List of AI Alignment Resources

Table of Contents

Book List

• The Age of AI (And Our Human Future) by Henry A. Kissinger, Eric Schmidt, and Daniel Huttenlocher

• Life 3.0 by Max Tegmark (signed copy)

• Our Mathematical Universe by Max Tegmark (not directly about AI)

• Superintelligence by Nick Bostrom

• I, Robot by Isaac Asimov

• Smarter Than Us: The Rise of Machine Intelligence by Stuart Armstrong (short read)

• Minds, Machines, and the Multiverse by Julian Brown

• Blondie24: Playing at the Edge of AI by David B. Fogel

• The Quest for Artificial Intelligence by Nils J. Nilsson

• Rationality: From AI to Zombies by Eliezer Yudkowsky

• Human Compatible: Artificial Intelligence and the Problem of Control by Stuart Russell

• The Self-Assembling Brain by Peter Robin Hiesinger

• Invention and Innovation: A Brief History of Hype and Failure by Vaclav Smil

Groups

• Machine Intelligence Research Institute (MIRI): a research non-profit developing formal tools for design and analysis of aligned general-purpose AI

• Harvard AI Safety Team (HAIST)

• Center for Human-compatible AI (CHAI) at UC Berkeley

• Berkeley Existential Risk Initiative

• Center for the Study of Existential Risk at the University of Cambridge

• David Krueger’s Lab at the University of Cambridge

• Future of Humanity Institute at the University of Oxford

• Stanford Existential Risks Initiative (SERI) at Stanford University

• Human-aligned AI Summer School: annual conference for AI alignment

• Redwood Research

Papers

• Attention Is All You Need by Vaswani et al., the original paper introducing the transformers architecture.

• QNRs: Toward Language for Intelligent Machines by K. Eric Drexler: quasilinguistic neural representations can upgrade natural language to embeddings which are more expressive and computationally tractable

• Language Models are Few-Shot Learners by Brown et al., the original paper introducing GPT-3, a 175 billion parameter model. “Scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches.”

• PaLM: Scaling Language Modeling with Pathways by Chowdhery et al., the 540 billion parameter Pathways Language Model (PaLM) from Google Research. The Pathways system enables efficiently training a single model across multiple TPU v4 pods. State-of-the-art few-shot performance across most language understanding + generation tasks.

• Why does deep and cheap learning work so well? by Lin, Tegmark, and Rolnick: NNs can approximate arbitrary functions, but the class of functions of practical interest can frequently be approximated through “cheap learning” with exponentially fewer parameters than generic functions.

Internet

• Industrial Society and Its Future by Theodore Kaczynski

• Papers on AI and Computation by David Chalmers

• The Google engineer who thinks the company’s AI has come to life: Google engineer Blake Lemoine claims that LaMDA is sentient.

• Interactive proof that NNs can compute any function by Michael Nielson

• Wikipedia
• The Mind-Body Problem: a philosophical debate concerning the relationship between thought and consciousness in the human mind, and the brain as part of the physical body.
• Orchestrated Objective Reduction: a theory introduced by Roger Penrose and Stuart Hameroff that consciousness arises from quantum effects in cellular structures called microtubules.
• AIXI: a theoretical mathematical formalism for artificial general intelligence reinforcement learning agent.
• Panpsychism: the view that a mindlike aspect is fundamental to reality, and everything in the universe has varying degrees of consciousness. Philip Goff on Lex Fridman
• Limits of Computation
• Bremermann's limit: a limit on the maximum rate of computation that can be achieved in a self-contained system in the material universe.