Towards Superintelligence Alignment

Front matter

1. Title page
2. Dedication
3. Acknowledgements
4. Preface
5. Introduction
6. Executive overview
7. Current status
8. Table of contents
9. List of figures
10. List of tables

Part 1 · chapters 1-5

The Alignment Problem Reframed

Reframes alignment as a dynamical guarantee for human-correctable processes.

1. 1The Wrong Object of Alignment
2. 2From Artificial Intelligence to Artificial Civilization
3. 3Alignment as a Dynamical Guarantee
4. 4Why Fixed Values Are the Wrong Target
5. 5Assumptions, Scope, and Failure Coverage

Part 2 · chapters 6-10

Agents, Boundaries, and Real Optimizers

Develops boundary discovery: find the real optimizer, not just the model; makes incentive tests boundary-relative.

1. 6What Is an Agent Without Anthropomorphism?
2. 7Finding the Boundary
3. 8Agents That Grow, Split, and Merge
4. 9The Real Agent May Be Composite
5. 10Agency Under Strategic Opacity

Part 3 · chapters 11-14

Capability Growth and Competence

Treats capability as boundary information that can outrun task ontology.

1. 11Measuring Capability Without Task Ontology
2. 12Capability Growth Is Boundary Expansion
3. 13The Coordination Bottleneck
4. 14When Intelligence Deepens Misalignment

Part 4 · chapters 15-20

Value Bundles

Introduces value bundles: learnable geometry plus fragile tradeoffs, bearers, and adversarial measurement pressure.

1. 15Values Are Compressed Control Signals
2. 16The Value-Bundle Model
3. 17When Low Dimensionality Helps Value Learning
4. 18What Values Apply To
5. 19Tradeoffs and Bundle Geometry
6. 20Measuring and Stress-Testing Bundle Geometry

Part 5 · chapters 21-24

Goal Inference and Transport

Upgrades goal inference into transport, relating reward/CIRL-style inference to it as a special case under bundle and bearer preservation.

1. 21From Rewards to Values
2. 22The Compression Test for Intention
3. 23Has the Goal Really Survived?
4. 24When the Words Survive but the Meaning Doesn't

Part 6 · chapters 25-29

Correction Channels

Shows correction is not feedback; vector CCI is defined as a certificate and then stress-tested under adversarial pressure, relating shutdown, interruptibility, low impact, quantilization, and corrigibility to it as special cases and separations.

1. 25Correction Is a Causal Channel
2. 26Correction-Channel Integrity
3. 27Correction Channels under Adversarial Pressure
4. 28Beyond Following Instruction
5. 29Manipulation, Domestication, and False Consent

Part 7 · chapters 30-33

Successors and Continuity

Makes successor creation the central inheritance test for alignment.

1. 30Successor Creation as the Central Alignment Test
2. 31Conserved Properties Across Successors
3. 32Better Self-Modeling Can Be Worse
4. 33Certification Without Construction

Part 8 · chapters 34-38

Attractor Basins and Selection

Tracks selection, preservation conditions, correction-audit evasion, attractor theory, and conductive artifacts for pivotal-process governance.

1. 34Alignment Is Selected or Destroyed by Its Environment
2. 35Multi-Agent Superintelligence and Inferential Coupling
3. 36Parasites in the Correction System
4. 37The Alignment Attractor
5. 38Conductive Artifacts and Pivotal Processes

Part 9 · chapters 39-44

Safety Cases and Adversaries

Turns the framework into adversarial measurement, relating debate, amplification, and ELK to it as narrower subchannels.

1. 39Passive Observation Is Not Enough
2. 40Detecting Goal Laundering
3. 41Checking a System at Every Level
4. 42A Safety Case for Superintelligence Alignment
5. 43What Survives an Adversary: Verifiability and Representability
6. 44Lethality Stress Test and Open Issues

Part 10 · chapters 45-48

Civilizational Limits

Reaches the civilizational limit: preserve the value-update envelope, not a final answer.

1. 45When Value Change Is the Thing at Stake
2. 46The End of Unconscious Value Drift
3. 47Who Still Counts After Transformation
4. 48Towards Superintelligence Alignment

Appendices

1. Notation Index
2. Bridges and the Field: A Crosswalk
3. Human Institutions as Alignment Translation Guide
4. Institutional Genesis, Memory, and Decay: Historical Case Studies
5. A Worked Example: The BioShield Deployment Gate
6. Operational Glossary
7. Research Program
8. Lean Proof Spine in Mathematical Form
9. Experimental Evidence: Findings by Line

Back matter

1. Bibliography
2. Index