Can artificial intelligence contribute to the implementation of the UN sustainability goals?

The United Nations has adopted the 17 Sustainable Development Goals (SDGs) 2015 as a guideline for global political and social development in the near future. The study published in Nature Communications examined the influence of AI on the 169 individual SDG targets.

"Artificial intelligence is already changing everyone’s lives in different ways," says Francesco Fuso-Nerini of KTH, one of the two leaders of the study. " This analysis provides the basis for a needed dialogue on what kind of future humanity should aim for with AI," says Ricardo Vinuesa of KTH, second director of the study.

The study was conducted under the direction of the KTH Royal Institute of Technology in Stockholm, Sweden, and with the participation of Dr. Sami Domisch from the Leibniz-Institute of Freshwater Ecology and Inland Fisheries (IGB) in Berlin, Germany, and Dr. Simone Langhans, former IGB scientist and currently Marie Skłodowska-Curie Fellow at the Basque Center for Climate Change in Bilbao, Spain, and the University of Otago, New Zealand.

The Sustainable Development Goals take aim at 169 individual targets running the gamut from economy and society to environment. The research results as a whole show that 134 of these targets could benefit from AI, while 59 of the targets could also be negatively influenced by AI. Calculated as a percentage, AI has a positive impact of 79 percent. According to the researchers, AI has the greatest potential to positively support the achievement of the environmental goals.

Positive influence on species and environmental protection

Goal 14 and 15 of the SDGs concern the protection of species in water and on land and the sustainable use of the terrestrial and aquatic environment. "We can use AI-based technologies to better capture biodiversity through automated procedures using remote sensing data. Machine learning plays a major role here. Computer algorithms are trained with existing data to recognize similar patterns using new data," explains Simone Langhans of the Basque Center for Climate Change.

Sami Domisch from IGB adds: "AI technologies can thus be a support in environmental protection. Because with a better understanding of species distribution patterns, the planning of protected areas can be made more efficient. This is particularly important for water protection and the threatened aquatic biodiversity. However, the regulated and transparent use of these technologies must be of great importance here, because AI can also be applied to simplify the use and exploitation of the environment".

Reducing energy consumption

Artificial intelligence shows a clear potential in reducing energy consumption and promoting clean affordable energy. For example, so-called “smart” electricity grids can match and balance the need for electricity against various renewable energy sources. However, overall demand for ICT solutions could increase data centers’ electricity usage to 20 percent of the global total by 2030, the study points out.

Risk of social instability

In the sphere of global political stability and justice, growing inequality, biased election outcomes, hatred for minorities and increased nationalism can result if AI is developed in the absence of ethical scrutiny, democracy and transparency, the study states. On an individual level, social media algorithms that display content for users tend to rely on the receivers’ own preconceived opinions. This type of AI can further polarize societies, Fuso-Nerini says. “There’s an underlying risk of prejudice when AI evaluates and predicts human behavior.”

“Regulatory oversight should be preceded by regulatory insight, where policymakers have sufficient understanding of AI challenges to be able to formulate sound policy. Developing such insight is even more urgent than oversight, since policy formulated without understanding is likely to be ineffective at best and counterproductive at worst,” says Max Tegmark, co-author and professor at the American Institute of Technology in Washington, D.C.

In the field of global political stability and justice, growing inequality, distorted election results, hatred of minorities and increased nationalism can occur when AI is developed without ethical control, democracy and transparency. "Social media algorithms that display content for users tend to rely on the preconceived opinions of recipients. There is a risk of prejudice when AI evaluates and predicts human behavior," says Francesco Fuso-Nerini. This type of AI can further polarize societies.

Appropriate policy and legal frameworks are necessary to apply AI for the benefit of individuals and the environment: "Regulatory oversight should be preceded by a regulated insight where policy makers have a sufficient understanding of the challenges of AI to make informed policy decisions. Developing such an understanding is even more urgent than supervision, as policy decisions taken without this understanding are likely to be ineffective at best and counterproductive at worst," says Max Tegmark, co-author and professor at the American Institute of Technology in Washington, D.C.