双语 | 让我们打造更好的人工智能
在人工智能领域快速发展的背景下,我们必须始终重视以下几个关键问题:要负责任地善用数字技术,谁来设计并实施全球行为准则和标准体系?如何做出决策?关键受众是谁?希望达成什么样的最终效果?
As the field develops rapidly, one must ponder on who designs global norms on the responsible use of digital technology, how are decisions made, and for whom.
”本文英文版首发于《中国日报》,作者徐浩良系联合国助理秘书长兼联合国开发计划署政策与方案支助局局长。
The English version was oriiginally published on China Daily. The author is XU Haoliang, assistant secretary-general of the United Nations and the director of the Bureau for Policy and Programme Support at the United Nations Development Programme.
WANG Xiaoying / CHINA DAILY
人工智能(artificial intelligence, AI)拥有巨大潜力,不仅可以推动我们实现可持续发展目标(SDGs),还能确保我们在发展过程中不断增强韧性。
Artificial intelligence presents tremendous potential to not only advance our progress toward achieving the Sustainable Development Goals, but also ensure that we continuously build resilience along the way.
比如,我们已经看到人工智能如何在灾害预防、气候适应和韧性建设方面发挥的重要作用。例如,地理空间卫星数据可以帮助识别并预防即将发生的灾害,也能帮助绘制灾害损失地图。人工智能解决方案也被用于提高信息真实性。例如,联合开发计划署(UNDP)开发了开源自动事实核查工具iVerify,可用于帮助识别虚假信息,继而防止或减少其传播。iVerify现已被列为数字公共产品,已经在赞比亚、肯尼亚、洪格拉斯且即将在利比里亚投入使用,以处理选举期间的不实信息。
We have seen, for example, how AI can play a vital role in disaster prevention, climate adaptation and resilience. For instance, geospatial satellite data can help to identify and prevent impending disasters, as well as aid in damage mapping. AI-enabled solutions are also being implemented to promote information integrity. For example, the United Nations Development Programme has developed iVerify, an open source, automated fact-checking tool that can be used to help identify false information and prevent and mitigate its spread. iVerify is now listed as a Digital Public Good and has been deployed in Zambia, Kenya and Honduras, and will soon be in Liberia, to tackle misinformation during elections.
我们正在见证人工智能如何支持大规模的行为改变,如何通过在关键时刻实现数据驱动决策来赋权于人。
We are witnessing how AI is supporting behavioral change at scale and empowering people by enabling data-driven decision-making at crucial junctures.
然而,我们同时也意识到,人工智能也可能造成潜在风险与危害,解决这些问题刻不容缓。例如,以不透明的方式收集和分析数据可能导致歧视性结果。
However, at the same time, we are also seeing an alarming need to address the potential risks and harms it can pose. For instance, data collected and analyzed in untransparent ways can lead to discriminatory outcomes.
举例而言,预测性执法和自动化司法决策在某些情况下,似乎对包括女性在内的少数群体、边缘化群体持有偏见。
Examples of this include predictive policing and automated judicial decision-making that in some cases appear to present a bias against minorities and marginalized groups, including women.
人工智能驱动的系统也能被用以破坏隐私权和基本自由。例如,人工智能驱动的去匿名化工具,可在私人或公共空间跨设备盗取私人数据并跟踪对象。
AI-driven systems can also be used in ways that undermine the right to privacy and fundamental freedoms. AI-powered de-anonymization tools can be used to disclose private data and track people across devices, in private and public spaces.
尽管世界上有些地区已经采取行动,政府在监管人工智能的风险和危害方面仍持续面临困境。
While efforts have been put in place in several parts of the world, governments continue to struggle when it comes to mitigating the risks and harms of AI.
UNDP已经认识到,数字技术并非一劳永逸的万能灵药,因此我们制定了具有雄心的愿景,即创建一个“数字技术助力人类和地球发展”的世界。我们与各国政府的交流与互动凸显了对数据和人工智能的治理与道德规范的迫切需求。
Recognizing that digital technology is not a panacea, the UNDP's digital strategy lays out a bold vision, to help create a world in which “digital is an empowering force for people and the planet”. Our observations and interactions with governments have underlined the urgent need for governance and ethical guidance on the use of data and AI.
我们必须推动并捍卫安全、负责、包容的人工智能治理。
We should promote and champion a safe, responsible, and inclusive governance of AI.
