Artificial Intelligence: An Accelerator for United Nations Sustainable Development Goals
Yi Zeng

 

Contactless delivery, contact tracing and analysis... We have realized the great power of Artificial Intelligence (AI), a disruptive technology, throughout its widespread applications to fight against the COVID-19 pandemics. Nevertheless, worries follow - will AI application increase income disparity? Will the collection of personal information cause risks related to information security? With these issues in mind, scientists and practitioners are also exploring ways in which AI can facilitate rather than hamper the efforts to achieve Sustainable Development Goals (SDGs). This is not only a strategic goal specified in China's New Generation Artificial Intelligence Development Plan, but also a clear vision laid out in the Governance Principles for the New Generation Artificial Intelligence of the country. Recently, it is also a core topic for the UN Secretary-General's Digital Cooperation Roadmap.

 

AI for sustainable development goals: a global consensus

The United Nations' (UN) SDGs are set to solve development issues in economic, social and environmental dimensions, and realize sustainable development by 2030. The SDGs call for action by "all countries - poor, rich and middle-income - to promote prosperity while protecting the planet." The SDGs specify that the countries shall commit to eradicating poverty, pursuing good health and well-being, ensuring quality education, achieving gender equality, ensuring clean drinking water and sanitation facilities, ensuring access to affordable and clean energy, fostering innovation and building infrastructure, reducing inequalities, developing sustainable cities and communities, combating climate change and protecting the environment, and more. In all these fields, AI has a great role to play.

 

Promoting the realization of UN sustainable development goals through AI has attracted wide attention and promotion from different countries and regions. The Oxford Initiative on AI×SDGs, launched by Oxford University, explores how current and future AI can be used to help support and advance the achievement of the SDGs. The UAE has established the AI4SDGs Centre in Dubai, building on its international communication platform in recent years on AI ethics and governance to promote international dialogue and technology for AI and sustainable development. The AI for Good Foundation, established in 2015, aims to advance the achievement of SDGs by coordinating AI research communities, policy makers and the general public, and has operations in the United States, the United Kingdom, Slovenia and other places.

 

AI as an enabling technology to promote the achievement of SDGs has formed an important consensus in the United Nations. Various UN member states, relevant academic institutions, industries, etc. are also working together to promote AI4SDGs. The United Nations International Telecommunication Union (ITU) is working annually with other UN entities to hold the AI for Good Global Summit, which aims to provide a platform for global dialogue on how AI can be used to accelerate the achievement of the UN's SDGs to help solve more global problems. ITU has also launched the AI repository to collect AI projects, research initiatives, think tanks and organizations related to the goal of promoting sustainable development. The UN Secretary-General's Digital Cooperation Roadmap, released on 12 June 2020, states that AI should be deployed in the best way to support the achievement of sustainable development goals and benefit the public.

 

In the pursuit of zero hunger, machine learning models can be adopted in combination with climate modeling to forecast crop yield in certain areas. Image processing technology can be used to monitor crop growth and send warnings about potential risks and threats of pests and diseases in their early stages, and help increase crop yield. In addition, smart logistics technology can help build a seamless connection between crops and the end consumers.

 

Climate change is a significant issue within the SDG agenda. Machine leaning and prediction studies based on climate science's big data will play an active role in deepening the understanding of climate change. The modeling of improvement strategies and their impacts through computational simulations or other approaches will offer us analytical results and support to mitigate global warming.

 

With regard to environmental protection, automated smart surveillance technology can be used to monitor natural disasters and practices causing environmental damage. Multimodal data processing can bolster the monitoring and prediction of water and air quality. Furthermore, the integration of mode recognition and automatic control technologies can realize automatic waste sorting and processing.

 

AI is also useful in preserving cultural heritage to promote the integrated development of human civilization. Three-dimensional reconstruction and character recognition technology can be applied in cultural heritage excavation, artwork protection, and cultural preservation. Image analysis and natural language processing and understanding can be adopted to build cultural connection and interaction. Furthermore, machine learning and automatic generation technology are expected to advance the creation of cultural and artistic works.

