AI-Based Game Design Patterns

This paper proposes a model for designing games around Artificial Intelligence (AI). AI-based games put AI in the foreground of the player experience rather than in a supporting role as is often the case in many commercial games. We analyze the use of AI in a number of existing games and identify design patterns for AI in games. We propose a generative ideation technique to combine a design pattern with an AI technique or capacity to make new AI-based games. Finally, we demonstrate this technique through two examples of AI-based game prototypes created using these patterns.

Treanor, Mike and Zook, Alexander and Eladhari, Mirjam P and Togelius,
Julian and Smith, Gillian and Cook, Michael and Thompson, Tommy and
Magerko, Brian and Levine, John and Smith, Adam (2015) AI-based game
design patterns. In: Proceedings of the 10th International Conference on
the Foundations of Digital Games 2015 (FDG 2015). Society for the
Advancement of Digital Games, Santa Cruz, CA. ISBN 9780991398249 ,
This version is available at https://strathprints.strath.ac.uk/57219/
Strathprints is designed to allow users to access the research output of the University of
Strathclyde. Unless otherwise explicitly stated on the manuscript, Copyright © and Moral Rights
for the papers on this site are retained by the individual authors and/or other copyright owners.
Please check the manuscript for details of any other licences that may have been applied. You
may not engage in further distribution of the material for any profitmaking activities or any
commercial gain. You may freely distribute both the url (https://strathprints.strath.ac.uk/) and the
content of this paper for research or private study, educational, or not-for-profit purposes without
prior permission or charge.
Any correspondence concerning this service should be sent to the Strathprints administrator:
[e-mail]
The Strathprints institutional repository (https://strathprints.strath.ac.uk) is a digital archive of University of Strathclyde research
outputs. It has been developed to disseminate open access research outputs, expose data about those outputs, and enable the
management and persistent access to Strathclyde's intellectual output.

AI-Based Game Design Patterns
Mike Treanor
American University
[e-mail]
Alexander Zook
Georgia Institute of Technology
[e-mail]
Mirjam P Eladhari
Otter Play
[e-mail]
Julian Togelius
NYU Polytechnic School of
Engineering
[e-mail]
Gillian Smith
Northeastern University
[e-mail]
Michael Cook
Goldsmiths College
[e-mail]
Tommy Thompson
University of Derby
[e-mail]
Brian Magerko
Georgia Institute of Technology
[e-mail]
John Levine
University of Strathclyde
[e-mail]
Adam Smith
University of Washington
[e-mail]
ABSTRACT
This paper proposes a model for designing games around
Artificial Intelligence (AI). AI-based games put AI in the
foreground of the player experience rather than in a supporting
role as is often the case in many commercial games. We analyze
the use of AI in a number of existing games and identify design
patterns for AI in games. We propose a generative ideation
technique to combine a design pattern with an AI technique or
capacity to make new AI-based games. Finally, we demonstrate
this technique through two examples of AI-based game prototypes
created using these patterns.
Categories and Subject Descriptors
I.2.1 [Artificial Intelligence] Applications and Expert Systems Ð
Games. K.8.0 [Personal Computing] General Ð Games.
General Terms
Design.
Keywords
Game design, Artificial Intelligence, Machine Learning
1. INTRODUCTION
Almost every game features some kind of Artificial intelligence
(AI). The most common role for AI in a game is controlling the
non-player characters (NPCs), usually adversaries to the player
character. Yet this opposing AI is often rudimentary because the
design of the game does not need more complex AI. The
perception of AI as controlling adversaries in turn results in games
designed to not need richer and more varied AI. However, there
are more roles AI can play. Using AI for controlling an
adversarial NPC is one of many design patterns for how AI can be
used in games. This paper proposes a model and ideation
technique for designing games around AI, or AI-based games. A
primary goal of this work is to aid discovering new types and
potential genres of games.
This paper focuses on AI that is foregrounded in the game, as
opposed to AI that operates in the background. We define
foreground AI as agents the player notices and can reason about.
