get headlined more than to really contribute to the field. We have yet to create properly working algorithm that aids software to initiate creative processes without human (or other) intervention. Or am I wrong here?
If it makes the front page of Slashdot in 2014, it has by definition already made the front page somewhere else in 2013.
systemd is Roko's Basilisk.
an artificial intelligence that 'understands' human storytelling in a way no species on Earth can match."
Is there any species on earth that understands human storytelling (besides humans)? I don't understand how this is a metric of success.
"First they came for the slanderers and i said nothing."
The hard part is making something that humans will like.
That's the easy part! Just make something they've seen before. Humans are idiots who like familiar things and hate unfamiliar things.
[moderate this one to flame-bait, it would be honest]
god, shut up. you obviously don't know what you read because you can't even qualify your verbs and shit or whatever. just shut up.
oh, hey, idiot: procedurally generated games have been out forever. there you go. the game was developed by "ai". fuck, what a fucktarded article.
"Stratigraphically the origin of agriculture and thermonuclear destruction will appear essentially simultaneous" -- Lee
That doesn't sound very creative.
And that somewhere else was probably also slashdot.
Slashdot: where don knuth is an idiot because he cant grasp the awesome power of php
it's 2014 you idiot
Ahh, the random little sardonic spike from an AC. What would Slashdot be without these?
Creativity does seem like a 'mesh of existing ideas', eh? It's even evident in the movies that we watch; the ones with budgets surmounting hundreds of millions. Yet, creativity is also being able to amalgamating existing ideas and building a relatively new one, in which regard it's not very different from Turing-completeness. Or is it?
Judging by the video, this looks like a random level generator for a Wolfenstein style 3D engine with largely random output. There were more useful algorithmic level generators for games already in 1984 (Elite). Not sure why this lame hack made the front page in 2013?
It almost makes me feel bad for the creator of the program as in reporting it got extremely overhyped. He could have introduced it as it is: "hey, I made this cool procedural level generator, have fun with it", and maybe gain a bunch of supportive comments from the indie gamedev community. Now it only makes him look worse as he is perceived as an overly exaggerating liar.
As future versions of the AI improve, the end result could be an artificial intelligence that "understands" human storytelling in a way no species on Earth can match.
Actually the above cannot be the end result. It is impossible to achieve with the current understanding we have.
> The hard part is making something that humans will like.
Reality TV has shown this is very easy. :)
Wow, I should not post when knackered.
Well, now that I did the unexpected and actually read the article, I have to take some of my words back. Apparently even his own goals clearly are more ambitious than just creating a level generator: to ultimately create an AI that can "design meaningful, intelligent and enjoyable games completely autonomously".
here I was thinking they could use Windows RT because MS can't seem to give it away. ;) until next galaxy note 3 acheter Galaxy note 3.
"Angelina chose the color of the walls, the textures, the ambient sound track."
fHueColour = iFloatRand(0.0f,360.0f);
Texture_Wall = iIntRand(0,10);
Music = iIntRand(0,5);
GameMode = iIntRand(0,10);
Ai, or, just basic random number generation?
Theres probably alot more under the hood. But, unless we can see how the code is written, it just looks like a random number based world generator to me.
So, AI experts in 20 or so years haven't been able to produce gaming bots which are "creative" in a multiplayer FPS
Getting this to work and to actually develop interesting game environments is going to be a pita, but if it can be done possibilities are interesting. Imagine a game that doesn't get to finish line or run out of quests etc but just keeps on evolving uniquely to match each gamers game-play. Even more interesting would be this in a MMORPG setting, imagine thousands of players affecting how game world evolves and expands. Each server might start with basic beginner world and end evolve from there, every server would be different, have different quests, maps etc. Large world maps are AI generated anyway these days, nobody goes around placing individual bushes, trees and whatnots everywhere, its more like: "this is forest area - randomly fill with vegetation type x". Applying this to other gameplay elements like quests, mob placements etc should be doable. And if you can take player actions as input this could result in very interesting results. It should be even possible to apply it to skills development.
I dunno if anyone else remembers it, but back in the 80s, Adventure Construction Set shipped with an option to generate an adventure from scratch, including the creation of new content (which, IIRC, was basically choosing some random values for things and rolling them into a new object). I obviously am not comparing the two -- this sounds considerably more advanced -- but the idea sounds the same and the results were probably about as interesting. That the AI relied on a pre-defined dictionary list of what is telling, too. Eventually, the understanding of consciousness will progress to the point where we can understand and analyze it in detail, but any AI is going to be dependent on that understanding before it is a true, complete, game production system.