 

The information perception, integration and decision-making technology in AI can boost the development of national and social governance systems. AI has a critical role to play in promoting peace and justice - facial recognition technology helps find missing children and search for criminals, thus safeguarding national and social security. Machine learning is conducive to monitoring and identifying fraud, tax evasion and money laundering in the financial sector, thus reducing fraud and corruption. Furthermore, AI will improve public health governance, for example, by building an early warning system for epidemics and other public health crises with perception and decision-making technologies, to enable public health emergency management and response.

 

Remaining vigilant against the negative impacts of AI and ensuring its positive use

The SDGs cover a wide range of areas. Nevertheless, for some AI applications, being relevant to the field of SDGs does not necessarily mean it will facilitate the field's sustainable development. AI applications in some SDG areas can have both positive and negative impacts. In other words, it can either promote or hinder the development in these areas. To assess the sustainability, it is important to focus on the long-term impacts that AI applications impose on society and ecology.

 

For instance, the integrated development of AI and Internet technologies will provide infrastructure and development opportunities for education by improving education quality and effectiveness with personalized study models, identifying individuals' full potential and encouraging them to reach the potential, and extending quality education resources to students of different genders and races. Applying facial recognition technology to identify students' expressions and assess their concentration in class, despite its many potential application scenarios, cannot be considered a sustainable use. Even if the expression features can be computed to indicate the students' engagement, this technology brings about the risk of privacy infringement and thus reduces human agency or even causes aversion to facial recognition. Furthermore, if the situation were to continue throughout the schooling period, students might gradually grow averse to AI technologies (e.g. some technologies are developed to shield facial features from camera detection). In this sense, this kind of AI application in education is inconsistent with the SDGs regarding education.

 

Another example is the AI application during the COVID-19 pandemic. The personal information collected, combined with big data analysis and decision-making technology, was useful in giving an early-warning about the public health situation and in epidemic management. This is a case of the deep integration of AI into public health management, which played a critically positive role in pandemic containment. Despite this, according to relevant specifications of the World Health Organization (WHO) and policies of China's Ministry of Civil Affairs and Ministry of Industry and Information Technology, Cyberspace Administration of China, and National Health Commission of China, the personal information collected during the pandemic can only be used in an appropriate manner when necessary. While in the post-pandemic period, before using such information for purposes other than pandemic containment, the acting party must ask for permission from the individuals again. Therefore, both the excessive collection of personal information and the use of personal information collected during the pandemic for other purposes go against government regulations and are incompatible with the vision of promoting sustainable development of public health.

 

As AI-related technologies are applied to achieve SDGs, they could potentially exacerbate inequality due to limited access provided for population in certain areas. The primary goal for fair sharing of benefits arising from technological advancements will be challenged by many issues to varying degrees: Can students in inaccessible regions have equal access to AI-augmented quality education? How will health codes, a QR code-based medical identification system, and relevant services benefit elderlies who are mostly inept phone users or have never even used phones at all?

 

In addition, AI models could acquire human prejudices from data and build them into applications, which may worsen inequality as well and run counter to greater equity. For instance, an AI-based recommendation system may show marked biases against people of different genders and racial backgrounds in job recommendations as a result of data-driven prejudices, which is at odds with the good intention of applying AI to finding employment. From the perspective of sustainable development, AI technology can be used to identify and tap into strengths of vulnerable populations so as to promote fair employment. Meanwhile, the facial recognition rate for females is significantly lower than that for males, and people of different colors and races face prejudices ingrained in the technology-assisted decision-making systems for public safety. All of the problems above calls for improved AI technology in the future to propel the sustainable development of fairness and equity. For example, employing AI-augmented technologies to enhance the capabilities of people with vision, hearing or speech impairment to perceive and communicate are important attempts through which AI can benefit vulnerable populations.

 

The pure replacement of human labor and certain types of jobs by AI and automation will only have negative implications, unless we do not only focus on the development of full automation and do not neglect the vision of providing decent work for humans in the SDGs. In the future, we must develop new occupations featuring human-machine collaboration, and help human beings and AI and automation systems to leverage their own strengths to achieve win-win outcomes. AI and robotics application should aim at saving time for human beings, and enabling them to perform more creative tasks to capitalize on their strengths. AI technology should also be able to take the burden off of workers and help them maintain work-life balance.

