Affiliation:
1. University of Alabama in Huntsville, Huntsville, AL, USA
Abstract
The concept of reliability is widely recognized across various academic disciplines. However, the conventional understanding of reliability varies by discipline, does not adequately address the intricate and ever-changing environments, and fails to account for the dynamic interactions between humans and artificial intelligence (AI) systems. To address these gaps, a new framework is proposed for considering reliability that accounts for performance on both supraordinate and subordinate objectives. By framing reliability in such a manner, the evaluation of systems can become more precise and research involving human-machine interactions can gain greater clarity. This is crucial for product designers and evaluators working to develop systems that meet end-use goals and comply with regulations. Researchers and practitioners alike need to rethink reliability in the context of AI systems, and this article proposes a new framework for understanding reliability.
Subject
General Medicine,General Chemistry