AI in Digital Entertainment
Instructor: Rand Waltzman
E-mail: [email protected]
Phone: 790 6882
Room: 1430, Lindstedtsvägen 3 4 point coursePeriods I and II
Who Cares?Who Cares?
To start with, I hope that you care just To start with, I hope that you care just because you think this stuff is fun.because you think this stuff is fun.
But wait, there’s more !!!But wait, there’s more !!!
Who Cares?Who Cares?Submitted Jul 19, 2006:Senior AI Programmer LucasArts
Major Responsibilities: The Senior AI Programmer will be responsible for designing an automated system to control the behaviors, short and long term goals, and reasoning of AI This person will be tasked with implementing path finding algorithms and solutions The person in this role will create data driven, customizable, flexible, and robust code, systems, and algorithms Provide technical expertise to develop games or technologies in support of games. …Assist in the sharing of ideas and exploration of new practices to continually improve the quality of software development for the Company …
Course LiteratureCourse Literature
Variety of web sites and electronically available Variety of web sites and electronically available papers.papers.Join American Association for Artificial Join American Association for Artificial Intellilgence (www.aaai.org)Intellilgence (www.aaai.org)– Online access to numerous conference proceedings Online access to numerous conference proceedings
including “Artificial Intelligence and Interactive Digital including “Artificial Intelligence and Interactive Digital Entertainment” proceedings.Entertainment” proceedings.
– One of the best sources of materials on all aspects of One of the best sources of materials on all aspects of AI.AI.
– International Student membership - $75 (about International Student membership - $75 (about 550SEK) 550SEK)
A bargain at twice the price!!A bargain at twice the price!!
AdministriviaAdministrivia
There is There is nono tentatenta for the course! for the course!
There is a final paper.There is a final paper.– Design and analysis of some type of digital based Design and analysis of some type of digital based
entertainment that uses some type of AI technology to entertainment that uses some type of AI technology to enhance the participants experience.enhance the participants experience.
Three homework assignments.Three homework assignments.
Details of the paper and the homework Details of the paper and the homework assignments will (soon) be found on the course assignments will (soon) be found on the course web site.web site.
AdministriviaAdministrivia
Course is graded on 6 degree scale: A-F.Course is graded on 6 degree scale: A-F.Each homework is worth 150 points.Each homework is worth 150 points.The final paper is worth 250 points.The final paper is worth 250 points.Final grades will be assigned as follows:Final grades will be assigned as follows:– A: >= 600 pointsA: >= 600 points– B: 500 – 600 pointsB: 500 – 600 points– C: 400 – 500 pointsC: 400 – 500 points– D: 300 – 400 pointsD: 300 – 400 points– E: 200 – 300 pointsE: 200 – 300 points– F: < 200 points.F: < 200 points.
“The holy grail of game design is to make a game where the challenges are never ending, the skills required are varied, and the difficulty curve is perfect and adjusts itself to exactly our skill level. Someone did this already, though, and its not always fun. It’s called life. Maybe you’ve played it?”
New PossibilitiesNew Possibilities
Application of AI techniques offer potential for Application of AI techniques offer potential for new:new:– MediaMedia– Design fieldDesign field– Art formArt form
Different dimensions to consider:Different dimensions to consider:– Cognitive psychologyCognitive psychology– Computer scienceComputer science– Environmental designEnvironmental design– StorytellingStorytelling
Example 1 – Subjective AvatarsExample 1 – Subjective Avatars
Example of work in Interactive Drama. Example of work in Interactive Drama. Goal is to create a story-like experience in which Goal is to create a story-like experience in which the focus is on interactions with characters, not the focus is on interactions with characters, not on solving puzzles. on solving puzzles. Offers an opportunity for a user to experience Offers an opportunity for a user to experience the world from a new viewpoint. the world from a new viewpoint. The hope is to combine the empathic The hope is to combine the empathic understanding of a character achieved by books understanding of a character achieved by books and movies with the intensity of a first-person and movies with the intensity of a first-person interaction. interaction.
Example 1 – Subjective AvatarExample 1 – Subjective Avatar
Allow the user to step into the shoes of a Allow the user to step into the shoes of a character, experiencing a story from this new character, experiencing a story from this new perspective. perspective. Allow the user to gain an empathic Allow the user to gain an empathic understanding of a character by being this understanding of a character by being this character. character. Allow the user to gain insight into the character Allow the user to gain insight into the character s/he is playing when s/he is controlling this s/he is playing when s/he is controlling this character’s actions. character’s actions. – If s/he were to immediately begin acting out of If s/he were to immediately begin acting out of
character, s/he will derail the story, effectively character, s/he will derail the story, effectively preventing any insight. preventing any insight.
