Slashdot Mirror
Artificial Intelligence for Computer Games
Craig Maloney writes "Artificial Intelligence (AI) is a very hot topic today in computer circles because of the interest in modeling behaviors on machines that we find in nature. Many books have been dedicated to studying and expanding the field of AI, but generally fall into two categories: those that concentrate on AI as a research topic, and those that concentrate on AI in the field of game development. Artificial Intelligence for Computer Games (AI for Computer Games) is unique in how it takes classical AI and merges that knowledge into AI for game development. It's an approach that will be fascinating to those currently studying AI, but the approach limits the usefulness of this book to a select audience of AI researchers interested in game development." Read on for the rest of Maloney's review.
Artificial Intelligence for Computer Games
author
John David Funge
pages
127
publisher
A K Peters, td.
rating
6
reviewer
Craig Maloney
ISBN
1568812086
summary
An introduction to Gaming Artifical Intelligence
AI for Computer Games begins with a brief introduction to the historic roles that AI has played in games such as Pac Man and Mario, and how these Non-Playable Characters (NPCs) achieved fame through their roles as NPCs. The NPCs play important roles in games, and their behavior can ultimately determine if the game is entertaining or frustrating. The author then describes the differences between the field of Artificial Intelligence as compared with Gaming Artificial Intelligence. Later he shows how these two fields can intertwine with each other, and how Gaming Artificial Intelligence can be useful to AI researchers via game-playing robots and other similar experiments. The author also introduces the architecture of the components of a game. They are:
Next, AI for Computer Games discusses NPC perception. Players in a gaming environment are hindered by what the renderer will display to them, so likewise, the NPCs should not have omniscience in the game. The author recommends a strategy for handling this for NPCs: use the render engine for determining the perception of the NPCs as well. This allows the players and NPCs to work from the same rules. The author also describes how NPCs can handle partial observability, as well as prediction.
The rest of the book deals with the NPCs' abilities to react, remember, search, and learn to the game environment. This is the heart of the book, and provides a good analysis of the various methods available to the developer to model complex behaviors. The section on learning is especially interesting, as the idea of rewarding the algorithm when it performs correctly seems both strange and obvious at the same time (although the author points out that sometimes the algorithm can do undesirable things in order to obtain that reward). There are many ideas in these sections for perfecting the AI of the game, and the author expertly describes each one and where each would best be used.
AI for Computer Games was both enlightening and frustrating at the same time. The author obviously possesses a lot of knowledge in the AI field; the frustration is in his telling of that knowledge. The book reads much like an academic paper on AI applications in games, and could put off many potential readers with its rather dense descriptions of complicated material. The book also suffers from being rather short. The book is 127 pages in total length with code snippets, diagrams, and other page artwork. The brevity makes the book easy to pick up and read for a bit, but the density ensures you'll be re-reading several chapters in order to catch what the author is trying to convey. The code snippets also suffer from brevity. The code snippets are in C++, but are primarily constructors, with precious few methods defined. The author has excellent ideas; using an environment where the player and the NPCs are equals removes much of the complexity for the example AI to handle. Unfortunately the execution in this book leaves me wanting more.
You can purchase Artificial Intelligence for Computer Games from bn.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.
AI for Computer Games begins with a brief introduction to the historic roles that AI has played in games such as Pac Man and Mario, and how these Non-Playable Characters (NPCs) achieved fame through their roles as NPCs. The NPCs play important roles in games, and their behavior can ultimately determine if the game is entertaining or frustrating. The author then describes the differences between the field of Artificial Intelligence as compared with Gaming Artificial Intelligence. Later he shows how these two fields can intertwine with each other, and how Gaming Artificial Intelligence can be useful to AI researchers via game-playing robots and other similar experiments. The author also introduces the architecture of the components of a game. They are:
- Game State: The current state of the world
- Simulator: Encodes the rules for how the game state changes, and the rules for the game (physics, etc.)
- Renderer: The display of the game
- Controllers: The player and NPC methods for interacting with the game.
Next, AI for Computer Games discusses NPC perception. Players in a gaming environment are hindered by what the renderer will display to them, so likewise, the NPCs should not have omniscience in the game. The author recommends a strategy for handling this for NPCs: use the render engine for determining the perception of the NPCs as well. This allows the players and NPCs to work from the same rules. The author also describes how NPCs can handle partial observability, as well as prediction.
