Using Algorithms to Make Ethical Judgements: METHAD vs. the ADC Model
Using Algorithms to Make Ethical Judgements: METHAD vs. the ADC Model
复制标题
使用算法做出道德判断：METHAD 与 ADC 模型
DOI：
10.1080/15265161.2022.2075967
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Dubljević, Veljko
中科院分区：
文献类型：
--
作者：
Coin, Allen;Dubljević, Veljko
在他们的论文“临床伦理决策的算法：概念证明”中，Meier等人。（2022）提出了一种概念验证临床伦理算法的设计和初步结果，他们声称可以使用机器学习对医疗保健中可能发生的道德困境提出有限的建议，使用Beauchamp和奇尔德里斯所支持的原则主义伦理框架。他们报告说，该算法取得了一些成功，测试结果在92%的训练数据集和75%的测试数据集中与伦理学家一致。该算法的一个明显局限性是，它的设计部分基于生物伦理学的原则主义方法，没有考虑到其他伦理框架。此外，作者描述的算法要求用户输入许多主观的数值变量，并根据在临床环境中使用算法的人的不同而导致不同的结果。此外，在输出中缺乏一个中立的中间点，导致作者在解释结果时偏向于干预。这篇评论的目的是传达和对比我们在自己的国家科学基金会资助的工作中学到的一些相关经验教训（#2043612），在那里我们结合了联合收割机的美德理论，义务论，和结果主义的方法，伦理决策算法内的代理人的行为后果（ADC）模型，沿着与人类决策者的实证研究结果。METHOD和ADC方法之间的主要区别在于，前者探讨了人工智能辅助的道德决策，而后者探讨了人工智能授权的道德决策是否可能。的
In their paper “Algorithms for Ethical Decision-Making in the Clinic: A Proof of Concept,” Meier et al.(2022) present the design and preliminary results of a proof-of-concept clinical ethics algorithm that they claim can use machine learning to make limited recommendations about moral dilemmas that may occur in healthcare, using the ethical framework of principlism as espoused by Beauchamp and Childress. They report some success for the algorithm, with tests producing results agreeing with ethicists in 92% of the training data set and 75% in the test set. A stated limitation of the algorithm is that it is designed based in part on the principlist approach to bioethics and does not account for other ethical frameworks. Additionally, the algorithm described by the authors requires the user to input a number of numerical variables that are subjective and would lead to variable results depending on the human utilizing the algorithm in a clinical setting. Moreover, the lack of a neutral median point in the output leads to bias toward intervention in the interpretation of the results by the authors.The purpose of this commentary is to relay and contrast some relevant lessons we have learned in our own National Science Foundation-funded work (# 2043612), where we combine the virtue theoretic, deontological, and consequentialist approaches for ethical decision-making algorithms within the Agent-Deed-Consequence (ADC) model, along with results of empirical research with human decision makers. The primary difference between METHAD and ADC approaches is that the first explores AI-assisted moral decision making, whereas the latter explores whether AI-empowered moral decision making is possible. The
登录
查看更多内容
用于老年人护理的护理机器人：技术、伦理和影响
DOI：
--
发表时间：
2021
期刊：
影响因子：
--
作者：
Allen Coin;Veljko Dubljević
通讯作者：
Veljko Dubljević
影响因子：
3.7
作者：
Dubljević V;Sattler S;Racine E
通讯作者：
Racine E
人类与人工智能团队的道德：原则和观点
DOI：
10.1007/s43681-022-00214-z
发表时间：
2022
期刊：
AI and Ethics
影响因子：
--
作者：
Pflanzer, Michael;Traylor, Zachary;Lyons, Joseph B.;Dubljević, Veljko;Nam, Chang S.
通讯作者：
Nam, Chang S.
影响因子：
3.8
作者：
Sattler, Sebastian;Dubljevic, Veljko;Racine, Eric
通讯作者：
Racine, Eric