Example 2 – Office PlantExample 2 – Office Plant
Walk into a typical, high tech office environment, Walk into a typical, high tech office environment, and, among the snaking network wires, glowing and, among the snaking network wires, glowing monitors, and clicking keyboards, you are likely monitors, and clicking keyboards, you are likely to see a plant. to see a plant. The silent presence of the plant fills an The silent presence of the plant fills an emotional niche. emotional niche. Unfortunately, this plant is often dying; it is not Unfortunately, this plant is often dying; it is not adapted to the fluorescent lighting, lack of water, adapted to the fluorescent lighting, lack of water, and climate controlled air of the office. and climate controlled air of the office. Office Plant is a technological object, adapted to Office Plant is a technological object, adapted to the office ecology, which fills the same social the office ecology, which fills the same social and emotional niche as a plant. and emotional niche as a plant.
Example 2 – Office PlantExample 2 – Office Plant
Office Plant is a cross between a companion Office Plant is a cross between a companion agent and an art object that someone would agent and an art object that someone would keep in their office. keep in their office. It shares a focus on long-term engagement with It shares a focus on long-term engagement with virtual pets such as Dogz and Catz virtual pets such as Dogz and Catz However, virtual pets are intended for However, virtual pets are intended for circumscribed, high-intensity interaction. circumscribed, high-intensity interaction. In contrast, Office Plant is always on, providing a In contrast, Office Plant is always on, providing a background ambient commentary on the day’s background ambient commentary on the day’s activity. activity.
Example 3 – Terminal TimeExample 3 – Terminal Time
A machine that constructs ideologically-biased A machine that constructs ideologically-biased documentary histories in response to audience documentary histories in response to audience feedback. feedback. A cinematic experience, designed for projection on a A cinematic experience, designed for projection on a large screen in a movie theater setting. large screen in a movie theater setting. At the beginning of the show, and at several points At the beginning of the show, and at several points during the show, the audience responds to multiple during the show, the audience responds to multiple choice questions reminiscent of marketing polls. E.g., choice questions reminiscent of marketing polls. E.g., – Which of these phrases do you feel best represents you:Which of these phrases do you feel best represents you:
A. Life was better in the time of my grandparents. A. Life was better in the time of my grandparents. B. Life is good and keeps getting better every day.B. Life is good and keeps getting better every day.
Example 3 – Terminal TimeExample 3 – Terminal Time
The audience selects answers to these The audience selects answers to these questions via an applause meter – the answer questions via an applause meter – the answer generating the most applause wins. generating the most applause wins. The answers to these questions allow the The answers to these questions allow the computer program to create historical narratives computer program to create historical narratives that attempt to mirror and often exaggerate the that attempt to mirror and often exaggerate the audience’s biases and desires. audience’s biases and desires. By exaggerating the ideological position implied By exaggerating the ideological position implied in the audience’s answers, Terminal Time in the audience’s answers, Terminal Time produces not the history that they want, but the produces not the history that they want, but the history that they deserve. history that they deserve.
Example 4 – Fable 2 Example 4 – Fable 2 DogDog
Peter Molyneuax’s laws of NPC Design:Peter Molyneuax’s laws of NPC Design:
NPCs (aka AIs) in games generally should NPCs (aka AIs) in games generally should be there for your entertainment. An NPC be there for your entertainment. An NPC should:should:– Not aggravate the player. Not aggravate the player. – Focus around the player. Focus around the player. – Look after itself. Look after itself.
Example 4 – Fable 2 Example 4 – Fable 2 DogDog
The dog does this by The dog does this by – getting out of your way but still running around getting out of your way but still running around
youyou– anticipating your movement and acting as a anticipating your movement and acting as a
guideguide– not barking too much… not barking too much…
Built using a BDI (Belief, Desire, Intention) Built using a BDI (Belief, Desire, Intention) architecture.architecture.
What is Fun?What is Fun?
A source of enjoyment.A source of enjoyment.All about making the brain feel good.All about making the brain feel good.– Release of Release of endorphinsendorphins into your system. into your system.– Same sorts of chemicals released bySame sorts of chemicals released by
Listening to music we resonate to.Listening to music we resonate to.Reading a great book.Reading a great book.Snorting cocaine.Snorting cocaine.Having an orgasm.Having an orgasm.Eating chocolate.Eating chocolate.
Fun is the feedback the brain gives us when we Fun is the feedback the brain gives us when we are absorbing are absorbing patternspatterns for learning purposes. for learning purposes.
Subtle ApproachSubtle Approach
One of the subtlest releases of chemicals is at One of the subtlest releases of chemicals is at the moment of triumph when wethe moment of triumph when we– Learn something Learn something – Master a taskMaster a task– Our bodies way of rewarding usOur bodies way of rewarding us
This is one of the most important ways we find This is one of the most important ways we find pleasure in pleasure in gamesgames..In games, learning is the drug.In games, learning is the drug.Boredom is the Boredom is the oppositeopposite. . – When the game stops teaching us, we feel bored.When the game stops teaching us, we feel bored.