The rest of the book deals with the NPCs' abilities to react, remember, search, and learn to the game environment. This is the heart of the book, and provides a good analysis of the various methods available to the developer to model complex behaviors. The section on learning is especially interesting, as the idea of rewarding the algorithm when it performs correctly seems both strange and obvious at the same time (although the author points out that sometimes the algorithm can do undesirable things in order to obtain that reward). There are many ideas in these sections for perfecting the AI of the game, and the author expertly describes each one and where each would best be used.
AI for Computer Games was both enlightening and frustrating at the same time. The author obviously possesses a lot of knowledge in the AI field; the frustration is in his telling of that knowledge. The book reads much like an academic paper on AI applications in games, and could put off many potential readers with its rather dense descriptions of complicated material. The book also suffers from being rather short. The book is 127 pages in total length with code snippets, diagrams, and other page artwork. The brevity makes the book easy to pick up and read for a bit, but the density ensures you'll be re-reading several chapters in order to catch what the author is trying to convey. The code snippets also suffer from brevity. The code snippets are in C++, but are primarily constructors, with precious few methods defined. The author has excellent ideas; using an environment where the player and the NPCs are equals removes much of the complexity for the example AI to handle. Unfortunately the execution in this book leaves me wanting more.
You can purchase Artificial Intelligence for Computer Games from bn.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.
Why do we always link to just one store? Why not link to a series of various places selling the book for those who are interested?
Yeah, it's an obligatory complaint. Mod me down for it.
Haec merda tauri est. Ceterum censeo Carthaginem esse delendam.
I just finished reading a preview of Spore, a new game from the creator of the sims, and it seems like AI in games is about to take another leap :)
Natural stupidity beats artificial intelligence every time
You cannot make something idiot-proof because idiots are too ingenious (variable). Unfortunately, I see much of AI as trying to impose order on chaos which cannot be done with deterministic methods. AI _can_ help with data reduction, but not understanding.
AI has been a hot topic for, what, the past 40 years?
Much like how AI is, atleast in it's current state.
Enlightening because even the most basic attempts at simulating intelligence in machines makes us realize how vastly superior Nature's machines are. And frustrating because of how difficult it is proving for us to reach an adequately satisfactory understanding of "real" intelligence/consciousness inspite of all the research/effort we've been putting in.
An Indian-American Hindu committed to non-violent thought/speech/action alarmed by the global explosion of radical Islam
I've never really liked calling a Game's 'AI' Artificial Intelligence for one reason - they don't learn. It's always seemed more of what I'd call Simulated Intelligence. There's always a stopping point, even if they train the computer to play themselves. A point where it's not learning anymore and the computer only seems to be acting intelligent.
:p
From the review, it seems this books touches on this a bit. Hopefully more game developers will start putting additional effort into making dynamic, learning Artificial Intelligence components to their games.
Of course, part of the problem is also building the AI to act Human. Humans make mistakes, and so should the computer. In warfare, there's always been that element of random chance where you can capitalize on an enemy's mistakes. Take in factos like morale, confidence, etc. It's no fun to play against a perfect oponent all the time
I think the first game company to get this careful balance right is going to be laughing all the way to the bank.
Government's view of the economy: If it moves, tax it. If it keeps moving,regulate it. If it stops moving, subsidize it.
True artificial intelligence is only ten years away - and has been for the past three decades. AI has been a huge disappointment. Most 'AI' problems that have been solved have been solved via brute force combined with the advance of Moore's law. From what I've seen of game 'AI' - it's more a mimicing of intelligence and not very impressive mimicing at that (not much more so than the 'Eliza' class programs of the 70's).
That's my 2 cents - flame away.
[Insert pithy quote here]
We are all machines.
It is still AI though. A Neural Network stops learning at some point, but it's still AI right? It's just a difference between per learning and lazy learning.
Maybe we need some modifiers for games, Active or Static AI.
