Collaborative Intelligence through smart Disobedience in Ethical Robots (CIDER)
Abstract
The rapid advancement of artificial intelligence and robotics has ushered in an era where human-robot interaction (HRI) is becoming increasingly prevalent across various domains. As robots assume more significant roles in industries such as healthcare and manufacturing, the frequency and complexity of interactions between humans and robots are set to rise substantially. This escalating integration raises critical concerns regarding the safety, ethics, and trustworthiness of these interactions. In this white paper, I propose the investigation of a cognitive capability known as intelligent disobedience - the ability of a robot or autonomous system to disobey commands or requests from other agents (humans or other robots) when following them would violate ethical or safety rules. This is a complex skill that depends on an ensemble of cognitive abilities, such as trust, Theory of Mind, and intention reading, along with awareness of the environment in which the agent is situated. By empowering robots to make informed decisions about when to disobey human commands, we aim to address pressing challenges in HRI, such as reducing human errors, avoiding potentially harmful situations, and mitigating unintended consequences. The successful implementation of this grant will not only pave the way for safer and more trustworthy robots, but it will also have far-reaching implications for a wide range of industries where human-robot collaboration is becoming increasingly prevalent. Embracing the potential of intelligent disobedience, I aim to establish a new paradigm in HRI that promotes user acceptance, societal integration, and harmonious cohabitation between humans and robots.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 05, 2025
- Source ID
- FA86552417046
Entities
People
- Samuele Vinanzi
Organizations
- Air Force Office of Scientific Research
- Sheffield Hallam University
- United States Air Force