Experience vs. DataExperience vs. Data
New data is used to flesh out a pattern.New data is used to flesh out a pattern.
New experience might force a whole new New experience might force a whole new system on the brain.system on the brain.– Potentially disruptive and not so much fun.Potentially disruptive and not so much fun.
Games must continually navigate between Games must continually navigate between – Deprivation vs. overloadDeprivation vs. overload– Excessive chaos vs. excessive orderExcessive chaos vs. excessive order– Silence vs. noiseSilence vs. noise
How to Make a Boring GameHow to Make a Boring Game
Player figures out whole game in first 5 minutes.Player figures out whole game in first 5 minutes.
Player might see that there are incredible Player might see that there are incredible number of possible permutations.number of possible permutations.– Require mastery of a ton of Require mastery of a ton of uninterestinguninteresting details. details.
Player fails to see any pattern whatsoever.Player fails to see any pattern whatsoever.
Pacing of the revelation of variations in the Pacing of the revelation of variations in the pattern too slow.pattern too slow.– Or too fast.Or too fast.
A Little Cognitive TheoryA Little Cognitive Theory
The brain is made to fill in the blanks.The brain is made to fill in the blanks.– E.g., see a face in a bunch of cartoony lines and interpret subtle E.g., see a face in a bunch of cartoony lines and interpret subtle
emotions from them.emotions from them.– Fantastic ability to make and apply Fantastic ability to make and apply assumptionsassumptions..
The brain is good at cutting out the irrelevant.The brain is good at cutting out the irrelevant.– Show somebody a movie with a lot of jugglers in it.Show somebody a movie with a lot of jugglers in it.– Tell them in advance to count all the jugglers.Tell them in advance to count all the jugglers.– They will probably miss the large pink gorilla in the background.They will probably miss the large pink gorilla in the background.
The brain notices a lot more than we think.The brain notices a lot more than we think.– Put somebody in a hypnotic trance and ask them to describe Put somebody in a hypnotic trance and ask them to describe
something vs.something vs.– Asking them on the street!Asking them on the street!
A Little More ...A Little More ...
The brain is actively hiding the real world from The brain is actively hiding the real world from us.us.– Ask somebody to draw something.Ask somebody to draw something.– More likely to get the generalized iconic version of the More likely to get the generalized iconic version of the
object ...object ...The one they keep in their head.The one they keep in their head.
– Rather than the actual object they have in front of Rather than the actual object they have in front of them.them.
Seeing what is actually in front of us is hard.Seeing what is actually in front of us is hard.– Most of us never learn how to do it.Most of us never learn how to do it.
ChunkingChunking
Compiling an action or set of actions into a Compiling an action or set of actions into a routine.routine.– Allows us to perform the action on autopilot.Allows us to perform the action on autopilot.– Burning a recipe into the neurons.Burning a recipe into the neurons.
Example: Describe how you get to work in the Example: Describe how you get to work in the morning.morning.– Get upGet up– Stumble to the bathroomStumble to the bathroom– Take a showerTake a shower– Get dressedGet dressed– Drive to work.Drive to work.
Easy enough, but ...Easy enough, but ...
ChunkingChunking
What if I ask you to describe What if I ask you to describe oneone of these steps? of these steps?Example: Getting dressed.Example: Getting dressed.– Tops or bottoms first?Tops or bottoms first?– Socks in top or second drawer?Socks in top or second drawer?– Which pant leg goes in first?Which pant leg goes in first?– Which hand touches the button of your shirt first?Which hand touches the button of your shirt first?
You could probably answer with You could probably answer with enoughenough thoughtthought..– This operation has been chunked. This operation has been chunked. – You would have to decompile and that would take You would have to decompile and that would take
timetime..
More on Chunking ...More on Chunking ...
We usually run on chunked patterns.We usually run on chunked patterns.– Most of what we see is a chunked pattern.Most of what we see is a chunked pattern.– We rarely look at the real world.We rarely look at the real world.
We usually recognize something chunked and leave it at We usually recognize something chunked and leave it at that.that.
When something in a chunk does When something in a chunk does notnot behave as behave as we expect we have we expect we have problemsproblems..– A car starts moving sideways on a road instead of A car starts moving sideways on a road instead of
forward.forward.– We no longer have a rapid response.We no longer have a rapid response.– Unfortunately, Unfortunately, consciousconscious thought is very thought is very inefficientinefficient..– If you have to think about what you are doing, you are If you have to think about what you are doing, you are
likely to screw it up.likely to screw it up.
