Artificial Intelligence and the Arts: Computational Creativity, Artistic Behavior, and Tools for Creatives

✍ Scribed by Penousal Machado, Juan Romero, Gary Greenfield

• Publisher: Springer
• Year: 2021
• Tongue: English
• Leaves: 383
• Series: Computational Synthesis and Creative Systems
• Edition: 1
• Category: Library

✦ Synopsis

Emotions, creativity, aesthetics, artistic behavior, divergent thoughts, and curiosity are both fundamental to the human experience and instrumental in the development of human-centered artificial intelligence systems that can relate, communicate, and understand human motivations, desires, and needs. In this book the editors put forward two core propositions: creative artistic behavior is one of the key challenges of artificial intelligence research, and computer-assisted creativity and human-centered artificial intelligence systems are the driving forces for research in this area.

The invited chapters examine computational creativity and more specifically systems that exhibit artistic behavior or can improve humans' creative and artistic abilities. The authors synthesize and reflect on current trends, identify core challenges and opportunities, and present novel contributions and applications in domains such as the visual arts, music, 3D environments, and games.

The book will be valuable for researchers, creatives, and others engaged with the relationship between artificial intelligence and the arts.

✦ Table of Contents

Preface
References
Contents
List of Contributors
Part I Visual Arts
1 Artificial Life and Artificial Intelligence Advances in the Visual Arts
1.1 Introduction
1.2 Swarm Art
1.2.1 Agent-Based Swarms
1.2.2 Ant Paintings
Abstract Ant Paintings
Representational Ant Paintings
Ant-Behavior Art
1.3 Robotics
1.3.1 Virtual Robots
1.3.2 Physical Robots
1.4 Neural Nets and Deep Learning
1.4.1 Neural Nets
1.4.2 Deep Learning
1.4.3 Artistic Analysis
1.5 Computational Aesthetics and Creativity
1.5.1 Computational Aesthetics
1.5.2 Computational Creativity
1.5.3 Knowledge-Based Agents
1.5.4 Artificial Societies
1.6 Conclusion
References
2 Aesthetics, Artificial Intelligence, and Search-Based Art
2.1 Introduction
2.2 Search, AI, and Art
2.2.1 What Drives Search in Creative Domains?
2.3 Aesthetic Theories
2.4 Imitation
2.4.1 Aesthetic Rescaling
2.4.2 Exposing Inner Workings
2.5 Skill and Expertise
2.6 Expression
2.6.1 Can Soulless Computers Express?
2.6.2 Expression Without Transmission
2.6.3 Expression, Emotion, and Expression Systems
2.6.4 Self-Contagion
2.6.5 Expression: Summary
2.7 Form
2.7.1 Form and Aesthetic Measure
2.7.2 Form and Corpus
2.7.3 Form and Multicriterion Optimisation
2.7.4 Form: Summary
2.8 Focus and Sake
2.9 Imaginative Experience
2.10 Criticism and the Artworld
2.11 Aesthetics as a Cluster Concept
2.12 Conclusions
2.12.1 Aesthetic Theories and AI Art
2.12.2 Expanding Theories for AI Art
2.12.3 Gaps and Assumptions
2.12.4 Understanding Human Art-Making
2.12.5 Future Directions
References
3 Applicability of Convolutional Neural Network Artistic Style Transfer Algorithms
3.1 Artistic Style Transfer
3.1.1 Concept
Formalization of Artistic Style Transfer
3.2 State of the Art Style Transfer
3.2.1 Analogy Style
3.2.2 Neural Style
Other Networks
Fast Style Transfer
3.2.3 Structural Semantic
3.3 Conclusion and Outlook
References
Images
Part II Music
4 Artificial Musical Intelligence: Computational Creativity in a Closed Cognitive World
4.1 Music
4.2 What Is Music?
4.3 What Are the Components of Musical Behaviour?
4.4 What Does Music Mean?
4.5 What Does the Future Hold for Arti cial Musical Intelligence?
4.6 Where Is Creativity in AMI?
References
5 Evolutionary Music, Deep Learning and Conceptual Blending: Enhancing User Involvement in Generative Music Systems
5.1 Introduction
