Applying AVWEWM to ethical decision-making during autonomous vehicle crashes

Area of Science:

• Computational Ethics
• Artificial Intelligence Decision-Making
• Autonomous Vehicle Safety

Background:

• Existing research on ethical dilemma decision-making lacks quantitative methods to weigh influencing factors.
• This gap hinders the development of computational models for resolving ethical dilemmas, particularly in autonomous systems.
• Autonomous vehicles (AVs) face complex ethical decision-making scenarios in crash situations.

Purpose of the Study:

• To develop and evaluate computational decision-making models for autonomous vehicle (AV) ethical dilemmas.
• To quantitatively assess the weight and role of various factors in ethical decision-making.
• To compare the effectiveness of the Entropy Weighted Method (EWM) and Attribute Value Weighted EWM (AVWEWM) in AV crash scenarios.

Main Methods:

• Applied the Entropy Weighted Method (EWM) and Attribute Value Weighted EWM (AVWEWM) to 84 ethical dilemmas.
• Developed decision-making models based on EWM and AVWEWM for AV crash situations.
• Tested the models using 40 distinct ethical dilemmas to assess decision clarity and public opinion alignment.

Main Results:

• Both EWM and AVWEWM models successfully provided clear decisions for 40 tested AV ethical dilemmas.
• The AVWEWM model demonstrated greater consistency with public opinion compared to the EWM model.
• Identified the weight and influence of different factors in ethical decision-making processes.

Conclusions:

• Computational models, particularly AVWEWM, can effectively address ethical dilemmas in autonomous vehicle crashes.
• AVWEWM offers a more public-opinion-aligned approach to ethical decision-making for AVs.
• Findings provide valuable insights for traffic management and legal bodies in formulating future AV regulations.