For example, AI that controls a NPC the player either interacts
with or observes for sufficient time to learn its behavior is
considered foregrounded. Meanwhile, AI that supports gameplay
such that the specifics of its behavior is not relevant to the player
is considered background AI. An example of background AI is the
NPC car behavior in Grand Theft Auto V that enables the player to
quickly speed down the road without crashing too often. Other
examples include the ÒfairerÓ random number generator in
Civilization IV that skews probabilities in the playerÕs favor, or
the NPC pathfinding systems in first person shooters. While such
background AI systems are important to gameplay and smooth the
player experience, their operation is not intended to be evident to
the player. This paper strives to advance the idea that putting AI
in the foreground can enable new types of gameplay experiences.
Accepting a broad definition of AI, games based on simulations of
physics can be considered AI-based games. For example, in Super
Mario Bros, the player must reason about how the system is going
to place the character in 2D space based on their input. While the
player may not know all of the specifics of how the game
simulates 2D physics, they build an approximate model and are
able to apply this model to predict how their input will affect the
game state while pursuing intentional acts. The physics simulation
is central to the gameplay experience.

Super Mario Bros makes use of what have been called Ògraphical
logicsÓ because the playerÕs understanding of its physics
simulation is achieved via visually represented entities moving
and interacting on a screen [14]. This can be considered
foreground AI, as the playerÕs understanding of the system is
central to how they make choices. This idea of putting a
visualization of the physics simulation central to a game suggests
that visualizations of other AI systems might make for interesting
games. For example, what types of games could be made when a
social simulation, or a learning algorithm are visualized and made
central to gameplay? We present several such design patterns for
how foreground AI can be used to make new types of games.
As part of this effort, we analyze how AI is used in several
existing games and identify design patterns for AI in games. We
propose a generative ideation technique to combine a design
pattern with an AI technique or capacity to make new AI-based
games. Finally, we demonstrate this technique through two
examples of AI-based game prototypes created using these
patterns.
2. RELATED WORK
Our argument for the value of foregrounding AI in game and our
design pattern-based taxonomy builds upon prior research in
design patterns, theoretical frameworks for understanding games,
and analysis of existing AI-based games.
A design pattern approach to describing games and game content
allows us to build a common vocabulary for discussing games,
identify common elements between games at the mechanical and
player levels, and reason about the structure of games [2, 9, 17,
19]. Design patterns are typically descriptive and informal, drawn
from a close analysis of multiple source games. The patterns
themselves typically have a short name, a description of how the
pattern is abstracted across games, and several motivating
examples to show the capacity of the pattern to describe a variety
of scenarios across multiple games. Our pattern taxonomy follows
the same model: the games we analyzed to extract our patterns
come from diverse developers, including large industry studios,
academic research, and independent development.
Though the primary purpose of design patterns is typically to
provide an analytical lens, they also have the potential to be used
generatively. Hullett and WhiteheadÕs [9] FPS level patterns were
evaluated via the deliberate design of levels that incorporate those
patterns. Dahlskog and TogeliusÕs [4] pattern-based platformer
level generator takes this one step further, formalizing the patterns
to the extent that a computer can perform the pattern-based
design. We pose that the patterns we have identified can be used
generatively during the ideation phase of design, to allow us to
consider new kinds of playable experiences.
Our taxonomy also builds on previous work in understanding the
role of AI in games and the potential it holds for the future of
games. Mateas [13] calls for the creation of Òexpressive AIÓ:
playable experiences with complex underlying AI systems where
all interaction is framed by the player needing to read meaning
into the AIÕs actions. Eladhari et al. [5] describe a process for
designing games where the AI system is an integral part of the
gameÕs design. They distill a common process followed during the
design of four games: the Pataphysic Institute [6], Prom Week
[15], Mismanor [18], and Endless Web [16]. With Endless Web,
Smith et al. pose that an AI-based game is one where the
mechanics, dynamics, and aesthetics of the game [10] are deeply
linked to the AI system. More recent games designed around their
AI system include HorswillÕs MKULTRA [8] and CookÕs A Rogue
Dream [3].
3. DESIGN PATTERNS
Below we discuss several design patterns for AI-based games.
These patterns illustrate ways to develop a game mechanic
starting from an AI technique (e.g., AI is Visualized) or starting
from an intended experience that requires AI (e.g., AI as Role-
model). The design patterns and example games are meant to be a
tool for thinking about creating AI-based games, rather than serve
as a comprehensive taxonomy of methods. Note also that multiple
techniques may apply to a single game: Table 1 provides an
overview of these patterns and game examples.