5.2 Music, Information and Evolutionary Exploration of Alternatives
5.2.1 Symbolic Feature Extraction and Music Information
5.2.2 Generating Musical Alternatives with Given Specific Qualities
5.3 Power to the User: Exploring Variation and Human Preference in Generative Systems
5.3.1 User Involvement in Generating Variations
Variation Using Features and Distance
Towards a User-Machine Creative Loop
5.3.2 Interactive Evolution: Capturing, Assessing and Expanding User Preference
5.4 Breaking New Ground: Evolution, Machine Learning and Conceptual Blending
5.4.1 Cognitive Models of Creativity and Conceptual Blending
5.4.2 Evolutionary Computation, Machine Learning and User-Driven Conceptual Blending for Melody Generation
Melodic Representation, Learning and Generation
Explicit Melodic Features
Implicit, Deep Machine Learning and System Architecture
Example Scenario
5.5 Conclusion
References
6 Representation Learning for the Arts: A Case Study Using Variational Autoencoders for Drum Loops
6.1 Introduction
6.2 Approaches and Issues in Representation Learning
6.2.1 Dimensionality Reduction
6.2.2 Manual Design of Representations
6.2.3 Representation Learning with Neural Networks
Dimensionality
Disentangled Representations
L2 Normalisation
6.2.4 Neural Networks and Music
6.2.5 Evaluating Representation Learning
6.3 Case Study: Representation Learning for MIDI Drum Tracks
6.3.1 Data and Preprocessing
6.3.2 Proposed Model
Hyperparameters
6.3.3 Evaluation
Loss During Training
Distribution of Latent Codes
Numerical Experiments
Samples
Listening Tests and Reconstructions
Interpolations
User Interface and User Experience
Disentanglement and the Beta VAE
6.4 Conclusions
6.4.1 Future Work
References
Part III 3D
7 Case Studies in Computer Graphics and AI
7.1 Introduction
7.2 Materials and Methods
NodeBox
Pattern
7.3 Results
7.3.1 Evolution (Agent-Based + Genetic Algorithm)
7.3.2 Creatures (Procedural Generation)
7.3.3 Perception (Knowledge Representation)
7.3.4 Timeline of Paintings (Knowledge Representation)
7.3.5 Valence (Human-Computer Interaction)
7.3.6 Virtual Underwater World (Human-Computer Interaction)
7.3.7 AI Stories (NLG + ML)
7.4 Analysis
Evidence-based design
Participatory design
7.5 Conclusions
References
8 Setting the Stage for 3D Compositing with Machine Learning
8.1 Introduction
8.2 Challenges
8.2.1 Data
8.2.2 Frameworks
8.2.3 Introspection
8.2.4 Evaluation
8.3 Case Study: 3D Compositing
8.3.1 Typical Compositing Work ows
8.3.2 A Machine Learning Driven Compositing Workflow
Perspective
Environment Light
Discrete Lighting
Outdoor Lighting
Indoor Lighting
Final Steps
8.4 User Feedback
8.5 Machine Learning Lessons
8.5.1 Representation
8.5.2 Training
8.5.3 Evaluation
8.5.4 Con dence Classi cation
8.6 Conclusion
References
Part IV Other Art Forms
9 Computational Models of Narrative Creativity
9.1 Introduction
9.1.1 Overview of Chapter Organization
9.2 Relevant Concepts from Related Fields
9.2.1 Narratology
9.2.2 Natural Language Generation
9.2.3 Cognitive Science
9.3 Building Fabula: Inventing Content
9.3.1 Plot-Based Fabula Generation
9.3.2 Character-Based Fabula Generation
Modelling Character Emotions
Modelling Affnities Between Characters
Modelling Social Norms
Modelling Conflict
9.4 Constructing Discourse: Arranging the Telling of Stories
9.4.1 Systems That Generate Discourse Based on Accounts of Plot
9.4.2 Systems That Generate Discourse Based on Neural Networks
9.4.3 Emergent Discourse in Interactive Systems
9.4.4 Systems That Compose a Discourse for a Given Fabula or Storyworld
Basic Concepts of Narrative Discourse
Simple Instances of Narrative Composition
Discourse Generation Based on Focalisation
Discourse Generation Based on Chronology
Discourse Generation Based on Suspense
9.4.5 Systems That Combine Fabula and Discourse Construction
