Autonomous Systems (ASs) interacting with human societies raises
complex social & ethical challenges. This paper argues that one way
of scaffolding human trust in ASs is through the encoding of ethical,
legal and social impact (ELSI) considerations in the ASs’ decisionmaking
processes. Existing ELSI-encoding efforts often focus on the
implementation of rule-based and risk-based approaches, leaving
key questions unanswered - what are the relationships between
ELSI-encoding software logic in ASs and human ethical practises;
what ethical approaches cannot be easily translated into software
rules and numeric risks; and what are the implications of this for
ethical AS?
To answer these questions, we review and discuss different ELSIencoding
approaches in ASs from a new perspective, i.e., their
relationships with classic human ethics philosophies. We also explore
the feasibility of large language models (LLMs)-based ELSIencoding
practices in overcoming the limitations of rule-based and
risk-based approaches and the associated challenges. To foster understanding,
facilitate knowledge exchange and inspire discussion
among cross-disciplinary research communities, we build and publish
the first online interactive playground demonstrating different
ELSI-encoding approaches on the same AS decision-making process.
We welcome feedback and contributions in making this platform
truly beneficial to trustworthy autonomous system research communities.