Value Alignment or Misalignment - What Will Keep Systems Accountable?

Machine learning’s advances have led to new ideas about the feasibility and importance of machine ethics keeping pace, with increasing emphasis on safety, containment, and alignment. This paper addresses a recent suggestion that inverse reinforcement learning (IRL) could be a means to so-called “value alignment.” We critically consider how such an approach can engage the social, norm-infused nature of ethical action and outline several features of ethical appraisal that go beyond simple models of behavior, including unavoidably temporal dimensions of norms and counterfactuals. We propose that a hybrid approach for computational architectures still offers the most promising avenue for machines acting in

[1]  Wendy Ju,et al.  The Design of Implicit Interactions , 2015, Synthesis Lectures on Human-Centered Informatics.

[2]  Susan M. Haack,et al.  Robot Ethics: The Ethical and Social Implications of Robotics , 2016 .

[3]  Ufuk Topcu,et al.  Robust control of uncertain Markov Decision Processes with temporal logic specifications , 2012, 2012 IEEE 51st IEEE Conference on Decision and Control (CDC).

[4]  Kenneth D. Forbus,et al.  An Integrated Reasoning Approach to Moral Decision-Making , 2008, AAAI.

[5]  James R. Cordy,et al.  A survey of grammatical inference in software engineering , 2014, Sci. Comput. Program..

[6]  Calin Belta,et al.  LTL Control in Uncertain Environments with Probabilistic Satisfaction Guarantees , 2011, ArXiv.

[7]  Patrick Lin,et al.  Robotics, Ethical Theory, and Metaethics: A Guide for the Perplexed , 2012 .

[8]  Matthias Scheutz,et al.  What to do and how to do it: Translating natural language directives into temporal and dynamic logic representation for goal management and action execution , 2009, 2009 IEEE International Conference on Robotics and Automation.

[9]  Eliezer Yudkowsky.  Artificial Intelligence as a Positive and Negative Factor in Global Risk , 2006 .

[10]  Toby Walsh.  The Singularity May Never Be Near , 2017, AI Mag..

[11]  Joanna Bryson,et al.  Patiency Is Not a Virtue: AI and the Design of Ethical Systems , 2016, AAAI Spring Symposia.

[12]  Matthias Scheutz,et al.  The Burden of Embodied Autonomy : Some Reflections on the Social and Ethical Implications of Autonomous Robots , 2007 .

[13]  L. Pereira,et al.  Counterfactuals, Logic Programming and Agent Morality , 2017 .

[14]  Kyle Kubler.  The Black Box Society: the secret algorithms that control money and information , 2016 .

[15]  Ryan Calo,et al.  There is a blind spot in AI research , 2016, Nature.

[16]  Ron Sun,et al.  Moral Judgment, Human Motivation, and Neural Networks , 2013, Cognitive Computation.

[17]  Stuart Armstrong,et al.  Motivated Value Selection for Artificial Agents , 2015, AAAI Workshop: AI and Ethics.

[18]  Michael Anderson,et al.  Machine Ethics , 2011 .

[19]  Nate Soares,et al.  The Value Learning Problem , 2018, Artificial Intelligence Safety and Security.

[20]  C. Bicchieri.  The grammar of society: the nature and dynamics of social norms , 2005 .

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

[22]  Eliezer Yudkowsky,et al.  The Ethics of Artificial Intelligence , 2014, Artificial Intelligence Safety and Security.

[23]  Matthias Scheutz,et al.  DIARC: A Testbed for Natural Human-Robot Interaction , 2006, AAAI.

[24]  Kenneth D. Forbus,et al.  Moral Decision-Making by Analogy: Generalizations versus Exemplars , 2015, AAAI.

[25]  Martin L. Puterman,et al.  Markov Decision Processes: Discrete Stochastic Dynamic Programming , 1994 .

[26]  A. Morton.  Shared Agency: A Planning Theory of Acting Together , 2015 .

[27]  Andreas Theodorou,et al.  Why is my robot behaving like that? Designing transparency for real time inspection of autonomous robots , 2016 .

[28]  James H. Moor,et al.  The Nature, Importance, and Difficulty of Machine Ethics , 2006, IEEE Intelligent Systems.

[29]  Demis Hassabis,et al.  Mastering the game of Go with deep neural networks and tree search , 2016, Nature.

[30]  Kenneth D. Forbus,et al.  Moral Decision-Making by Analogy: Generalizations versus Exemplars , 2015, AAAI.

[31]  Gina Neff,et al.  Talking to Bots: Symbiotic Agency and the Case of Tay , 2016 .

[32]  C. Allen,et al.  Artificial Morality: Top-down, Bottom-up, and Hybrid Approaches , 2005, Ethics and Information Technology.

[33]  Matthias Scheutz,et al.  Against the moral Turing test: accountable design and the moral reasoning of autonomous systems , 2016, Ethics and Information Technology.

[34]  Michael L. Littman,et al.  Reinforcement Learning as a Framework for Ethical Decision Making , 2016, AAAI Workshop: AI, Ethics, and Society.

[35]  Selmer Bringsjord,et al.  Toward a General Logicist Methodology for Engineering Ethically Correct Robots , 2006, IEEE Intelligent Systems.

[36]  Mark O. Riedl,et al.  Using Stories to Teach Human Values to Artificial Agents , 2016, AAAI Workshop: AI, Ethics, and Society.

[37]  Roman V. Yampolskiy,et al.  Artificial Intelligence Safety Engineering: Why Machine Ethics Is a Wrong Approach , 2011, PT-AI.

[38]  Stuart J. Russell,et al.  Research Priorities for Robust and Beneficial Artificial Intelligence , 2015, AI Mag..