Chess as a Testing Grounds for the Oracle Approach to AI Safety

no code implementations • 6 Oct 2020 • James D. Miller, Roman Yampolskiy, Olle Haggstrom, Stuart Armstrong

To reduce the danger of powerful super-intelligent AIs, we might make the first such AIs oracles that can only send and receive messages.

BIG-bench Machine Learning

An AGI Modifying Its Utility Function in Violation of the Orthogonality Thesis

no code implementations • 31 Jan 2020 • James D. Miller, Roman Yampolskiy, Olle Häggström

An artificial general intelligence (AGI) might have an instrumental drive to modify its utility function to improve its ability to cooperate, bargain, promise, threaten, and resist and engage in blackmail.

An AGI with Time-Inconsistent Preferences

no code implementations • 23 Jun 2019 • James D. Miller, Roman Yampolskiy

This paper reveals a trap for artificial general intelligence (AGI) theorists who use economists' standard method of discounting.

Decision Making