人工智能技术的伦理问题及其治理路径研究

当前人工智能技术正在高速地发展并深刻地改变着世界,它迅速渗入和应用到社会各领域中,给人类带来便利的同时也带来新的挑战。本文对人工智能技术应用的伦理问题进行了探索性研究,审视其应用过程中所带来的一系列伦理风险问题,并提出合理规避路径及治理策略,以期促进人工智能技术应用与伦理道德的统一,服务于人类整体的价值追求与伦理准则,使人工智能技术应用增进人类福祉。     

欧盟人工智能的政策、伦理准则及规制路径研究

本文在简要梳理欧盟人工智能战略框架的基础上,从比较的视角具体阐释了欧盟人工智能伦理准则及规制路径的核心内涵,运用“规范力量”与“市场力量”两个概念,分析了其相关政策出台动因及欧盟在全球人工智能竞争与合作中的地位。结合欧盟在此方面的举措,本文进一步思考了其对中国人工智能发展的相关启示,提出我国应保持相对优势,并在竞争环境中立足国情,加快人工智能伦理治理实践,积极推动人工智能治理的跨学科参与和国际合作。     

伦理智能体及其设计:现状和展望

智能体一直是人工智能的主要研究领域之一,任何独立的能够同环境交互并自主决策的实体都可以抽象为智能体.随着人工智能从计算智能到感知智能,再到认知智能的发展,智能体已逐步渗透到无人驾驶、服务机器人、智能家居、智慧医疗、战争武器等人类生活密切相关的领域.这些应用中,智能体与环境、尤其是与人类和社会的交互愈来愈突出,其中的伦理和道德问题日益凸显.人工智能应用的伦理风险和挑战引起了人们的普遍关注,伦理智能体是人工智能伦理的重要研究内容.本文从人工智能伦理的工程设计与实现角度,对伦理智能体及图灵测试、伦理智能体的设计范式、伦理智能体的逻辑程序设计、伦理智能体的形式化验证、伦理困境及分析等进行了介绍和讨论.同时,对伦理智能体及设计所面临的挑战和进一步研究方向进行了述评和展望.     

人工智能的伦理困境与出路

人工智能经过60多年的发展取得了举世瞩目的成就,成为当代及未来发展的驱动力之一。伴随着人工智能的蓬勃发展,其伦理方面的问题也渐渐浮现了出来:人工智能带来劳动异化问题、人工智能的技术形态意识化问题、人工智能的权利和责任问题、人工智能发展中的环境伦理问题等。鉴于人工智能发展的巨大影响力,人工智能伦理问题必须引起重视,增强研发主体的道德责任感、积极推进人工智能发展规范化、法律化,加强各国的交流合作,确保人工智能在造福人类的轨道上发展,还人工智能一颗“良芯”。     

面向人工智能伦理计算的中文道德词典构建方法研究

道德词典资源建设是人工智能伦理计算的一个研究重点。由于道德行为复杂多样,现有的英文道德词典分类体系并不完善,而中文方面目前尚未有相关的词典资源,理论体系和构建方法仍待探究。针对以上问题,该文提出了面向人工智能伦理计算的中文道德词典构建任务,设计了四类标签和四种类型,得到包含25 012个词的中文道德词典资源。实验结果表明,该词典资源不仅能够使机器学会道德知识、判断词的道德标签和类型,而且能够为句子级别的道德文本分析提供数据支持。     

人工智能大模型的隐私保护与数据安全技术研究

随着人工智能（AI）技术的快速发展,AI大模型在各个领域的应用越来越广泛。然而,这些模型在伦理、安全和治理方面面临着诸多挑战。本文旨在探讨人工智能大模型在这些方面的挑战以及相应的应对策略。首先,本文分析了人工智能大模型在伦理方面可能引发的问题,如数据隐私、算法歧视和决策透明度等。针对这些问题,提出加强数据保护、改进算法设计和提高透明度等措施。其次,本文讨论了人工智能大模型在安全方面的挑战,包括对抗性攻击、模型泄露和恶意使用等。为了应对这些挑战,提出加强安全防护、建立安全审计机制和制定严格的使用规范等建议。最后,本文探讨了如何建立有效的监管框架以确保人工智能大模型的合规性和可持续发展,提出推动社区合作等策略,期望通过采取适当的措施,能够在一定程度上降低人工智能大模型在伦理、安全和治理方面的风险。     

人工智能安全框架

随着人工智能时代的到来,各行各业均开始结合自身业务需要部署人工智能系统,这全面加速了全球范围内人工智能规模化部署和应用进程.然而,人工智能基础设施、设计研发以及融合应用过程中面临的安全风险也随之而来.为了充分规避风险,世界各国纷纷采取制定人工智能伦理准则、完善法律法规和行业管理等方式来进行人工智能安全治理.在人工智能安全治理中,人工智能安全技术体系具有重要指导意义.具体而言,人工智能安全技术体系是人工智能安全治理的重要组成部分,是落实人工智能伦理规范和法律监管要求的重要支撑,更是人工智能产业健康有序发展的重要保障.然而,在当前阶段,全球范围内人工智能安全框架普遍缺失,安全风险突出且分立,迫切需要对人工智能各生命周期存在的安全风险进行总结与归纳.为解决上述问题,文中提出了涵盖人工智能安全目标、人工智能安全分级能力、人工智能安全技术和管理体系的人工智能安全框架,期待为社会各界提升人工智能安全防护能力提供有益参考.     

