Practical wisdom and intelligent machines: toward AI-human socio-technical decision making and resource allocation

Abstract

This project, The Machine Wisdom Project, aims to develop a framework for defining and enhancing practical wisdom in human-machine interactions. The notion of practical wisdom highlights both (1) the current limitations in Artificial Intelligence and Machine Learning Systems (AI/ML) and (2) the dangers entailed when humans must make real-time adjustments to AI/ML systems with only a partial understanding of how they work. The Machine Wisdom Project envisages strategies for mitigating both (1) and (2) that recognize the lack of self-awareness in current and foreseeable AI/ML technologies and the need to promote designs that make those limitations apparent. Feedback loops between the actions of people and artificially intelligent machines constitute socio-technical systems with the potential to alter (positively and negatively) the capacity of individuals to act ethically. The project researchers seek to deliver principles and recommendations for the design of socio-technical systems that minimize the likelihood that people adapt to the rigid incomprehension of machines in ways that restrict their ethical autonomy and responsibility, that minimize the potential for bad actors to exploit these limitations, and that improve the capacity of individuals to act ethically within the socio-technical system. This research article presents the Ethical, Multimodal, Moral Agent (EMMA) system, an HVAC socio-technical decision-making test bed for the Machine Wisdom Project.