3 Levels of Thought3 Levels of Thought
Conscious thought.Conscious thought.– LogicalLogical– Works on a basically mathematical level.Works on a basically mathematical level.– Assigns values and makes lists.Assigns values and makes lists.– Very slow!Very slow!
Integrative, associative and intuitive.Integrative, associative and intuitive.
NonNon-thinking thought.-thinking thought.– You stick your hand in a fire.You stick your hand in a fire.– You pull it out You pull it out beforebefore you have time to you have time to thinkthink about it. about it.
Integrative ThoughtIntegrative Thought
Part of the brain that does the chunking.Part of the brain that does the chunking.
Can’t normally access this part of the brain Can’t normally access this part of the brain directly.directly.
It is frequently It is frequently wrongwrong..
It is the source of common sense.It is the source of common sense.– Often self-contradictory.Often self-contradictory.
““look before you leap”look before you leap”
““he who hesitates is lost”he who hesitates is lost”
This is where approximations of reality are built.This is where approximations of reality are built.
Appeal to Their IntelligenceAppeal to Their Intelligencess
Some basic types of intelligence that Some basic types of intelligence that entertainment can entertainment can appealappeal to:to:– LinguisticLinguistic– Logical-MathematicalLogical-Mathematical– Bodily-KinestheticBodily-Kinesthetic– SpatialSpatial– MusicalMusical– InterpersonalInterpersonal– Intrapersonal Intrapersonal
Internally directedInternally directedSelf motivatedSelf motivated
Fun is EducationalFun is Educational
Learn to calculate odds.Learn to calculate odds.– Prediction of events.Prediction of events.– Qualitative probability.Qualitative probability.
Learn about power and status.Learn about power and status.– Not surprisingly of interest since we are basically Not surprisingly of interest since we are basically
hierarchical and strongly tribal primates.hierarchical and strongly tribal primates.
Learn to examine environment or space around Learn to examine environment or space around us.us.– Spatial relationships are critically important.Spatial relationships are critically important.– Classifying, collating and exercising power over the Classifying, collating and exercising power over the
contents of space is crucial element of many games.contents of space is crucial element of many games.Using spatial relations as basis for predictive models.Using spatial relations as basis for predictive models.
Fun is Educational ...Fun is Educational ...
Learn to explore conceptual spaces.Learn to explore conceptual spaces.– Understanding rules is not enough.Understanding rules is not enough.– To exercise power over a conceptual space we need to know To exercise power over a conceptual space we need to know
how it reacts to change.how it reacts to change.– Exploring a possibility space is an excellent way to learn about it.Exploring a possibility space is an excellent way to learn about it.
Memory plays an essential role.Memory plays an essential role.E.g., recalling and managing very long and complex chains of E.g., recalling and managing very long and complex chains of information.information.
– Provide tools for exploration. But, the Provide tools for exploration. But, the tricktrick is to strike a balance is to strike a balance betweenbetween
Teaching players to rely on tools to overcome their own limitations Teaching players to rely on tools to overcome their own limitations VSVSMaking people so dependent on tools that they can’t function Making people so dependent on tools that they can’t function without them.without them.
Fun is Educational ...Fun is Educational ...
Learn basic skills:Learn basic skills:– Quick reaction time.Quick reaction time.– Tactical AwarenessTactical Awareness– Assessing the weakness of an opponent.Assessing the weakness of an opponent.– Judging when to strike.Judging when to strike.– Network building.Network building.
A very modern skill.A very modern skill.
As opposed to basic cave-man skills.As opposed to basic cave-man skills.
Good EntertainmentGood Entertainment
Thought provokingThought provokingRevelatory Revelatory – Good portrayal of human conditionGood portrayal of human condition– Provides insightProvides insight
Contributes to betterment of society.Contributes to betterment of society.Forces us to reexamine assumptions.Forces us to reexamine assumptions.Gives us different experiences each time we participate.Gives us different experiences each time we participate.Allows each of us to approach it in his/her own way.Allows each of us to approach it in his/her own way.Forgives misinterpretationsForgives misinterpretations– Maybe even encourages themMaybe even encourages them
Does not dictate.Does not dictate.Immerses and imposes a world view.Immerses and imposes a world view.
From Game to ArtFrom Game to Art
For games to reach art, the mechanics For games to reach art, the mechanics mustmust (one point (one point of view) be revelatory of the human condition.of view) be revelatory of the human condition.– Create games where the formal mechanics are about climbing a Create games where the formal mechanics are about climbing a
ladder of success.ladder of success.E.g., mechanics simulate not only the projection of power, but E.g., mechanics simulate not only the projection of power, but concepts like duty, love, honor, responsibility.concepts like duty, love, honor, responsibility.