多模态数据融合下智能教育伦理问题及应对策略

近年来,人工智能在教育领域的发展取得了重大突破,其中多模态数据融合与智能教育相结合的应用愈加广泛,其可能引发的伦理风险也备受社会关注。本文详细阐述了多模态数据融合下智能教育的发展现状,深入剖析了智能教育可能引发的伦理问题,最后提出解决现有伦理困境的政策措施,以期推动智能教育向善发展,更好地服务于人类和社会。     

符合伦理的人工智能应用的价值敏感设计:现状与展望

价值敏感设计是一种将伦理嵌入人工智能设计的高效方法,尤其是其独特的三方方法为人工智能应用提供了规避伦理问题和道德风险的手段,因此有着广阔的应用前景.本文系统地介绍了价值敏感设计的内涵、原理和方法;详细描述了价值敏感设计在国内外的研究现状,并对典型研究内容进行了对比分析;总结了近年来价值敏感设计在智能机器人、智能运载工具...     

互联网+时代的商业伦理课程教育研究

为了使高校商业伦理实践更好地与"互联网+"时代融合,需要改变高等教育界对商业伦理课程教学深度与授课范围的固有认知.结合大数据、人工智能等技术发展对市场环境提出的新商业伦理诉求,本文对商业伦理教育领域的发文量进行时间序列图分析,运用Vosviewer对该领域文献采取关键词共现分析.结果表明,商业伦理教育应抓住思政需求,紧...     

基于社会新闻数据集的伦理行为判别方法

人工智能(artificial intelligence,AI)应用的伦理风险和挑战引起了人们的普遍关注,如何从技术实现角度开发出遵守人类价值观和伦理规范的AI系统,即,符合伦理的AI设计,是亟需解决的重要问题之一.基于机器学习的伦理与道德判别是此方面的有益探索.社会新闻数据具有丰富的伦理和道德的内容及知识,为机器学习的训练数据开发提供了可能.鉴于此,本文构建了具有人类行为伦理和道德属性的社会新闻数据集,附之以法律与行为规范数据集,用以机器学习的训练和测试;建立了基于使用信息实体的增强语言表示(enhanced language representation of information entities,ERNIE)和卷积神经网络(convolutional neural network,CNN)的伦理行为判别模型ERNIE-CNN,通过词的向量表示计算语义相似度来提取关于行为的伦理判断.实验结果表明,提出的模型具有比基准模型更好的性能,验证了方法和模型的有效性.     

机制主义人工智能理论——一种通用的人工智能理论

现行人工智能研究取得了许多进展,但存在“深度上浅层化、广度上碎片化和体系上封闭化”的重要缺陷。这不是改进算法或者提高硬件性能所能解决的问题,而是要在科学观方法论上寻找根源。本文依据“科学观→方法论→研究模型→研究途径→基本概念→基本原理”这个顶天立地的研究纲领,总结了信息科学的科学观,提炼了信息生态方法论;在新的科学观和方法论指导下构筑了体现智能生长全过程的研究模型,发现了智能生长的共性机制,确立了机制主义研究途径,进而澄清和匡正了信息（特别是语义信息）、感知、知识、认知、基础意识、情感、理智、综合决策等一系列基础概念,总结了实现信息－知识－智能转换的一组基本原理,创建了机制主义人工智能理论。而且证明了:长期三分而立的结构主义（人工神经网络）、功能主义（专家系统）、行为主义（感知动作系统）三大人工智能理论可在机制主义人工智能理论框架内实现和谐统一;机制主义是生成基础意识、情感、理智三位一体高等人工智能的科学途径;机制主义人工智能理论是通用型的人工智能理论。     

Ethical decision making in technology development: a case study of participation in a large-scale information systems development project

Advanced systems engineering has traditionally paid little attention to ethical concerns relative to other technical and non-technical issues. This is particularly evident in systems analysis, design, and development methodologies. This paper asks if it is possible that the lack of emphasis upon ethical considerations in development methodologies can result in the failure of advanced technology development projects? In order to explore this contention, the paper sets out the findings of a case study of a large-scale advanced technology project in a multinational engineering company involving the implementation of an enterprise resource planning system. The research examined the extent to which ethical issues emerged in the project and assesses the impact of ethical considerations upon the technology development process and its outcomes. Evidence is presented which shows how ethical concerns clearly impacted upon the outcome of the project, supporting the contention that ethics was a success factor in the case presented. However, it was also clear that the kinds of ethical considerations that emerged were highly complex, and associated with an “ethics of care”. The findings suggested that researchers should examine the potential of an “ethics of care” as a way of complimenting the “ethics of rights” currently dominant within engineering ethics.     