"Not knowing when the dawn will come, I open every door." - Emily Dickinson
Err ... 'evolution' results, if all goes well, in a local-maximum. Nothing about the process of evolution, real or simulation, says you'll get "the best" anything. Only that, on average, things will tend to improve in a way that matches the particular constraints at the time, according to available variations. If the constraints change, or there were several ways around a problem ("hack",) or there wasn't sufficient diversity, or a bad trait just happened to get rewarded along with a good trait, you may wind up with a terribly bad result.
This is why I get so annoyed when scientists (and creationists) ask "what is this organ useful for?" expecting that every animal's every organ is entirely well-suited to its environment -- because either evolution or the hand of god made it perfect. That's not what the theory(!) of evolution predicts. Narwhals wound up with a long tooth, and sure, maybe they use it to impress the females now -- but is that why they have it in the first place, do they maybe only use it because they have it? Could it be that somewhere along the line, some freak just happened to survive an accident when others didn't, and passed on the freak gene causing this tooth to be a horn?
Weird results from genetic algorithms are even more likely in small-population scenarios like games. You can only send so many 'test' enemies at the player before he gets bored. Particularly considering that in most games, either the player surives or the computer survives, I'm not seeing at what point you can reward the AI by letting it reproduce except when the player loses (at least in an FPS setting.) In that case, the game only gets better if the player loses a lot -- and most games try to make sure the player doesn't lose too much, but is instead always on the brink of losing (to keep him hopeful but challenged.) On the other hand, if you train them in the studio, you'll have to be careful to not train them to be good only against the testers. You don't want to release a game in which the AI is really good -- so long as you don't lure it into getting stuck in a corner, just because no tester thought to do that often enough to breed it out.
But genetic algorithms are certainly not guaranteed to produce good results. They merely might.
Artificial Intelligence (AI) is a very hot topic today in computer circles because of the interest in modeling behaviors on machines that we find in nature.
AI is not a hot topic. It hasn't recovered from the 1970s snake oil peddling stage, and it is still looked down as an overpromiser and underdeliverer of goods, even though some real neat and exciting stuff is going on there.
They'll have to fight hard to get rid of that image and that starts by continuing the shift from shoddy-but-cool-sounding work towards more measurable science. Certain areas like theorem proving have fully made the shift, others like onthologies seems still mired in the promise-lots deliver-little stage.
Doesn't part of the problem come from the fact that true intelligence is based on a life time of learning and experience?? Every time someone is faced with a decision, they subconsciously compare the situation to previous experiences to help make the correct choice.. With game A.I., the programmers are sort of trying to cram a life time of experience into so many lines of code.. Seems an exercise in futility to me.
Enlightening because even the most basic attempts at simulating intelligence in machines makes us realize how vastly superior Nature's machines are. And frustrating because of how difficult it is proving for us to reach an adequately satisfactory understanding of "real" intelligence/consciousness inspite of all the research/effort we've been putting in.
Not really. People just keep raising the bar for what counts as "intelligence", redefining it to mean: "what people can do, and machines can't".
People who could do hard computations used to be considered "smart". Until calculators were invented. Calculators aren't smart.
Chess used to be the sign of a profound thinker; someone wise and deep. Until a computer beat the world's best chess grand master; now it's just a considered a "toy problem", not "real AI".
Speech recognition used to be considered a sign of true AI. Now that we've got machines answering telephones and directing calls better than the average minimum wage immigrant secretary, we're going to start calling that "not real AI" soon, too.
Systems that could learn were considered AI. Now we've got machines that can teach themselves to walk every time they're turned on; but that's not "real AI", either.
We'll never have "real AI" until we stop moving the goalposts, and stop glorifying nature unfairly.
Let's face it, Mother Nature also made some really *stupid* creatures. Dogs that chase their tail, catch it, and then get angry and confused because something just bit them. Rabbits that try to "hide" in the middle of a highway, thinking if they crouch down on the pavement the cars can't see them. Moose that derail trains, trying out outrun them, instead of moving off the tracks.
AI's got a long way to go, granted. But it's not fair to just think of nature's geniuses when you see what she's accomplished. For every Einstein out there, there's a hundred village idiots. If an AI can outperform the idiots who wreck the curve, it may well average better performance than nature does.
For example, if 90% of car crashes are due to drunk driving, an AI driving system that replaces human drivers will be safer, *even if* it is only 50% as good at handling traffic accidents as a regular person is still safer: because it eliminates the greatest risk to traffic safety by getting the drunks off the road. Nature can't easily perform that kind of optimization. We can. We should.