 

When applying AI to SDG-related areas, we need to pay attention to the risks and ethical issues involved, so as to ensure that the technology is applied for a good cause. Only if AI continues to promote the implementation of SDGs while we pay attention to and avoid potential technological and social risks, can AI exploration and application truly be a driving force for sustainable development.

 

An opportunity of inclusiveness for and learning lessons from various cultures

It is a globally shared vision that AI can facilitate sustainable development; the vision is closely associated with culture and geography in many areas related to sustainable development, such as education, fairness and equity, and urban development. The combination of culture and AI technology can offer insights into sustainable development from distinct perspectives, and give us a greater understanding of long-term sustainable development and its implementation. Therefore, we need to embrace different cultures, learn from each other, and seek common ground while putting aside differences, be in harmony, yet be different, and standing together in the face of challenges.

 

The rapid development of AI technology is usually only seen in technologically and economically advanced countries, which is very likely to quickly widen the wealth disparity as AI develops and potentially accelerate the Matthew Effect as a result of AI application. Thus, countries and regions with advanced AI technology should promote the equitable and universal access to AI technology and services through open and shared development and global collaboration, and enable low- and middle-income countries and regions to leverage the AI revolution to facilitate their social and economic development, thereby achieving the SDGs globally as soon as possible.

 

As a responsible country, China is in the frontiers in the R&D and industrial development of AI. Therefore, China stands in a much better position to contribute to globally equitable access to AI. On the basis of the New Generation Artificial Intelligence Development Plan, China should help to further flesh out the SDGs and promote the application of AI in various areas of sustainable development. Meanwhile, AI for SDGs think tanks, online case databases, resource-sharing libraries, and public service platforms should be established for global benifits. Non-profit AI for SDGs research programs should be supported by Industry-University-Research collaboration and accept applications worldwide, so as to make AI an enabling technology to promote the SDGs and make research findings available to the whole world. In particular, China should support low- and middle-income countries and regions and those involved in the Belt and Road Initiative, and collaborate deeply with countries that with more advanced AI capabilities to tackle hard problems for the human races in different areas of sustainable development.

 

Moreover, measures should be taken to encourage innovation, such as national demonstration zones for new-generation AI innovation, national new-generation AI open innovation platforms, pilot zones for AI innovation and application, and AI-related research institutes, to carry out in-depth collaboration, combine top-level designs with strategic planning, and actively utilize AI for sustainable development. Chinese AI enterprises and research institutes should also be encouraged to develop a model of industry-research cooperation. They should also cooperate with other global research institutes and businesses to draw on each other's strengths for open and win-win development, jointly adopt new-generation AI as an enabling technology to achieve sustainable development in social, economic and environmental dimensions across the globe, and build a community of shared future for humankind.

 

(The original article was published in Chinese in Guangming Daily, a national Chinese-language daily newspaper, June 18th, 2020)

 

Link to the original Chinese version:

http://news.gmw.cn/2020-06/18/content_33920719.htm


Yi ZENG is the founding Director for the Research Center on AI Ethics and Sustainable Development, Beijing Academy of Artificial Intelligence. He is a Professor and Deputy Director at Research Center for Brain-inspired Artificial Intelligence, and Co-Director of the China-UK Research Centre for AI Ethics and Governance, both at the Institute of Automation, Chinese Academy of Sciences. He is a board member for the National Governance Committee of Next Generation Artificial Intelligence, Ministry of Science and Technology China. He is an Expert in the Ad Hoc Expert Group on AI Ethics, UNESCO. He is in the Expert Group on the Ethics/Governance of Artificial Intelligence in Health, World Health Organization (WHO). He is the lead drafter of Beijing Artificial Intelligence Principles, and one of the major drafters for the National Governance Principles of New Generation Artificial Intelligence, China. His major research interests focus on Brain-inspired Artificial Intelligence, AI Ethics and Governance, and AI for Sustainable Development.