3.1 AI is Visualized
Pattern: Provide a visual representation of the underlying AI
state, making gameplay revolve around explicit manipulation of
the AI state.
Explanation: Many AI techniques revolve around an estimation
of the value of actions or game states. Typically these values are
hidden from players to promote the sense that an opposing AI
agent possesses an intelligence motivating its actions. Visualizing
the state of a system or agent enables gameplay as the system is
now exposed as a potential obstacle to player progress.
Example: Third Eye Crime [11] is a stealth game that illustrates
this pattern by visualizing the guard AI position tracking and
estimation system. Gameplay involves avoiding guards or
throwing distractions to manipulate the guardsÕ predictions of
player location. The direct visualization of AI state allows a
designer to build a game around manipulating, understanding, and
mentally modeling how the AI state changes.
3.2 AI as Role-model
Pattern: Provide one or more AI agents for the player to behave
similarly to.
Explanation: AI techniques to date often demonstrate strongly
patterned behavior that players come to predict: e.g., finite state
machines (FSMs) follow fixed routines that can often be easily
noticed. Rather than attempt to make agent behavior more
unpredictable, this pattern leverages the behavioral rigidity of a
technique to set a stage for the player to act on. Gameplay in this
pattern involves acting to mimic the behaviors of AI agents,
leading to an Òimitation gameÓ judged by an in-game system or
opposing players.
Example: Spy Party is a game where one player is a spy at a party
populated by FSM agents and the opposing player is a sniper
watching the party with a single shot to kill the spy. Gameplay for
the spy centers on the player attempting to act similarly to the
party agents while discreetly performing tasks in the environment
like planting a bug or reading a code from a book. Gameplay for
the sniper focuses on discerning the human player from AI agents
by looking for behavioral cues that differentiate the two. An
imitation game thus forces players to explicitly reason about the
processes followed by an AI technique.
3.3 AI as Trainee
Pattern: Have player actions train an AI agent to perform tasks
central to gameplay.
Explanation: Machine learning techniques revolve around
learning new behaviors using examples. By using player actions
as a source of examples an AI agent can learn to perform tasks,

with this indirect control (or automation) becoming central to
player activity in a game. Note that many paradigms for training
exist. Supervised learning requires players to explicitly provide
feedback by labeling examples as indicative of a behavior.
Unsupervised learning abstracts from examples without explicit
guidance. Reinforcement learning uses feedback about the value
of actions, rather than labels describing what an action was. Each
of these paradigms provides opportunities for different kinds of
player action in a game to indirectly control game outcomes.
Example: Black & White [1] is a god game where the player
trains a creature to act as an autonomous assistant in spatial
regions where the player cannot take direct action. The creature
learns sets of behaviors through a reward signal based on a needs
model; the creature also takes direct feedback through player
action (e.g., slapping or petting the creature after it takes actions).
Players cannot directly control the actions of the creature, but
instead rely on feedback to train the creature to perform actions
that align with the playerÕs desired strategy. By training AI agents
players are required to consider behind how an agent learns, even
without direct representation of that process.
3.4 AI is Editable
Pattern: Have the player directly change elements of an AI agent
that is central to gameplay.
Explanation: Most AI techniques have parameters (e.g., weights
in a neural network) or other data structures (e.g., nodes in a
behavior tree) that can be directly changed to manipulate agent
behavior. Typically these are created and tuned at design time and
remain fixed and hidden during the running time of a game.
Providing players with direct access to manipulate these
parameters can be the foundation for a game about player indirect
action via an AI agent. Note that direct player editing can
supplement indirect player training (e.g., as an alternate gameplay
mode or resource-gated mechanic).
Example: Galactic Arms Race [7] is a space shooter where how
the player uses different weapons evolves an underlying neural
network representation to change weapon firing behavior. Base
gameplay revolves around finding a set of firing behaviors that
together enable a player to succeed at destroying opposition
(another example of the AI as Trainee pattern). One gameplay
mode allows the player to explicitly manipulate the network
weights on weapons, allowing more precise control over the firing
patterns of the evolved weapons. This control enables players to
more finely explore the space of parameterizations, leading to an
indirect way to understand the processes of the AI system.