– Create games that are about the loneliness of being at the top.Create games that are about the loneliness of being at the top.– Sample TitlesSample Titles
Hamlet: The GameHamlet: The GameWorking for the ManWorking for the ManSim GhandiSim GhandiAgainst RacisimAgainst RacisimCustody BattleCustody Battle
ExampleExample
Your goal is the overall survival of your tribe.Your goal is the overall survival of your tribe.You gain power to act based on how many people you control.You gain power to act based on how many people you control.You gain power to heal yourself based on how many friends you You gain power to heal yourself based on how many friends you havehaveFriends tend to fall away as you gain power.Friends tend to fall away as you gain power.So:So:– Being at the top and having no allies is a choice.Being at the top and having no allies is a choice.– Being lower in the status hierarchy is also a choiceBeing lower in the status hierarchy is also a choice
Perhaps more effectivePerhaps more effective
Feedback:Feedback:– Reward players for sacrificing themselves for the good of the tribe.Reward players for sacrificing themselves for the good of the tribe.– If they are captured during the game, they may no longer act directly but If they are captured during the game, they may no longer act directly but
still score points based on the actions of the players they used to rule.still score points based on the actions of the players they used to rule.– This could represent their legacy.This could represent their legacy.
What is Artificial IntelligenceWhat is Artificial Intelligence
Can Machines Have Minds?Can Machines Have Minds?
Two Types of GoalsTwo Types of Goals
AI and Computer ScienceAI and Computer Science
Examples of AI ResearchExamples of AI Research
Other AI Research AreasOther AI Research Areas
AI is AI is InherentlyInherently Multi-Disciplinary Multi-Disciplinary
Different Strokes for Different AI Different Strokes for Different AI FolksFolks
AI ProgrammingAI Programming
ACM Computing ClassificationACM Computing ClassificationI.2.0 General
Cognitive simulation Philosophical foundations
I.2.1 Applications and Expert Systems Cartography Games Industrial automation Law Medicine and science Natural language interfaces Office automation
I.2.2 Automatic Programming Automatic analysis of algorithms Program modification Program synthesis Program transformation Program verification
ACM Computing ClassificationACM Computing Classification
•I.2.3 Deduction and Theorem Proving •Answer/reason extraction •Deduction (e.g., natural, rule-based) •Inference engines •Logic programming •Mathematical induction •Metatheory •Nonmonotonic reasoning and belief revision •Resolution •Uncertainty, ``fuzzy,'' and probabilistic reasoning
ACM Computing ClassificationACM Computing Classification
•I.2.4 Knowledge Representation Formalisms and Methods •Frames and scripts •Modal logic •Predicate logic •Relation systems •Representation languages •Representations (procedural and rule-based) •Semantic networks •Temporal logic
•I.2.5 Programming Languages and Software •Expert system tools and techniques
ACM Computing ClassificationACM Computing Classification
I.2.6 Learning Analogies Concept learning Connectionism and neural nets Induction Knowledge acquisition Language acquisition Parameter learning
ACM Computing ClassificationACM Computing Classification
I.2.7 Natural Language Processing Discourse Language generation Language models Language parsing and understanding Machine translation Speech recognition and synthesis Text analysis
ACM Computing ClassificationACM Computing Classification
•I.2.8 Problem Solving, Control Methods, and Search •Backtracking •Control theory •Dynamic programming •Graph and tree search strategies •Heuristic methods •Plan execution, formation, and generation •Scheduling
ACM Computing ClassificationACM Computing Classification
•I.2.9 Robotics •Autonomous vehicles •Commercial robots and applications •Kinematics and dynamics •Manipulators •Operator interfaces •Propelling mechanisms •Sensors •Workcell organization and planning
ACM Computing ClassificationACM Computing Classification
•I.2.10 Vision and Scene Understanding •3D/stereo scene analysis •Architecture and control structures •Intensity, color, photometry, and thresholding •Modeling and recovery of physical attributes •Motion •Perceptual reasoning •Representations, data structures, and transforms •Shape •Texture •Video analysis
ACM Computing ClassificationACM Computing Classification
•I.2.11 Distributed Artificial Intelligence •Coherence and coordination •Intelligent agents •Languages and structures •Multiagent systems
Quality bars of the near-futureQuality bars of the near-future
““Emergent” behaviorsEmergent” behaviors
Believable charactersBelievable characters
100x physics100x physics
Portable avatars, persistent assetsPortable avatars, persistent assets
CommunitiesCommunities
Economies and moneyEconomies and money
Camera POV and LOD drives game-play Camera POV and LOD drives game-play
Collaborative and dynamic intelligencesCollaborative and dynamic intelligences
AI will be a “killer app” feature of AI will be a “killer app” feature of next generation of gamesnext generation of games
Characters:Characters:– AwarenessAwareness– MemoryMemory– Complex motives, simple commandsComplex motives, simple commands
Must be co-developed with animators!Must be co-developed with animators!– Game AI must be acted out and seenGame AI must be acted out and seen– Expressions & gesturesExpressions & gestures
NPC – Non-Player CharactersNPC – Non-Player Characters
““In fact, it seems the In fact, it seems the more realisticmore realistic you you make them, the more their flaws stand out make them, the more their flaws stand out and make them and make them less believableless believable. There's a . There's a difference between realistic and difference between realistic and believable. Slinging a fireball is not believable. Slinging a fireball is not realistic. But if you do it right in a fantasy realistic. But if you do it right in a fantasy game, it's entirely believable. The point is game, it's entirely believable. The point is not to make games (or characters) more not to make games (or characters) more realistic, but more believable.” realistic, but more believable.”
