Three areas of research on the superintelligence control problem

Abstract

The control problem of superintelligent artificial intelligence (AI) is explored in this work through the description of three main areas of research: technical foresight, strategy, and system design. Technical foresight aims to comprehend the possible properties of superintelligent AI to determine potential risks and benefits. Strategy explores societal responses to potential risks, with a focus on strategic planning, response development, and domain-specific strategy. Finally, system design involves the precise specification and implementation of reliable, robust, and value-aligned AI systems. A sober assessment of the current state of research reveals that it is relatively early and many open questions remain. Further progress may challenge current understandings of risks from superintelligent AI, warranting continued research efforts and interdisciplinary collaboration. – AI-generated abstract.