3.5 AI is Guided
Pattern: The player assists a simple or brittle AI agent that is
threatened with self-destruction.
Explanation: Many AI algorithms are brittle and likely to break
unless constrained to highly limited environment. Rather than
avoid exposing the AI to situations where its behavior would be
detrimental, build gameplay around the player acting to avoid
those situations. Gameplay then emphasizes players acting around
the AI to protect it or directly acting to continually maintain the
AI in the face of gradual degradation.
Example: The Sims addressed the problem of Òhuman-likeÓ
agents in a social world by making gameplay revolve around the
player addressing the needs of simple agents. AI agents have a set
of needs and desires they attempt to pursue while players
intervene to provide for the needs of the agents through food,
shelter, work, socialization, and eventually more grand life
aspirations. By having players care for the AI, players come to (at
least indirectly) model some of the processes used by the AI.
3.6 AI as Co-creator
Pattern: Involve the player in a creative task where an AI agent
directly contributes to the task as an equal partner.
Table 1. An overview of AI-based game design patterns and game examples.
Pattern What player(s) do Role of AI (in relation to player) Example(s)
AI is Visualized Observe AI state Gives (strategic) information, showing states Third Eye Crime
AI as Role-model Imitate AI Show agent actionas and behaviors, agents as
puzzles
Spy Party
AI as Trainee Teach AI Child/student Black & White
AI is Editable Edit AI Artifact/agent that player can author/manipulate Galactic Arms Race
AI is Guided Guide/manage the AI Partly independent inhabitants, with players as
their Gods
The Sims
AI as Co-creator Make artifacts assisted
by AI
Co-creator, making artifacts ViewPoints AI
AI as Adversary Play game against the
opponent
Opponent (symmetric) Chess, Go
AI as Villain Combat the Villain(s) Villain in game; mob, boss mob, NPC
(asymmetric)
Alien Isolation
AI as Spectacle Observe Spectacle, enacting simulated society Nowhere

Explanation: In games based around performance or creation of
an artifact an AI system can directly contribute to that creation.
Rather than leave all content creation in the hands of players, an
AI system can participate in content creation, creating gameplay
around the shared construction. Enabling AI participation in the
process affords gameplay around the ongoing negotiation of
meaning and goals for an artifact or performance, rather than
traditional game structures based on overcoming obstacles.
Example: Viewpoints AI [12] has a human and AI performer
create a shared movement experience through improvised, turn-
based interactions. Players are projected into a 2D plane shared
with an AI agent where the pair act and react to one anotherÕs
movements. Having players share authoring an AI agent guides
players to consider the processes being used by a system.
3.7 AI as Adversary
Pattern: Require players to overcome an (embodied or not) AI
opponent in a contest.
Explanation: This pattern is arguably one of the oldest uses of AI
in games: providing players with opponents when none (or few)
may be found. Many games depend on multiplayer competition
where it is impossible to play without a computer opponent.
Gameplay against AI adversaries revolves around understanding
the strategies and tactics the AI executes to succeed at the
competition.
Example: Computer Chess is a classic example of this pattern,
where AI agents enable people to play the game at any time and
against an opponent with adjustable capabilities (ÒdifficultyÓ).
Within global chess tournaments AI agents have had sweeping
influence: players study the strategies of AI systems and use these
to achieve high-level play. This influence has reached the point
that high-level chess success requires players to innovate in ways
that take advantage of the reasoning processes used by these AI
systems, evidencing the deep understanding of AI processes the
AI as Adversary pattern can induce.
3.8 AI as Villain
Pattern: Require players to complete a task or overcome an AI
opponent where the AI is aiming to create an experience (e.g.,
tension or excitement) rather than defeat the player.
Explanation: In games developed around players overcoming
opposition the AI agents can be Òpulling punchesÓ to intentionally
create a desired experience for the player. Rather than the AI
being a character in the game world, it is an actor attempting to
create an experience for the player while maintaining a facade of
being a character. For the player, gameplay still revolves around
defeating the opponent, yet for the opponent, gameplay revolves
around shaping player behavior in a desired way.