9.5 Producing Renderings for Proto-literary Drafts
9.5.1 Systems That Generate Text Directly
9.5.2 Systems That Generate Renderings for a Given Discourse
Prose
Dialogue
Visual Discourse
9.6 Modelling Cognitive Aspects of Literary Creation
9.6.1 Computational Appreciation of Proto-Literary Drafts
9.6.2 Generation Systems Based on Cognitive Accounts of Writing
9.6.3 Evolutionary Generation Systems
9.7 Conclusions
References
10 Learning from Responses to Automated Videogame Design
10.1 Introduction
10.1.1 Stakeholders
10.1.2 Background
10.2 Press
10.2.1 Replacing Humans
10.2.2 Gender and Names
10.2.3 Secondary Sources
10.3 Public
10.3.1 Expectations
Mass Production of Creative Artefacts
Fear
10.3.2 Exhibition Structure
10.4 Peers
10.4.1 Optimism
10.4.2 Peer Communication
10.5 Conclusions
10.5.1 Everything Is Context
10.5.2 Stakeholder Goals
10.5.3 Relationships
References
11 Artificial Intelligence for Designing Games
11.1 Introduction
11.2 Games as a Multi-faceted Creative Domain
11.2.1 Visuals
11.2.2 Audio
11.2.3 Narrative
11.2.4 Levels
11.2.5 Rules
11.2.6 Gameplay
11.3 Orchestration
11.4 Procedural Content Generation in Games
11.5 Arti cial Intelligence and Game Design
11.5.1 Generating Content Through AI
11.5.2 Orchestrating Game Generation
11.5.3 Cases of AI-Based Game Design
11.6 Conclusions
11.6.1 Future Directions
11.6.2 Challenges
11.6.3 Parting Words
References
Part V Artistic Perspectives
12 Bees Select Flowers, Humans Select Images: New Designs for Open-Ended Interactive Evolutionary Computation Inspired by Pollination Ecology
12.1 An Introduction to the Open-Ended Interactive Evolution of Images
12.1.1 What Is IEC?
12.1.2 What Are the Limitations of IEC?
12.1.3 What Is the Goal of IEC?
12.2 Creativity in Art and Evolution
12.2.1 Defining Creativity
12.2.2 Transformational Creativity and Painting Styles
12.2.3 Is Painting Open-Ended, or Dead?
12.2.4 Exploratory Creativity in Painting
12.2.5 Combinational Creativity
12.3 Creativity, Biological Evolution and Pollination Ecology
12.3.1 Evolution Doesn't Prefer Novelty (But Produces It All the Same)
12.3.2 Humans Prefer Novelty (But IEC Software Struggles to Produce It)
12.4 Creativity and Open-Endedness of Syles in Digital Image-Making
12.4.1 Rendering Lines with Turtle LOGO
12.4.2 Rendering Re ections, Ray-Tracing
12.4.3 Rendering Complexity, Fractals
12.4.4 Image Synthesis Techniques as Visual Styles
12.5 Open-Ended IEC
12.5.1 Are Existing IEC Implementations Open-Ended?
12.5.2 A Fundamental Diffculty of Navigating the Space of Images and Styles
12.5.3 Learning from the Bees to See Beyond the Limitations of Existing IEC
12.5.4 Insect-Pollinated Flowering Plants Form Part of a Community
12.5.5 Bees learn
12.5.6 Bees remember
12.5.7 Bee decision-making
12.5.8 Bees search collectively
12.6 Conclusion
References
13 Breaking the Black Box: Procedural Reading, Creation of Meaning, and Closure in Computational Artworks
13.1 Introduction
13.2 Computational Artworks
13.3 Subface-Surface Duality
13.4 Ergodicity
13.5 Open-Ended Works
13.6 Meaning
13.7 Closure
13.8 Closure in Computational Media
13.9 Procedural Reading
13.10 Development of a Theory of the System
13.11 Consequences of Procedural Reading
13.12 Procedural Reading and Aesthetic Pleasure
13.13 Intersubjectivity and Computational Art
References
14 Organism-Machine Hybrids
14.1 Introduction: Designing with Nature
14.2 Creating with Nature
14.2.1 The Organism as Programmable Builder
14.2.2 Trivariate Co-Creation
14.3 Classification of Approaches
14.3.1 Nano-Scale Interventions
14.3.2 Bio-Scaffolds
14.3.3 Fabricated Host Materials
14.3.4 Tropisms
14.3.5 Robotic Augmentation
14.4 Case Study: Mycelium Experiments
14.4.1 Material Explorations
14.4.2 Design Tectonics
14.4.3 Computational Simulation
14.4.4 Results
14.5 Conclusions
References