NPCsNPCs
Give The AI PerceptionGive The AI Perception
Let the AI recognize what's going on around it. Let the AI recognize what's going on around it.
Go beyond Go beyond justjust perceiving perceiving– Make inferences as to what is happening. Make inferences as to what is happening. – As simple as the AI noticing that the player is trying to As simple as the AI noticing that the player is trying to
steal from them. steal from them. – Or as complex as the AI trying to infer intention out of Or as complex as the AI trying to infer intention out of
the player's actions, or to try and figure out WHY the player's actions, or to try and figure out WHY certain things are happening.certain things are happening.
NPCsNPCs
Give the AI EmotionGive the AI EmotionThe The appearanceappearance of emotion. of emotion. This can be as simple as getting them This can be as simple as getting them mad at you if you try and steal from them.mad at you if you try and steal from them.Or it can be infinitely more complex. Or it can be infinitely more complex. But our brains respond so well to But our brains respond so well to perception of emotion that we will attribute perception of emotion that we will attribute emotion when none is there to inanimate emotion when none is there to inanimate objects. objects.
NPCsNPCs
Give the AI MemoryGive the AI MemoryClosely related to perception.Closely related to perception.
Giving the NPC a memory of past interactions Giving the NPC a memory of past interactions with the player is a powerful tool for making with the player is a powerful tool for making them seem more 'believable.' them seem more 'believable.'
E.g., if the AI won't trust you (emotion) because E.g., if the AI won't trust you (emotion) because it remembers (memory) that you tried to steal a it remembers (memory) that you tried to steal a pot from it (perception), you've got something pot from it (perception), you've got something close to a believable character in the game. close to a believable character in the game.
NPCsNPCs
Have the NPCs communicate with each other Have the NPCs communicate with each other about events - particularly those involving the about events - particularly those involving the player. player. It's pretty common these days in RPGs to have It's pretty common these days in RPGs to have NPC's parrot back major plot events that the NPC's parrot back major plot events that the player triggered or performed. player triggered or performed. This is cool and all, but it feels artificial.This is cool and all, but it feels artificial.But having the NPCs talk about little incidental But having the NPCs talk about little incidental gossipy stuff - perhaps even something that the gossipy stuff - perhaps even something that the player only witnessed - is even cooler. player only witnessed - is even cooler.
NPCsNPCs
Give the AI Some Kind of Personality QuirkGive the AI Some Kind of Personality QuirkThe human mind loves to categorize things, and it takes The human mind loves to categorize things, and it takes very little for a person to be pigeonholed in the minds of very little for a person to be pigeonholed in the minds of others. others. Just think back on your High School experiences for Just think back on your High School experiences for endless examples. endless examples. – One little sentence was all it could take for someone to get One little sentence was all it could take for someone to get
classified as a geek, or a snob, or a ditz. classified as a geek, or a snob, or a ditz.
If you give your NPCs just one small but noticeable If you give your NPCs just one small but noticeable quirk, it is often enough to make them stand out and for quirk, it is often enough to make them stand out and for the player to attribute all kinds of emotion and subtext to the player to attribute all kinds of emotion and subtext to conversations where none really existed. conversations where none really existed.
NPCsNPCs
The player’s perception The player’s perception is everything!is everything!
The Madden Test The Madden Test (of game AI)(of game AI)
1985: “That’s not football!”1985: “That’s not football!”
1990: “I’d fire the coach!”1990: “I’d fire the coach!”
1995: “What are those guys doing?”1995: “What are those guys doing?”
2000: “Rookie, you’re cut!”2000: “Rookie, you’re cut!”
2005: “That’s the way I designed it!”2005: “That’s the way I designed it!”
