Open-Endedness is Essential for Artificial Superhuman Intelligence

arxiv:2406.04268

Original source · Paper JSON

Agents, open-endedness, AGI · 2024-06-06

Cites (6) · Cited by (4)

Complete cleaned Markdown · Source Markdown including bibliography

Read in parts

1. Part 1
2. Part 2
3. Part 3
4. Part 4
5. Part 5

Contents

• Open-Endedness is Essential for Artificial Superhuman Intelligence
• Abstract
• Keywords:
• 1 Introduction
• 2 Defining Open-Endedness
• 2.1 Formal Definition
• 2.2 Related Definitions
• 2.3 Types of Observer
• 2.4 Examples
• 3 Open-Ended Foundation Models
• 3.1 Reinforcement Learning
• 3.2 Self-Improvement
• 3.3 Task Generation
• 3.4 Evolutionary Algorithms
• 4 Achieving ASI Responsibly
• 4.1 AI Creation and Agency
• 4.2 Humans Understanding AI Creations
• 4.3 Humans Guiding AI Creation
• 4.4 Human Society Adapting
• 4.5 Emergent Risks of Open-Ended Systems
• 5 Conclusion and Outlook
• Impact Statement
• Acknowledgements
• Appendix A Illustrating Open-Endedness
• A.1 An Informal Example
• A.2 Definitional Subtleties
• Appendix B Alternative Definition
• Appendix C Further Related Work
• Instructions for reporting errors