首先,采取以人为本、以权利为基础的手段,来管理数据的可获取性、质量、透明度和问责制度,作为合乎伦理地使用人工智能的前提。
First, adopt a people-centered, rights-based approach to data availability, quality, transparency, and accountability as a precondition to an ethical deployment of AI.
例如,UNDP是联合国人工智能跨机构工作组(IAWG-AI)的成员。在联合国教科文组织(UNESCO)的领导下,工作组提出了关于人工智能伦理原则的指导意见,包括遵守国际法与人权法,涵盖了隐私权、公平与非歧视权、数字责任。
For example, the UNDP is part of the United Nations Inter-Agency Working Group on AI, which, thanks to leadership from the United Nations Educational, Scientific and Cultural Organization, has put forward guidance on the Ethical Principles on AI. These include abiding by international and human rights laws, including the right to privacy, fairness and nondiscrimination, and data responsibility.
这至少意味着,人应当有权不受制于画像分析等完全基于自动程序的决策。例如,在通过自动算法确定永久或临时住房的优先次序和资格时,人们也应该有权要求人工干预。
This means — among other things — that people should have the right not to be subject to decisions based solely on automated processing, including profiling. Look, for instance, at automated decision-making being used to prioritize and determine eligibility for permanent or temporary housing, where people should have the right to request human intervention.
第二,采用主动开放的做法,不仅是促进政企合作,还要把民间团体、学术界和所有公民都联合起来。这对于填补如今的数字鸿沟至关重要:目前,仍有27亿人无法接触互联网,其中95%都生活在发展中国家。
Second, take an intentionally inclusive approach which brings together not just governments and businesses but also civil society, academia and citizens. This is critical to address the current digital divide with 2.7 billion people offline; 95 percent of whom live in developing countries.
UNDP正在探索人工智能就绪度评价方法,以帮助各国政府借助包括边缘化群体在内的多利益相关方的参与,了解人工智能使用情况。
The UNDP is developing an AI Readiness Assessment to help governments understand the current state of AI adoption through multi-stakeholder engagement, including with marginalized groups.
第三,支持能力建设,包括加深公众对人工智能、技术治理以及数据价值的认识与理解。
Third, support capacity-building, including raising public awareness and understanding of AI technologies, governance, and the value of data.
为解决这一关键问题,UNDP和国际电信联盟(ITU)启动了一项数字能力发展联合基金,来帮助在现有努力下仍缺乏数字服务的群体。
To address this key issue, the UNDP has launched a Joint Facility for Digital Capacity Development for those underserved by existing efforts.
我们还在支持摩尔多瓦、塞内加尔、毛利塔利亚、肯尼亚等国制定数据治理框架,以帮助他们在政策制定中负责任地、包容地使用数据。
We are also supporting countries such as Moldova, Senegal, Mauritania and Kenya to develop Data Governance frameworks to support responsible and inclusive use of data in policymaking.
此外,UNDP目前也在为数字身份系统开发一个示范性治理框架,可方便地以开源形式供政府采用。
Additionally, the UNDP is currently developing a model governance framework for digital identity systems, which will be open-source and made easily accessible for governments to adopt.
上述能力建设的行动和工具,将确保输入人工智能系统的数据都能接受符合国际规范与伦理的、透明的、负责任的管理。
These capacity-building efforts and tools will ensure that data that feed into AI systems are governed in an ethical, transparent and accountable manner in accordance with international standards.
尽管这些努力均旨在支持政府和人民以积极的方式利用人工智能的力量,我们仍然需要更广泛的参与。
While these efforts are aimed toward supporting governments and people to harness the power of AI in positive ways, wider engagement is still required.
在人工智能领域快速发展的背景下,我们必须始终重视以下几个关键问题:要负责任地善用数字技术,谁来设计并实施全球行为准则和标准体系?如何做出决策?关键受众是谁?希望达成什么样的最终效果?
As the field develops at a rapid pace, we must continue asking ourselves critical questions, such as, who is invited to the table to design and implement global norms and standards on the responsible use of digital technology, how decisions are made, and for whom and to what end?
我们必须一起努力,共建一个惠及所有人、所有地方的数字未来。
We should work together to co-develop a digital future that benefits everyone, everywhere.
往期回顾
有奖互动 | COP15闭幕,“生生万物”上线,邀您共赴生物多样性之旅
AI Forum | Web3助力实现可持续发展目标
AI Forum丨正视人工智能引发的性别歧视