Publications

Academic Publications

• Robustness to fundamental uncertainty in AGI alignment, appearing in Journal of Consciousness Studies, Volume 27, Numbers 1-2, 2019, pp. 225-241(17)
  • blog post summary of paper

Technical Communications

• Comparing AI Alignment Approaches to Minimize False Positive Risk
• Goodhart’s Curse and Limitations on AI Alignment
• A developmentally-situated approach to teaching normative behavior to AI - Winner of the EthicsNet Guardians’ Challenge
• Avoiding AGI races through self-regulation (general audience version) - Runner up to the Solving the AI Race General AI Challenge
• Formally Stating the AI Alignment Problem

Research Notes

• Finding the Wisdom to Build Safe AI
• Dangers of Closed-Loop AI
• Bootstrapped Alignment
• Formal Alignment Introduction
  • Deconfusing Human Values Research Agenda v1
  • Towards deconfusing values
  • Values, Valence, and Alignment
  • Minimization of prediction error as a foundation for human values in AI alignment
• TAISU 2019 Field Report
• Let Values Drift
• HLAI 2018 Field Report
• Safety in Machine Learning
• Thoughts on “AI Safety via Debate”
• How safe “safe” AI development?
• Phenomenological AI Alignment Introduction
  • Computational Complexity of P-Zombies
  • AI Alignment and Phenomenal Consciousness
  • Towards and Axiological Approach to AI Alignment