arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2001.09750cs.CY
收藏

走向理解人工智能的社会和伦理含义的框架

Towards a framework for understanding societal and ethical implications of Artificial Intelligence

Richard Benjamins, Idoia Salazar

摘要:

人工智能(AI)是当今讨论最多的技术之一。有许多创新的应用,如癌症的诊断和治疗,客户体验,新业务,教育,传染病传播和人道主义灾难管理的优化。然而,伴随着所有这些机会,也有责任确保人工智能的良好和公平实践。本文的目的是确定大规模采用人工智能所隐含的主要社会和伦理挑战。我们调查了文献中最常见的挑战,并将其分为七组:1)非预期效果,2)责任,3)未知后果,4)人与机器人的关系,5)权力和财富的集中,6)故意的不良使用,以及7)武器和战争的人工智能。这些挑战应该根据其起源以不同的方式处理;有些挑战有技术解决方案,而另一些则需要道德,社会或政治答案。根据不同的来源,不同的利益攸关方可能需要采取行动。无论确定的利益攸关方是谁,不处理这些问题将导致不确定性和不可预见的后果,可能产生巨大的负面社会影响,特别伤害社会中最脆弱的群体。技术正在帮助做出更好的决策,人工智能正在促进数据驱动的决策,以及基于经验和直觉的讨论,并有许多改进。然而,这项技术的负面影响需要很好地理解,并在我们将其大规模推向世界之前采取行动。

英文摘要:

Artificial Intelligence (AI) is one of the most discussed technologies today. There are many innovative applications such as the diagnosis and treatment of cancer, customer experience, new business, education, contagious diseases propagation and optimization of the management of humanitarian catastrophes. However, with all those opportunities also comes great responsibility to ensure good and fair practice of AI. The objective of this paper is to identify the main societal and ethical challenges implied by a massive uptake of AI. We have surveyed the literature for the most common challenges and classified them in seven groups: 1) Non-desired effects, 2) Liability, 3) Unknown consequences, 4) Relation people-robots, 5) Concentration of power and wealth, 6) Intentional bad uses, and 7) AI for weapons and warfare. The challenges should be dealt with in different ways depending on their origin; some have technological solutions, while others require ethical, societal, or political answers. Depending on the origin, different stakeholders might need to act. Whatever the identified stakeholder, not treating those issues will lead to uncertainty and unforeseen consequences with potentially large negative societal impact, hurting especially the most vulnerable groups of societies. Technology is helping to take better decisions, and AI is promoting data-driven decisions in addition to experience- and intuition-based discussion, with many improvements happening. However, the negative side effects of this technology need to be well understood and acted upon before we launch them massively into the world.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