Example: Alien: Isolation is a first-person survival horror game
where the opposing alien was designed to harass the player
without using an optimal strategy that would always kill the
player directly. The enemy alien spends the game hunting the
player, displaying behaviors of seeking the playerÕs location (a
weak version of AI is Visualized), and gradually learning from
tactics the player uses repeatedly (an oppositional application of
AI as Trainee). By having players continually reason on what the
alien has learned and where it will go the player is forced to
consider the state of the AI and (after repeated play) the processes
involved in the AI learning.
3.9 AI as Spectacle
Pattern: Have an AI or group of AI agents implement a complex
system, such as a social hierarchy, that the player may observe or
interfere with.
Explanation: AI agent architectures are often capable of acting
and interacting fully autonomously. This pattern leverages AI
autonomy to create a game experience around watching the
unfolding of an AI society and potentially intervening in
controlled ways to observe the outcomes. For the player,
gameplay revolves around watching events unfold and
formulating ideas about how the agent society functions.
Example: Nowhere is a Ôpsychedelic RPGÕ that aims to
implement a complex society of AI agents that the player can
influence at any point in its history, and at any level in its
hierarchy. The game is designed to be confusing and alien to the
player Ð communication with the AI agents is through an Ôalien
vocabularyÕ of 27 words, so the player is forced into unfamiliar
experiences. The scale and complexity of the society the player is
faced with is part of the gameÕs design aesthetic.
4. USING PATTERNS
4.1 Generative Ideation Technique
AI is used to achieve ends: automated classification of data, object
recognition, planning, and language generation are just a few AI
capacities. We have found AI-based games can be described as a
combination of one or more patterns and one or more AI
capacities. For example, Spy Party combines the ÔAI as Role-
modelÕ pattern with AI's capacity to manage character behavior
with finite state machines. The social simulation game Prom Week
combines the AI is Visualized and the AI is Editable patterns with
the AI capacities to select actions and generate language.
We believe the space of AI-based games is vast and
underexplored: the patterns above can be used generatively to
explore possibilities for other AI-based games. Combining each
pattern with an AI capacity yields many potential games. For
instance, starting with the AI as Trainee pattern and the AI
capacity for facial expression recognition, we can imagine a game
where the player sits in front of a camera with make up, and tries
the trick the game into thinking they are someone they are not
(e.g., ÒTry to be an old angry man!Ó). This is just one of many
potential games that could come from this combination.
Below are two concrete examples of game prototypes we
developed with this approach in mind.
4.2 Examples
The design patterns we presented are largely descriptive of how
AI-based games have been created. To understand how these
patterns might be used generatively we developed two AI-based
games in short hackathon sessions. These games were used to
explore different ways of instantiating and combining some of the
patterns from the table and explore the challenges of making game
mechanics, theme, and content choices when developing AI-based
games.
4.2.1 Contrabot
In Contrabot1 players attempt to send crates of contraband
material past an inspector to a partner (see Figure 1). The playerÕs
goal is to get as many crates collected by their partner as possible,
while minimizing the crates intercepted by the inspector. To
1 code available at: https://github.com/gamesbyangelina/contrabot

achieve this goal the player stamps crates with a code indicating
to their partner that the crate should be opened. The crates, and
their associated codes, are viewed first by the inspector and then,
if the inspector does not confiscate it, the playerÕs partner. The
core AI in the game involves how agents learn the codes sent by
the player, creating gameplay around managing what the inspector
learns to check as opposed to what the partner learns to check.
The game mechanics revolve around how agents learn to check
codes based on codes they have seen. Agents have two main
processes: learning codes and matching new codes against their
learned code. To do so agents posses a memory of previously
viewed codes and a learned code that generalizes over those
remembered codes.
Learning proceeds as follows. When the inspector or partner agent
looks at a crate they add the code to their memory. The agent then
uses a simplified form of least general generalization to produce a
code that can match any of the crates. Codes take the form of a
grid of tiles all set to a binary value of white or black (0 or 1,
respectively). Learning considers each tile location: if all codes
from memory have the tile white (or black) the learned code
considers that position white (black). If the tile has been both
white and black in the agentÕs memory the tile is assigned grey,
indicating a wildcard. When new codes pass the agent they
attempt to match the new code against their learned code.
