机器道德伦理在人与人工智能交互中的应用研究

发布时间：2020-12-05 18:30

　　本研究旨在以人工智能道德建设的角度评估人们对人机交互的态度与偏好,探讨如何在不同伦理困境情境中设计自动化和决策方法,并确定在意外发生时人们的责任追求倾向。笔者首先对30名受访者进行初步访谈,以收集大众对未来人类与人工智能交互场景的见解与担忧。访谈结果表明智能家居、自动驾驶、专业服务及军事战斗为人们心中未来人工智能技术应用最广泛的领域。笔者随后根据初步访谈所得到的这四大类领域以及受访者对其所提出的担忧设计出了之后问卷调查所需的情境,并通过对103名参与者进行基于8个不同情境的问卷调查,以收集定量数据。笔者在制定问卷过程中还充分考虑对人类决策过程有影响的三个要素:事件的严重性、主体决策的时间限制和主体与事件的相关性,而正是这三个因素构建了这8种不同水平组合的伦理困境。另外通过针对30名参与者进行深度访谈来收集用于支持问卷结果的相关定性数据。问卷调查数据的重复测量方差分析结果显示,这三个因素显著影响着受众对自动化水平,决策方式和责任分配的选择。首先,在与参与者相关,且决策可用时间较短的情况下,参与者对监控自动化满意度最高,而在决策可用时间较长的情况下,参与者对授权自动化最为满意。在与参与者无... 

【文章来源】：清华大学北京市 211工程院校 985工程院校 教育部直属院校

【文章页数】：110 页

【学位级别】：硕士

【文章目录】：
摘要
ABSTRACT
1 INTRODUCTION
    1.1 Research background
    1.2 Research objective and questions
2 LITERATURE REVIEW
    2.1 Automation and decision support systems
    2.2 Morality
        2.2.1 Machine ethics
        2.2.2 Moral status of machines
        2.2.3 Moral dilemma and sources of ethical standards
    2.3 General factors related to human-machine interaction
        2.3.1 Comfort
        2.3.2 Trust
        2.3.3 Self-confidence
        2.3.4 Factors affecting decision-making process
3 RESEARCH FRAMEWORK
    3.1 Research orientation: the people
    3.2 Research focus and variables
    3.3 Research methodology
        3.3.1 Preliminary interview on public interest in future human-AI interactions
        3.3.2 Ethical dilemma scenario survey
4 PRELIMINARY INTERVIEW ON PUBLIC INTERESTS IN FUTURE HUMAN-AI INTERACTIONS
    4.1 Methodology of the preliminary interview
        4.1.1 Part 1: Design project
        4.1.2 Part 2: Concerns for future AI technology
        4.1.3 Participants and practical details
    4.2 Results of the preliminary interview
        4.2.1 Part 1: Design project
        4.2.2 Part 2: Concerns for future AI technology
        4.2.3 Brief summary of the results
5 ETHICAL DILEMMA SCENARIO SURVEY
    5.1 Methodology of the online questionnaire
        5.1.1 Part 1: Basic information
        5.1.2 Part 2: Moral dilemma scenarios
    5.2 Results of the online questionnaire
        5.2.1 Participants’ profile
        5.2.2 Automation preference
        5.2.3 Decision-making approach
        5.2.4 Responsibility allocation
        5.2.5 Brief summary of the results and limitations
    5.3 Methodology for the in-depth interview
    5.4 Results from the in-depth interview
        5.4.1 Participants’ profile
        5.4.2 Insights from in-depth interviews
    5.5 Brief summary of the overall survey results
6 CONCLUSION
REFERENCE
ACKNOWLEDGEMENT
APPENDIX
    APPENDIX 1: Design project description
    APPENDIX 2: Online questionnaire for the construction of artificial intelligence morality standard
RESUME



本文编号：2899895