Madden Football GenesisMadden Football Genesis23%23%
Madden 97 PlaystationMadden 97 Playstation54%54%
Madden 2001 Playstation 2Madden 2001 Playstation 263%63%
Madden 2005 PS2Madden 2005 PS283%83%
When AI is applauded…When AI is applauded…
Appropriate NPC behaviorsAppropriate NPC behaviors– Dynamic adjustmentsDynamic adjustments– Satisfying mistakesSatisfying mistakes– Coordinated attacks/retreatsCoordinated attacks/retreats
Challenging opponentChallenging opponent
When AI is punished…When AI is punished…
Too easyToo easy– No coverNo cover
Too dumbToo dumb– Low awarenessLow awareness– ““Deer in headlights”Deer in headlights”
Observation #1
Maximize the ratio of perceived intelligence to
internal complexity.
SHRDLU - Winograd
-Put a red pyramid on a green block.>OK-Pick up a blue block>OK
-Men are all alike. >IN WHAT WAY?
-They're always bugging us about something or other. >CAN YOU THINK OF A SPECIFIC EXAMPLE?
-Well, my boyfriend made me come here. >YOUR BOYFRIEND MADE YOU COME HERE
-He says I'm depressed much of the time. >I AM SORRY TO HEAR YOU ARE DEPRESSED
-It's true. I am unhappy >DO YOU THINK COMING HERE WILL HELP YOU NOT TO BE UNHAPPY
Eliza - Weizenbaum
SHRDLU
-Men are all alike. >IN WHAT WAY?
-They're always bugging us about something or other. >CAN YOU THINK OF A SPECIFIC EXAMPLE?
-Well, my boyfriend made me come here. >YOUR BOYFRIEND MADE YOU COME HERE
-He says I'm depressed much of the time. >I AM SORRY TO HEAR YOU ARE DEPRESSED
-It's true. I am unhappy >DO YOU THINK COMING HERE WILL HELP YOU NOT TO BE UNHAPPY
Eliza
SHRDLU
-Men are all alike. >IN WHAT WAY?
-They're always bugging us about something or other. >CAN YOU THINK OF A SPECIFIC EXAMPLE?
-Well, my boyfriend made me come here. >YOUR BOYFRIEND MADE YOU COME HERE
-He says I'm depressed much of the time. >I AM SORRY TO HEAR YOU ARE DEPRESSED
-It's true. I am unhappy >DO YOU THINK COMING HERE WILL HELP YOU NOT TO BE UNHAPPY
Eliza
PerceivedComplexity
ActualComplexity
5
1
1
5
Observation #2
The player will build an internal model of your system.If you don’t help them build it, they’ll probably build the wrong one.
Observation #3
The flow of information about a system has a huge impact on the players perception of it’s intelligence.
Observation #4
From the players point of view there is a fine line between complex behavior and random behavior.Visibility of causal chains usually makes the difference.
Observation #5
Mimicking human intelligence and maximizing the intelligence of an artificial system are 2 very different tasks.
Observation #6
There are many applications of AI in games that don’t involve Opponents, Avatars or even human-like intelligence.
Meta AI
Peer AI
Sub AI
Meta AI
Peer AI
Sub AI Simulation
Agents
Experience
- Behavior- Opponents/Avatars- Complex Player Model
- Physics- Tactile- Intuitive Player Model
- Information Flow- Pacing- Simple Player Model
Meta
Peer
Sub
Sub
Peer
Meta
Peer
Sub
Meta
SimCity The Sims Spore
MetaPeerSub
MetaPeerSub
MetaPeerSub
Observation #7
Building a system that collects and reflects natural intelligence might be easier than replicating that intelligence.
45
Observation #8
Building a robust, internal model of the player can have huge potential value.
AI Research AI Research & IE Practice& IE Practice
IE has strong interest for systems that think, IE has strong interest for systems that think, behave and interact like people.behave and interact like people.– Autonomous agents as supporting cast roles.Autonomous agents as supporting cast roles.
Virtual Worlds:Virtual Worlds:– NPCsNPCs
Real WorldsReal Worlds– CompanionsCompanions
– CollaboratorsCollaborators
– OpponentsOpponents
Good news for AI research community.Good news for AI research community.– No simple non-AI engineering solution. No simple non-AI engineering solution.
Some Daunting ChallengesSome Daunting Challenges
Significant difference in the rate of development in AI Significant difference in the rate of development in AI and IE.and IE.– Progress in AI is slow – slower than ever.Progress in AI is slow – slower than ever.– IE experiencing explosive growth in IE experiencing explosive growth in bothboth academia academia andand industry. industry.
