Ethics Policy
A new system teaches robots to operate according to ethical principles
Researchers have developed a system called Principles2Plan, which helps transform general ethical principles into concrete operational guidelines for robots and other automated systems. The goal is to support the ethical operation of robots in environments where they interact with humans.
Current automated planning tools are already capable of creating action plans, but they poorly consider aspects like privacy protection or harm avoidance. On the other hand, manually defining ethical rules is laborious and highly case-specific: the same principle can mean different things in different situations.
Principles2Plan introduces an interactive layer where human expertise and a large language model work together to form rules suitable for the situation. The system receives a description from the user about the task world being planned, the problem at hand, and key high-level principles such as beneficence and privacy.
Based on this, the system produces more detailed, practically implementable ethical rules that align with the given principles. The user can review and, if necessary, modify the rules and set their priorities. After that, the rules are used to guide the automated planner so that the action plans it produces are ethically considered.
According to the researchers, a similar system that supports users in structuring ethical principles for planning has not been previously introduced. Principles2Plan is a prototype that demonstrates how a human and a language model can together build situation-specific ethical guidelines as part of robots' decision-making.
Source: Principles2Plan: LLM-Guided System for Operationalising Ethical Principles into Plans, ArXiv (AI).
Current automated planning tools are already capable of creating action plans, but they poorly consider aspects like privacy protection or harm avoidance. On the other hand, manually defining ethical rules is laborious and highly case-specific: the same principle can mean different things in different situations.
Principles2Plan introduces an interactive layer where human expertise and a large language model work together to form rules suitable for the situation. The system receives a description from the user about the task world being planned, the problem at hand, and key high-level principles such as beneficence and privacy.
Based on this, the system produces more detailed, practically implementable ethical rules that align with the given principles. The user can review and, if necessary, modify the rules and set their priorities. After that, the rules are used to guide the automated planner so that the action plans it produces are ethically considered.
According to the researchers, a similar system that supports users in structuring ethical principles for planning has not been previously introduced. Principles2Plan is a prototype that demonstrates how a human and a language model can together build situation-specific ethical guidelines as part of robots' decision-making.
Source: Principles2Plan: LLM-Guided System for Operationalising Ethical Principles into Plans, ArXiv (AI).
This text was generated with AI assistance and may contain errors. Please verify details from the original source.
Original research: Principles2Plan: LLM-Guided System for Operationalising Ethical Principles into Plans
Publisher: ArXiv (AI)
Authors: Tammy Zhong, Yang Song, Maurice Pagnucco
December 27, 2025
Read original →