Is there room for AI in Sustainable Development?

Whether you are for or against the implementation of AI in our daily lives, we are now at the stage of acceptance. Look around, and you will find that most technology and data-based companies are now jumping aboard the fast-moving AI train. The debate between the dangers and benefits of AI will continue to heat up as the technology continues to evolve and advance. And we can argue that this is one of the most volatile pieces of technological advancement we have ever created; it has the extraordinary potential for being harmful, yet it also has the extraordinary potential for good.

AI helps us detect patterns in data; identifying anomalies and similarities allows us to spot correlations. Plus, adding historical databases allows AI to predict future outcomes based on its dataset more accurately. However, one of the key benefits of AI-based research is that AI may be able to spot crucial links that we may have missed. AI-led initiatives such as FireAld, developed by the World Economic Forum, help predict wildfires and allow faster emergency response or preventative measures. The European Space Agency AI-powered Destination Earth (DestinE), a complex climate modeling initiative, allows improved strategies for tackling climate change.

One significant relationship spotted by AI-aided research is between two seemingly unrelated variables: daily temperature variations and household wealth estimates. Using machine learning, researchers were able to spot a correlation between temperature variations and 130 low—and middle-income countries. The results showed that higher temperature variations often led to greater poverty, a crucial factor that may have been overlooked without the use of AI.

Using SDGs as the Key Indicator for AI

So, the question is whether AI can become a key tool in tackling the world’s most pressing developmental issues. One key measure of success would be the extent to which AI technology assists us in achieving our Sustainable Development Goals (SDGs). In May last year, McKinsey & Company published a report titled “AI for Social Good.” The study looked at how AI is already a key part of impacting SGDs and its potential as a catalyst in all 17 SDGs. 

The study concluded that AI would affect all SDGs, but five SDGs, in particular, have the highest potential for AI to be a key difference maker. These are Good Health and Well-Being (SDG 3), Quality Education (SDG 4), Climate Action (SDG 13), Affordable and Clean Energy (SDG 7), and Sustainable Cities and Communities (SDG 11). This reflects the fact that 60 percent of not-for-profit AI for social good initiatives were in these areas.

In December 2023, the United Nations initiated its own research on the implications of AI for social good with its report “Governing AI for Humanity.” The report argues that despite AI’s huge potential for good in science, clean energy, education, and public health, severe risks need to be mitigated. One of the main risks is the sustainability of AI. 

It is important to first distinguish the difference between AI for sustainability and sustainable AI. The term AI for sustainability is the application of AI towards SDGs, whereas sustainable AI refers to minimizing the environmental impact of developing and operating AI systems. Data centers that contain AI servers consume large amounts of water to cool down the servers; a study estimates that global AI-related infrastructure could consume six times more water than Denmark. AI servers also use massive amounts of electricity. A prompt requested through ChatGPT consumes 10 times more electricity than a Google Search, as reported by the International Energy Agency. 

Computer microchips are also expensive due to the rarity of the elements required to produce them; these microchips that power AI need rare earth elements, which are often mined unsustainably, according to UNEP. So, while AI does have a large potential to address our most urgent developmental issues, does the benefit of using AI outweigh its excessive requirements? Does the reduced greenhouse gas emission due to AI application offset its own carbon emissions? These questions are part of a relatively new research field that requires a more prolonged period of time to assess the results of AI implementation towards these SDGs.

Moving Forward: The Need for AI Literacy

It could be that to improve our relationship with AI universally for the greater good, we need to increase global AI literacy. While it may improve the general public familiarity and comfort with the ongoing AI revolution, increasing global AI literacy would be a more effective way to build trust and understanding of the technology. This would allow more people to apply AI for the greater good, ultimately resulting in a more holistic approach to achieving our SDGs.

From 2018 to 2023, a mere 10 percent of grants for AI initiatives were given to organizations based in low or middle-income countries. Looking forward, we need to direct more focus on AI literacy towards benefitting lower-income countries where the need and potential of AI on SDGs are among the highest. The goal is to ensure that we use AI to reduce inequality and not deepen the divide that already exists.