Slow progress of AI will Slow progress of AI will notnot keep pace with academic keep pace with academic and industrial interests.and industrial interests.E.g., autonomous virtual animated characters.E.g., autonomous virtual animated characters.– Graphics researchers have provided animated character bodies Graphics researchers have provided animated character bodies
approaching realism in visualization and animation.approaching realism in visualization and animation.– Capacities for autonomous planning, control, conversation, and Capacities for autonomous planning, control, conversation, and
interaction are barely passable for most IE applications. interaction are barely passable for most IE applications.
Industry Can’t WaitIndustry Can’t Wait
IE has had to rely on fully scripted interactions IE has had to rely on fully scripted interactions with human players to support complex with human players to support complex interactions.interactions.– Exception: Basic CombatException: Basic Combat
One approach:One approach:– Have supporting cast members played by real Have supporting cast members played by real
humans.humans.– In many ways, the rise of multiplayer and massively In many ways, the rise of multiplayer and massively
multiplayer IE forms has greatly reduced industry multiplayer IE forms has greatly reduced industry need for human-level AI.need for human-level AI.
– But in other ways it has But in other ways it has increasedincreased the need. the need.E.g., how to help orient and retain new participants in a E.g., how to help orient and retain new participants in a confusing virtual world.confusing virtual world.
Social PreferencesSocial Preferences
Interacting with other humans in a Interacting with other humans in a distributed online environment might be distributed online environment might be preferable for many.preferable for many.Result is increased interest in research in Result is increased interest in research in sociology and social psychology.sociology and social psychology.– Social network analysis.Social network analysis.– Personality profiling.Personality profiling.– Perhaps more important than the fidelity of Perhaps more important than the fidelity of
NPCs.NPCs.
Advice to AI CommunityAdvice to AI Community
Be happy that some of the pressure is Be happy that some of the pressure is being relieved!being relieved!
Broaden the scope of your expertise to Broaden the scope of your expertise to include elements of the social sciences.include elements of the social sciences.
Follow the Money!Follow the Money!
IE Industry probably has no intention of IE Industry probably has no intention of funding funding basicbasic AI research. AI research.Traditional flow of software content:Traditional flow of software content:– Small developers Small developers – Filtered through hardware manufacturers Filtered through hardware manufacturers – Large publishers.Large publishers.
None of these has incentive to support None of these has incentive to support individual basic research projects.individual basic research projects.– Not for industry-research collaboration either.Not for industry-research collaboration either.
Follow the Money!Follow the Money!
Developers probably have most to gain. But ..Developers probably have most to gain. But ..– Tight deadlines.Tight deadlines.– Slim profit margins.Slim profit margins.– Clash with academic models of Clash with academic models of high riskhigh risk investigation. investigation.
IdeasIdeas more likely to cross the divide than code. more likely to cross the divide than code.– Expect to see increased interest in academic Expect to see increased interest in academic
prototypesprototypes..– Implies importance of research funding for prototypes.Implies importance of research funding for prototypes.– Where will this funding come from?Where will this funding come from?– Wait (!!) – it is the Wait (!!) – it is the cavalry to the rescuecavalry to the rescue ... ...
Necessity is the Mother of InventionNecessity is the Mother of Invention
The military has been the most consistent source of AI The military has been the most consistent source of AI research funding throughout its entire history.research funding throughout its entire history.Increasing reliance on automation and information Increasing reliance on automation and information technology superiority.technology superiority.Steadily increasing interest in IE.Steadily increasing interest in IE.– E.g., computer game technology for militaryE.g., computer game technology for military
SimulationsSimulationsTrainingTrainingRecruitmentRecruitment
Existing Existing comfort levelcomfort level with AI research has made it with AI research has made it easier for military IE projects to have significant AI easier for military IE projects to have significant AI components.components.And the happy news is ..And the happy news is ..– the military is heavily into the tradition of the research prototype!the military is heavily into the tradition of the research prototype!
A Couple of SuggestionsA Couple of Suggestions
AI should take advantage of the reduced need AI should take advantage of the reduced need for human-level AI brought about by increased for human-level AI brought about by increased interest in multiplayer and massively multiplayer interest in multiplayer and massively multiplayer systems.systems.– Use research-grade AI systems in the automation of Use research-grade AI systems in the automation of
supporting cast member roles that most humans supporting cast member roles that most humans would not find entertaining to play.would not find entertaining to play.
Computational linguisticsComputational linguistics has been a notable has been a notable exception in the slow pace of AI research.exception in the slow pace of AI research.– Fueled by empirical and statistical methods.Fueled by empirical and statistical methods.– Few IE researchers have capitalized on the potential Few IE researchers have capitalized on the potential
offered by current technology.offered by current technology.
A Final WordA Final Word
If anything you have heard today has If anything you have heard today has upset or discouraged you in any way, upset or discouraged you in any way, remember remember The Guide’sThe Guide’s most important bit most important bit of advice:of advice:
Don’t Panic!Don’t Panic!