Decision Theory and the Future of AI

This book appeals to students, researchers and professionals working in philosophy and related fields on decision theory applied to artificial intelligence. These chapters stem from the topical conference series, 'Decision Theory and the Future of AI' which began in 2017 as a collaboration between the Leverhulme Centre for the Future of Intelligence (CFI) and the Centre for the Study of Existential Risk (CSER) at Cambridge, and the Munich Center for Mathematical Philosophy (MCMP) at LMU Munich. The range of topics, and even more so the range of authors and their home disciplines and affiliations, are a tribute to the richness of the territory, both in intellectual and in community-building terms.
Previously published in Synthese Volume 198, supplement issue 27, November 2021

Chapter Approval-directed agency and the decision theory of Newcomb-like problems is available open access under a Creative CommonsAttribution 4.0 International License via link.springer.com.

Reveals how thinking on the future of AI needs contributions across disciplines Shows how decision theory is ideally placed to make contributions on multiple levels Studies the standards by which AI decisions are evaluated and improved

Auteur

Dr. Yang Liu Upon completion of a Master s degree at Chinese Academy of Sciences and Yangzhou University, Liu pursued a PhD degree at the University of British Columbia. During my PhD, he joined Sorbonne University for one year to construct a theoretical model. Since 2017, he has begun postdoc research at the University of British Columbia. By employing multidisciplinary approaches, his long-term research goal seeks to reveal a role for demographic, ecological, and genetic mechanisms in plant adaptive evolution, and how adaptive and nonadaptive evolutionary processes have contributed to biodiversification.

Stephan Hartmann completed a Diploma in Physics (1991), a Master in Philosophy (1991), and a PhD degree in Philosophy (1995), each at Justus-Liebig University Giessen. In autumn 2012 he became Chair of Philosophy of Science, Alexander von Humboldt Professor, and Head of the Munich Center for Mathematical Philosophy at the LMU Munich. Before that he taught at Tilburg University, the London School of Economics, and the University of Konstanz. He had visiting appointments at the University of California at Irvine and Lund University and was a Visiting Fellow at the Center for Philosophy of Science at the University of Pittsburgh. He was President of the European Philosophy of Science Association (EPSA, 2013-2017) and of the European Society for Analytical Philosophy (ESAP, 2014-2017).

Huw Price retired in 2020 after nine years as Bertrand Russell Professor of Philosophy and a Fellow of Trinity College at the University of Cambridge. He was Academic Director of the Leverhulme Centre for the Future of Intelligence from its launch in 2016 until October 2021, and he remains Chair of the CFI Strategy Group. He was also co-founder, with Martin Rees and Jaan Tallinn, of the Centre for the Study of Existential Risk. In January 2019 he joined the inaugural Board of the new Ada Lovelace Institute. Before moving to Cambridge in 2011 he was ARC Federation Fellow and Challis Professor of Philosophy at the University of Sydney, where he was founding Director of the Centre for Time.

Contenu
Editorial to Decision theory and the future of AI.- A classification of Newcomb problems and decision theories.- Reward tampering problems and solutions in reinforcement learning: a causal influence diagram perspective.- Intuition, intelligence, data compression.- Approval-directed agency and the decision theory of Newcomb-like problems.- Causal concepts and temporal ordering.- Desirability foundations of robust rational decision making.- Subjective causal networks and indeterminate suppositional credences.