I somehow enjoy reading about computer AI.
Cut him some slack. Until I read your post I too thought it was 2013. Holiday food and booze induced memory loss.
For (i = 1; i GameSize; i++) {
Character("Space Marine");
Display("Brown");
}
I believe this AI code has been in widespread use for at least 15 years.
"artificial intelligence that 'understands' human storytelling"
Could it be that the AI politician is close at hand?
what side do you want?
1. United States
2. Russia
3. United Kingdom
4. France
5. China
6. India
7. Pakistan
8. North Korea
9. Israel
"Computer ScientistS InventS Game-Developing Computer AI"
If food induces memory loss, then our obesity problem in America is two-fold!
The hard part is making something that humans will like.
No it's not - add titties. Game of the Year right there.
Now all we need is AI to play shovelware games for us.
I'm Mike, the chap behind this research. I'm glad to see a healthy dose of skepticism in the comments here! I just wanted to clear up a few points: first, I'm not claiming to have designed anything world-changing, this is just another step in the very early days of a very, very long road. Over the next few years I hope to get ANGELINA inventing game mechanics, designing graphical styles, commentating on its own developments, and producing a wider variety of games than ever before. But I hope you'll all still be asking questions and being critical, nevertheless. If you'd like to follow the project and let me know what you think of how things turn out, I blog at gamesbyangelina.org and I tweet @mtrc. Thanks for sharing the work!
This causes amazing ideas to race through my head. Imagine a much more mature version of this - one that could take input and even pull from other resources. For example, try to imagine a system that understands surfaces and textures, and even has access to the internet (or a built-in library) to derive work from. You tell the program to create a 3D castle similar to one in Scotland. It pulls information (either provided or on it's own) and from that it can create procedural textures, 3D surfaces to map those textures to, and then boom...a castle. Eventually it would understand night and day, sounds that are associated with certain environments, lighting, varied architecture, natural terrain...the possiblities are endless. And instead of having a slew of developers and artists trying to create a cohesive image over weeks or months, you have a singular entity creating this content on the fly, limited only by the amount of CPU cycles at your disposal. What an amazing science fiction story that could play itself out in reality if this idea was matured.
Judging from the game description, Angelina has already created something light years more creative than anything Michael Bay has ever done.
You mean to tell me all those shitty movie-themed games from LJN back in the 80s weren't generated by some computer? I'm shocked.
It's impressive that an AI can create a game, but that AI is coded by a person so it is always going to be restricted to what the human has created.
What if Angelina 1.0 could create Angelina 2.0 and Angelina 2.0 could create Angelina 3.0?
What would be the result of letting that run for a few years?
"Angelina chose the color of the walls, the textures, the ambient sound track" So all it did was randomly choose a few things. Completely stupid. That isn't "AI" by any stretch.
Wait, that monster *IS* my biology professor! WTF
(Ender's Game had that Giant video game which added content on its own...)
Wait, that monster trying to kill me *IS* my biology professor!
* The reference: Ender's Game had a game called Giant's Drink which used an AI to develop content tailored to the user.
I could see how games like CoD, Wolfenstien, Doom, Diablo, even side quests in D&D-style games like WoW and DDO could be generated by an AI. Go back through gaming history, though, and you'll find titles with stories SOOOO far out there that it's amazing that even a human though of it: Ultima 5,6,7,& 7 part 2, Starflight, some of the Sierra titles - specifically, Kings Quest III & IV, the Space Quest and Hero's Quest in which sarcasm and style plays an important role in the story-telling. I think an AI will have trouble with style and sarcasm. Baldur's Gate could have had an AI helping with side quests, but the main story-arc and back-story would be a stretch for AI.
Stephen Thaler's creativity machine is proof of the potential of machine creativity.
Reinvent the wheel only at either a lower cost, greater effectiveness, or your own personal enrichment and satisfaction.
http://www.pyongyangracer.co/
-- 29A the number of the Beast
Please ignore this test comment.
(Ptolemarch, 1, root-level)
Please ignore this other test comment.
(Ptolemarch, 2, reply)