Accepted author manuscript, 814 KB, PDF document
Available under license: CC BY: Creative Commons Attribution 4.0 International License
Final published version
Licence: CC BY: Creative Commons Attribution 4.0 International License
Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Conference contribution/Paper › peer-review
Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Conference contribution/Paper › peer-review
}
TY - GEN
T1 - Encoding Social & Ethical Values in Autonomous Navigation
T2 - TAS ’24, September 16–18, 2024, Austin, TX, USA
AU - Tang, Yun
AU - Moffat, Luke
AU - Guo, Weisi
AU - May-Chahal, Corinne
AU - Deville, Joe
AU - Tsourdos, Antonios
PY - 2024/9/16
Y1 - 2024/9/16
N2 - Autonomous Systems (ASs) interacting with human societies raisescomplex social & ethical challenges. This paper argues that one wayof scaffolding human trust in ASs is through the encoding of ethical,legal and social impact (ELSI) considerations in the ASs’ decisionmakingprocesses. Existing ELSI-encoding efforts often focus on theimplementation of rule-based and risk-based approaches, leavingkey questions unanswered - what are the relationships betweenELSI-encoding software logic in ASs and human ethical practises;what ethical approaches cannot be easily translated into softwarerules and numeric risks; and what are the implications of this forethical AS?To answer these questions, we review and discuss different ELSIencodingapproaches in ASs from a new perspective, i.e., theirrelationships with classic human ethics philosophies. We also explorethe feasibility of large language models (LLMs)-based ELSIencodingpractices in overcoming the limitations of rule-based andrisk-based approaches and the associated challenges. To foster understanding,facilitate knowledge exchange and inspire discussionamong cross-disciplinary research communities, we build and publishthe first online interactive playground demonstrating differentELSI-encoding approaches on the same AS decision-making process.We welcome feedback and contributions in making this platformtruly beneficial to trustworthy autonomous system research communities.
AB - Autonomous Systems (ASs) interacting with human societies raisescomplex social & ethical challenges. This paper argues that one wayof scaffolding human trust in ASs is through the encoding of ethical,legal and social impact (ELSI) considerations in the ASs’ decisionmakingprocesses. Existing ELSI-encoding efforts often focus on theimplementation of rule-based and risk-based approaches, leavingkey questions unanswered - what are the relationships betweenELSI-encoding software logic in ASs and human ethical practises;what ethical approaches cannot be easily translated into softwarerules and numeric risks; and what are the implications of this forethical AS?To answer these questions, we review and discuss different ELSIencodingapproaches in ASs from a new perspective, i.e., theirrelationships with classic human ethics philosophies. We also explorethe feasibility of large language models (LLMs)-based ELSIencodingpractices in overcoming the limitations of rule-based andrisk-based approaches and the associated challenges. To foster understanding,facilitate knowledge exchange and inspire discussionamong cross-disciplinary research communities, we build and publishthe first online interactive playground demonstrating differentELSI-encoding approaches on the same AS decision-making process.We welcome feedback and contributions in making this platformtruly beneficial to trustworthy autonomous system research communities.
KW - Trustworthy Autonomous Systems, ELSI-encoding, Large Language Model, Demonstration
U2 - 10.1145/3686038.3686044
DO - 10.1145/3686038.3686044
M3 - Conference contribution/Paper
T3 - Communications of the ACM
SP - 1
EP - 9
BT - TAS '24: Proceedings of the Second International Symposium on Trustworthy Autonomous Systems
PB - ACM
CY - New York
Y2 - 16 September 2024 through 18 September 2024
ER -