Matching checks whether every tile matches the color of the
learned code as white or black, with grey allowing either white or
black (i.e., generalization to match any option). If a code matches,
the agent adds the code to their memory.
We designed Contrabot gameplay around risk-reward
considerations for the player. To do so we started both agents with
an empty learned code and empty memory. The partner learns
from the first crate inspected and only retains a memory of the 4
most recent codes inspected. The inspector instead has a random
chance to inspect creates Ð until the inspector randomly chooses to
check a crate, the inspector will not match any crates. However,
the inspector has an infinite memory, meaning the inspector will
eventually learn a code that matches any new code (this ends the
game). After both agents have learned some code gameplay
revolves around creating codes that the inspector will not check
(modulo random checks), but that the partner will check. The
number of crates the player passes to the partner serves as a score
as the game will eventually end once the inspector learns to match
all crates.
Contrabot implements three AI-based game patterns. AI is
Visualized is implemented through showing the codes learned by
both the inspector and partner. This supports player reasoning
about what codes the agents have learned to consider new codes
to create. AI as Adversary is implemented through the inspector
agent checking crates and learning to match their codes. Players
must reason about how the inspector (and partner) learns codes,
how the inspector (and partner) match codes, and recognize the
random inspection chance for the inspector. AI is Guided is
implemented through the partner, where the player must manage
the limited partner memory and simple learning mechanism. By
providing a limited partner memory buffer we force players to
recognize the simplicity of the learning technique, particularly
under conditions with limited codes to learn from. This game is
highly extensible and customizable: with a variety of gameplay
difficulties and options that can be achieved by implementing
target numbers of received packages, limiting the number of
packages sent by the player and manipulating the size of both the
inspector and smugglers memories.
We chose these patterns as they highlight how a simple learning
technique, when visualized, can readily become the core of
gameplay. Once this is established there are many other patterns
to use to create different gameplay loops Ð in this case we took a
metaphor of code transmission and interception and used it to
choose patterns for both adversary (inspector) and teammate
(partner). Many other combinations are possible Ð below we
discuss an alternative approach where we explored visualization
of a different learning mechanism, but provided players with the
ability to both directly and indirectly manipulate the learned
information.
Figure 1. The Contrabot game interface. The leftmost tiles indicate the code the player will send, the middle tiles are the learned
code for the inspector, and the rightmost tiles are the learned code for the partner. The smaller boxes adjacent to the larger tiles
indicate the 4 most recent codes in the memory of the inspector and partner, respectively.

4.2.2 What are you doing?
In What are you doing?2 the player plays as a parent shrub in
charge of protecting and guiding one or several child shrubs. The
challenge the player faces is that the children will not do as they
are told, but instead learn from and mimic the player. This can be
positively lethal for them, however the player possesses a limited
power to edit their minds to unlearn some of the bad habits they've
learned.
The game is turn-based and plays out on a grid as shown in Figure
2. There are five types of entities in the world: the player
character (parent shrub), childshrubs, stones, strawberries, and
ponds. The parent shrub can move in any of the four cardinal
directions, eat or pick up whatever is in the direction it is facing,
or drop something it is carrying. The parent has an energy level
which can be recharged by eating. Child shrubs have mostly the
same actions available, but they have somewhat different effects:
in particular, if the child picks up stone it might be crushed under
its weight, and it will drown if moving into a pool. The energy
levels of the child shrubs are also constantly decreasing. Both
stones and ponds block the way, but stones can be picked up and
moved elsewhere. If a stone is placed into a pond both the pond
and stone disappears, so the tile becomes empty (passable).
Strawberries can be eaten by parents or kids and will increase
energy. As children will rarely seek out strawberries, the parent
can pick up strawberries and drop them on child shrubs so they
eat.
In a pane to the right of the main game area the shared mind of the
children is visualized. Their mind is initially empty, but as the
children watch the parent act, they learn rules. Rules are of the
form "when faced with certain surroundings, perform a certain
action". For example, they could learn a rule that when standing
with a stone to the left and a pond behind, move forward. Or if in
front of a strawberry, pick up. Rules are re-learned every ten
turns, and there is no limit to how many rules can be learned Ð this
2 code available at:
https://github.com/gamesbyangelina/whatareyoudoing
depends only on how many regularities the rule learning
algorithm can find within the recent history of what the parent has
done. The rule learning algorithm is a simple brute force search
for all rules that can be constructed from recent activity; if using
longer action histories or more actions, this would have to be
replaced with an a priori implementation.
