AI Safety and Reproducibility: Establishing Robust Foundations for the Neuropsychology of Human Values

We propose the creation of a systematic effort to identify and replicate key findings in neuropsychology and allied fields related to understanding human values. Our aim is to ensure that research underpinning the value alignment problem of artificial intelligence has been sufficiently validated to play a role in the design of AI systems.

[1]  Brian A. Nosek,et al.  An open investigation of the reproducibility of cancer biology research , 2014, eLife.

[2]  Noah D. Goodman,et al.  Learning the Preferences of Ignorant, Inconsistent Agents , 2015, AAAI.

[3]  G. Sarma.  Doing things twice (or differently): Strategies to identify studies for targeted validation , 2018 .

[4]  Andreas Theodorou,et al.  What Does the Robot Think? Transparency as a Fundamental Design Requirement for Intelligent Systems , 2016, IJCAI 2016.

[5]  Gopal P. Sarma,et al.  Mammalian Value Systems , 2016, Informatica.

[6]  J. Stevenson.  The cultural origins of human cognition , 2001 .

[7]  Malin Björnsdotter,et al.  Vicarious Responses to Social Touch in Posterior Insular Cortex Are Tuned to Pleasant Caressing Speeds , 2011, The Journal of Neuroscience.

[8]  John Schulman,et al.  Concrete Problems in AI Safety , 2016, ArXiv.

[9]  Joseph E LeDoux,et al.  Using Neuroscience to Help Understand Fear and Anxiety: A Two-System Framework. , 2016, The American journal of psychiatry.

[10]  S. Paradiso,et al.  Book Review: Affective Neuroscience: The Foundations of Human and Animal Emotions , 2000 .

[11]  L. F. Barrett.  How Emotions Are Made: The Secret Life of the Brain , 2017 .

[12]  John P. A. Ioannidis,et al.  A manifesto for reproducible science , 2017, Nature Human Behaviour.

[13]  John P. Sullins.  When Is a Robot a Moral Agent , 2006 .

[14]  Nick Bostrom,et al.  Superintelligence: Paths, Dangers, Strategies , 2014 .

[15]  Bernice B Brown,et al.  DELPHI PROCESS: A METHODOLOGY USED FOR THE ELICITATION OF OPINIONS OF EXPERTS , 1968 .

[16]  Owain Evans.  Learning the Preferences of Bounded Agents , 2015 .

[17]  Richard Horton,et al.  Offline: What is medicine's 5 sigma? , 2015, The Lancet.

[18]  Stephen W Porges,et al.  The Early Development of the Autonomic Nervous System Provides a Neural Platform for Social Behavior: A Polyvagal Perspective. , 2011, Infant and child development.

[19]  Luciano Floridi,et al.  Transparent, explainable, and accountable AI for robotics , 2017, Science Robotics.

[20]  M. Tsakiris,et al.  ‘Bodily precision’: a predictive coding account of individual differences in interoceptive accuracy , 2016, Philosophical Transactions of the Royal Society B: Biological Sciences.

[21]  Elizabeth Gilbert,et al.  Reproducibility Project: Results (Part of symposium called "The Reproducibility Project: Estimating the Reproducibility of Psychological Science") , 2014 .

[22]  John Salvatier,et al.  When Will AI Exceed Human Performance? Evidence from AI Experts , 2017, ArXiv.

[23]  A. Damasio.  Self comes to mind : constructing the conscious brain , 2010 .

[24]  P. Ekman,et al.  Emotion in the Human Face: Guidelines for Research and an Integration of Findings , 1972 .

[25]  Mark Solms,et al.  The Brain and the Inner World: An Introduction to the Neuroscience of Subjective Experience , 2002 .

[26]  Seth D. Baum.  Reconciliation between factions focused on near-term and long-term artificial intelligence , 2017, AI & SOCIETY.

[27]  Kaj Sotala,et al.  Defining Human Values for Value Learners , 2016, AAAI Workshop: AI, Ethics, and Society.

[28]  Stuart Russell.  Should We Fear Supersmart Robots? , 2016, Scientific American.

[29]  Dalai Lama Xiv Bstan-ʾdzin-rgya-mtsho,et al.  The Dalai Lama at MIT , 2006 .