Ethical norms and issues in crowdsourcing practices: A Habermasian analysis

Crowdsourcing practices have generated much discussion on their ethics and fairness, yet these topics have received little scholarly investigation. Some have criticized crowdsourcing for worker exploitation and for undermining workplace regulations. Others have lauded crowdsourcing for enabling workers' autonomy and allowing disadvantaged people to access previously unreachable job markets. In this paper, we examine the ethics in crowdsourcing practices by focusing on three questions: (a) What ethical issues exist in crowdsourcing practices? (b) are ethical norms emerging or are issues emerging that require ethical norms? and, more generally, (c) how can the ethics of crowdsourcing practices be established? We answer these questions by engaging with Jürgen Habermas' discourse ethics theory to interpret findings from a longitudinal field study (from 2013 to 2016) involving key crowdsourcing participants (workers, platform organizers, and requesters) of three crowdsourcing communities. Grounded in this empirical study, we identify ethical concerns and discuss the ones for which ethical norms have emerged as well as others which remain unresolved and problematic in crowdsourcing practices. Furthermore, we provide normative considerations of how ethical concerns can be identified, discussed, and resolved based on the principles of discourse ethics.     

Ethical theory, codes of ethics and IS practice

Ethical issues, with respect to computer-based information systems, are important to the individual IS practitioner. These same issues also have an important impact on the moral well-being of organizations and societies. Considerable discussion has taken place in the Information Systems (IS) literature on specific ethical issues, but there is little published work which relates these issues to mainstream ethical theory. This paper describes a range of ethical theories drawn from philosophy literature and uses these theories to critique aspects of the newly revised ACM Code of Ethics. Some in the IS field on problematic ethical issues which are not resolved by the Code are then identified and discussed. The paper draws some implications and conclusions on the value of ethical theory and codes of practice, and on further work to develop existing ethical themes and to promote new initiatives in the ethical domain.     

网络的发展给社带会来的伦理道德问题

随着计算机网络的发展,也带来了许多政治、法律伦理道德和社会问题研究探讨网络发展所带来的伦理道德问题,已经成为国内外各界人士普遍重视的前沿性课题。道德范畴主要包含三个方面内容：道德意识、道德规范、道德行为、本文探讨了计算机网络发展对道德的复杂性影响。     

网络的发展给社带会来的伦理道德问题

随着计算机网络的发展,也带来了许多政治、法律伦理道德和社会问题研究探讨网络发展所带来的伦理道德问题,已经成为国内外各界人士普遍重视的前沿性课题。道德范畴主要包含三个方面内容:道德意识、道德规范、道德行为。本文探讨了计算机网络发展对道德的复杂性影响。     

Tomorrow’s ethics and today’s response: An investigation into the ways information systems professionals perceive and address emerging ethical issues

This paper explores the question of how foresight and futures research can identify and address ethical issues in the field of Information Systems (IS). Starting from the premise that such IS are part of socio-technical systems, the interaction between technology and human actors raise ethical concerns. Early recognition of these concerns can address ethical issues and improve the use of the technology for a range of social and organisational goals. This paper discusses research conducted in two futures research projects. Both projects investigated emerging information and communication technologies (ICTs) and ethics. The first project established approaches for identifying future technologies and their related ethical concerns. This led to the identification of 11 emerging ICTs and their associated ethical concerns. The second project took these general ethical concerns and focused on their role in IS. Specifically, how IS professionals view future emerging technologies, their associated ethical concerns, and how they think these concerns could be addressed. The key findings are that IS professionals are primarily interested in the job at hand and less so in the ethical concerns that the job might bring; ethics is a concern that is best left for others to deal with. This paper considers the implications of research on ethics in emerging ICTs and draws general conclusions about the relevance of future technologies research in IS.     

Computational Models of Ethical Reasoning: Challenges, Initial Steps, and Future Directions

Computational models of ethical reasoning are in their infancy in the field of artificial intelligence. Ethical reasoning is a particularly challenging area of human behavior for AI scientists and engineers because of its reliance on abstract principles, philosophical theories not easily rendered computational, and deep-seated, even religious, beliefs. A further issue is this endeavor's ethical dimension: Is it even appropriate for scientists to try to imbue computers with ethical-reasoning powers? A look at attempts to build computational models of ethical reasoning illustrates this task's challenges. In particular, the Truth-Teller and SIROCCO programs incorporate AI computational models of ethical reasoning, both of which model the ethical approach known as casuistry. Truth-Teller compares pairs of truth-telling cases; SIROCCO retrieves relevant past cases and principles when presented with a new ethical dilemma. The computational model underlying Truth-Teller could serve as the basis for an intelligent tutor for ethics.This article is part of a special issue on Machine Ethics.