AI-GAs: AI-generating algorithms, an alternate paradigm for producing general artificial intelligence

arxiv:1905.10985

Original source · Paper JSON

Agents, open-endedness, AGI · 2019-05-27

Cites (4) · Cited by (7)

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
6. Part 6
7. Part 7
8. Part 8

Contents

• AI-GAs: AI-generating algorithms, an alternate paradigm for producing general artificial intelligence
• Abstract
• 1 Two approaches to producing general AI: the manual approach vs. AI-generating algorithms
• 1.1 The manual AI approach
• 1.2 A different approach: AI-generating algorithms (AI-GAs)
• 1.3 The Three Pillars required to produce an AI-GA
• 1.4 AI-GAs: A new grand challenge for AI research
• 1.5 Both the manual and AI-GA paths are worth pursuing, irrespective of which is more likely to be the fastest path to general AI
• 1.6 Which path is more likely to produce general AI first?
• 2 Research into the Three Pillars
• 2.1 The First Pillar: Meta-learning Architectures
• 2.2 The Second Pillar: Meta-learning the learning algorithms
• 2.3 The Third Pillar: Generating effective learning environments and training data
• 2.3.1 Encouraging behavioral diversity
• 2.3.2 Quality Diversity algorithms
• 2.3.3 Environment-generating quality-diversity algorithms
• 2.3.4 More expressive environment encodings
• 2.3.5 The viability of the Third Pillar
• 3 Discussion
• 4 Safety and ethical considerations
• 5 Conclusions
• Acknowledgements