The player has the option at any turn to remove one or more rules
from the mind of the children. This, however, costs energy for the
parent, so this possibility must be used sparingly. In order to
replace energy, strawberries can be eaten Ð but then the same
strawberries cannot be used to feed your children. The reason for
removing rules from the childrenÕs mind is that they would
otherwise hurt themselves, for example drowning in a pond or
picking up a stone and crushing themselves.
Tension is created in the game by scarce resources (energy and
strawberries), but also by the nature of the learning algorithm.
There could be many good courses of action which should be
avoided because your children are watching and could learn the
wrong things. For example, one might want to approach a stone
from the opposite direction than what the shortest path would
suggest, to avoid that the kids see regularities. However, this takes
more time and allows the children to learn more bad behavior in
the meantime. It is also taxing the player's memory to try to
remember what situations the parent has been in recently.
However it is done, there is a clear advantage to reasoning about
the AI, which is reinforced by the visualization of the kids' mind
and the possibility of editing it.
The child shrubs in What Are You Doing? exhibit both the AI as
Trainee pattern and the AI is Editable pattern. In the former case,
player actions are observed by the shrubs, and so the player knows
that by solving problems in the game world they are also giving
new knowledge to the AI agents. In the latter case, pruning
knowledge from the shrubs is a restricted form of editing, in that
knowledge can be explicitly removed from the agents (although
new knowledge cannot be explicitly added, only implicitly
through behaving in a certain way). Because the learned
knowledge is displayed to the player through a visualization of the
Figure 2. The What are you doing? game interface.

rules, the game also exhibits the AI is Visualized pattern, through
which the player can understand what knowledge has been picked
up by the shrubs, and then edit it appropriately to remove
dangerous behavior.
The game raises interesting problems with presenting machine
learned knowledge to the player. Our method of machine learning
learned partial rules that may not be thought of as ÔknowledgeÕ to
a human player unfamiliar with machine learning (such as what to
do when a rock is to your left, but not what to do in the general
case of being adjacent to rocks). This also means that editing
knowledge can be exhausting, since there may be multiple cases
that are all undesirable, which need to be manually removed when
to the player they may seem to all stem from the same original
case. The presentation of machine learned knowledge both in
Contrabot and What Are You Doing? presents interesting
possibilities of how to use these technique in future game projects,
and indicates the richness of new ideas to be found in applying
these patterns to new aspects of gameplay.
5. CONCLUSION
A potential critique of most of the design patterns we present is
that human players can replace the AI to produce a comparable or
improved experience. Why not have people act as adversaries or
be the targets for imitation? We are not claiming that AI agents
can pass a game-based version of the Turing test and thereby
provide new or improved play experiences. Rather, we believe
that a serious consideration of the strengths and limitations of
various AI techniques can be the foundation for new kinds of
games. By using AI as the core of gameplay experiences we can
leverage how people reason about other agents (e.g., adversaries,
ÒpeopleÓ to imitate, or creatures to raise) and create gameplay
based around thinking about how agents work. Just as much as
human puppeteers could play the roles of AI agents (consider
Jason RohrerÕs Sleep Is Death that replaces a disembodied AI as
Villain with a human), humans could also manually fill in the
physics used in many 2D platformer games. By creating
automated processes, gameplay can be based on the reliability of
these underlying systems derived from their algorithmic structure.
Designing games that use AI techniques in a new way as a core of
their gameplay diversifies and enriches the role of artificial
intelligence in games. This not only improves the breadth of the
medium with new games and genres, but also opens up new
research questions, as players begin to interact with software in
novel ways. Developing AI-based games also pushes us to tackle
existing research problems from a new, practical perspective.
Building AI agents capable of taking over from absent players in
online games or developing agents capable of assisting and
enhancing the creativity of other players are research problems
that are very relevant to the modern games industry. By building
games in which these problems are approached as a question of
game design, we can evaluate solutions directly, in contexts where
the problem is the very focus of the player activity, rather than
being one element in a much larger game. This might prove to be
a new and effective lens through which to examine other problems
in game AI research.