So, really, are we that far from "real intelligence" in our systems? I'm not sure we are. The real problem is that a great deal of what's considered intelligence is largely a form of cultural prejudice.
Early cave men probably considered the first farmers "stupid" for not eating food when they had the chance. Instead, they saved seeds that they could have eaten, despite the fact that plants grew anyway. You can almost hear our ancestors now: "Stupid agriculture! A waste of time! Eat while you can, or you'll die! Stupid farmers!!!"
In retrospect, we can see how agriculture worked out well for us (it's one of our first successful technologies). But it's not hard to see why it could be considered "stupid" to our ancestors. So, if it's not obvious what's "stupid" and what's "smart", it's hard to say when we've build something with "intelligence", let alone "real intelligence". Intelligence often just means: "this behaviour refelects what I (or people of my culture) would do in this scenario".
... until they realize biological intelligence is the place where AI really needs to start. It seems pretty ass backward to model biological behaviour on a machine when you don't understand the mechanics of Biological or human consciousness.
My guesss is 'real' true AI, that see's the world like us and senses it like us wont be understood for a long time. Because lets face it, what AI really gives us is precise tools, to do all the jobs we cannot more precisely, but the fact is these greater functions were made by us, and are still reliant on our wetware brains for their superior formation and organization, and are only as good as we build them to be.
AI is supposed to do more and be more then just an automaton running algorithms, what I mean is, it has to self-aware environment like we are, when we are very young when we are first born we run algorithms that build some the foundation of the mind, none of us remember learning to walk or our first words within our first year or two, our self-awareness, 'we' as we experience ourselves don't wake up until between 2 and 4 year of age, this can vary somewhat depending on how fast the brain develops but the biology does most of the 'plumbing work' for us to build a foundation / mature the brain structures, to the point where enabling self-awareness makes sense.
Babies may seem alive and self-aware when they are very young but they are not, they do not experience the world at all like 6 year old, they are effectively asleep until the brain has matured to the point where self-awareness is achievable, we do not understand this process and until this is understood AI will be a wet dream if we truly want to create intelligences like human beings that are not merely machines not aware of their own existence responding in like a live human being in every respect, but not really alive. You can only really be said to be alive if you awake.
People in their sleep speak and move and do all sorts of 'living' things but they are not aware that they are doing so, it's all automatic, this is basically what most if not all AI will be like until we understand the threshold of what causes human self-awareness.
The thing is, unless you are talking about NPCs in an RPG, AI doesn't have to model all aspects of the human brain. It just has to model enough in order to act out the limited role that that character has to play in the specific game.
For instance, in a racing game, the AI of the other drivers only should know how to drive the car. You don't need to give the driver a fullblown consciousness.
So you have to scale your AI approach to the game.
It can be argued that once we know something can be programmed we stop thinking of it AI. A few years ago many would have claimed that a computer really should have achieved something intelligent when it beats a grandmaster at chess. Now, after the fact, we only think of it as a clever search routine.
There's some truth to that, but I think people want AI to be done the "right" way, like a chess program that was an outgrowth of a general purpose intelligence rather than a specialized one-trick-wonder.
If Deep Blue (or whatever the latest champion is) could also pick out a face in a variety of lighting conditions, have a half-acceptable conversation, or even learn to play a decent game of Checkers or Stratego (all using the same basic ideas that let it play chess so well not just bolt-on independent programs), then we'd have Artificial Intelligence, not just Artificial Idiot Savants.
Or if something like Cyc could LEARN to play a decent game of chess after being told the rules. (Heh...actually in some ways Atari 2600 Chess with its habit of rearranging the board (during the screenblanking "thinking" period) when under "stressful" situations almost seems more human than something like Deep Blue.)
What's also interesting is reading about how the human grandmasters deal with high end programs. Humans play chess by "chunking" the pieces on the board and applying pattern recognition. Similarly, humans can kind of "chunk" the strategies that specific AIs fall into, and try to counter them, even though it's still damned hard to do so. Conversely, I'm not sure if we have a serious AI that could even take things to the metalevel like that.
SO YOU'RE GOING TO DIE: The Comic for Dealing with Death