6. ACKNOWLEDGMENTS
The ideas in this paper were formed in a working group at the
Dagstuhl seminar 15051: Artificial and Computational
Intelligence In Games Ð Integration.
7. REFERENCES
[1] Barnes, J. 2002. Testing Undefined Behavior as a Result
of Learning. AI Game Programming Wisdom.
[2] Bjork, S. and Holopainen, J. 2004. Patterns in Game
Design (Game Development Series). Charles River
Media.
[3] Cook, M. and Colton, S. 2014. A Rogue Dream:
Automatically Generating Meaningful Content For
Games. (2014).
[4] Dahlskog, S. and Togelius, J. 2012. Patterns and
Procedural Content Generation. Proceedings of the
Workshop on Design Patterns in Games ({{DPG}}
2012), co-located with the Foundations of Digital Games
2012 conference (Raleigh, NC, May 2012).
[5] Eladhari, M.P. et al. 2011. {{AI}-Based} Game Design:
Enabling New Playable Experiences.
[6] Eladhari, M.P. 2014. The Mind Module: Using an Affect
and Personality Computational Model as a {Game-Play}
Element. Affective Computing, IEEE Transactions on. 5,
1 (Jan. 2014), 3Ð16.
[7] Hastings, E.J. and Stanley, K.O. 2010. Galactic Arms
Race: An Experiment in Evolving Video Game Content.
SIGEVOlution. 4, 4 (Mar. 2010), 2Ð10.
[8] Horswill, I. 2014. Game Design for Classical {AI}.
Tenth Artificial Intelligence and Interactive Digital
Entertainment Conference (2014).
[9] Hullett, K. and Whitehead, J. 2010. Design Patterns in
{FPS} Levels. Proceedings of the 2010 International
Conference on the Foundations of Digital Games
({{FDG}} 2010) (Monterey, CA, Jun. 2010).
[10] Hunicke, R. et al. 2004. MDA : A Formal Approach to
Game Design and Game Research. Challenges in Game
AI Workshop Nineteenth National Conference on
Artificial Intelligence (2004).
[11] Isla, D. 2013. Third Eye Crime: Building a Stealth Game
Around Occupancy Maps. Ninth Artificial Intelligence
and Interactive Digital Entertainment Conference
(2013).
[12] Magerko, B, Permar, J, Jacob, M, Comerford, M, Smith,
J. 2014. An Overview of Computational Co-creative
Pretend Play with a Human. Proceedings of the Playful
Characters workshop at the Fourteenth Annual
Conference on Intelligent Virtual Agents (2014).
[13] Mateas, M. 2007. Expressive Intelligence: Artificial
Intelligence, Games and New Media. {AI*IA} 2007:
Artificial Intelligence and {Human-Oriented}
Computing. Springer Berlin Heidelberg. 2.
[14] Mateas, M. and Wardrip-Fruin, N. 2009. Defining
Operational Logics. Digital Games Research Association
(DiGRA) (2009).

[15] McCoy, J. et al. 2011. Prom Week: Social Physics as
Gameplay. The Foundations of Digital Games
Conference (Poster) (2011).
[16] Smith, G. et al. 2012. PCG-Based Game Design:
Creating Endless Web. Foundations of Digital Games
Conference (FDG 2012) (2012).
[17] Smith, G. et al. 2011. Situating Quests: Design Patterns
for Quest and Level Design in {Role-Playing} Games.
Proceedings of the 2011 International Conference on
Interactive Digital Storytelling (Vancouver, BC, Canada,
Dec. 2011).
[18] Sullivan, A. et al. 2012. The Design of Mismanor:
Creating a Playable Quest-based Story Game.
Proceedings of the International Conference on the
Foundations of Digital Games (New York, NY, USA,
2012), 180Ð187.
[19] Zagal, J.P. and Bruckman, A. 2008. The game ontology
project: Supporting learning while contributing
authentically to game studies. Proceedings of the 8th
international conference on International conference for
the learning sciences-Volume 2 (2008), 499Ð506.