Are AI and ESG Compatible? The answer is yes, but human oversight and management are required.

Environmental Impact. A major issue in AI deployment is energy consumption. According to The New York Times, citing a peer-reviewed study published October 10, 2023, in 50 Joule, “servers could use between 85 to 134 terawatt hours (Twh) annually. That’s similar to what Argentina, the Netherlands and Sweden each use in a year, and is about 0.5 percent of the world’s current electricity use.” Erdenesanaa, Delger, A.I. Could Soon Need as Much Electricity as an Entire Country, New York Times, October 23, 2023.

Where will the additional energy come from? If it is from electricity produced using hydrocarbons, then greenhouse gas emissions will increase unless processing becomes more efficient. Using renewable energy would lower incremental emissions, but clearing land to install solar or wind farms carries its own environmental impacts.

Further, there is the issue of the solar scrap generated when users replace older panels with newer, more efficient ones. Aasavari Joshi wrote in an article dated February 21, 2023, appearing on the website of the Danish think tank, Circular Innovation Lab, that advances in recycling could lead to up to 95% of a solar panel’s value being recoverable and reusable, but for now the technology for recycling is in its early stages and needs to be ramped up.

On the positive side, AI can be used to optimize energy consumption and reduce waste, and, in precision agriculture to reduce input waste. AI can analyze weather data and help predict potentially dangerous environmental events. It can also be used in the development of sustainable materials and technologies. 

Social Impact. AI models can perpetuate discrimination. Algorithms can be trained on data that reflects historical biases against minorities, women, older adults, and people with disabilities. When biased data is encoded into the AI models, discriminatory outputs follow. In addition, if underrepresented groups are underrepresented in the AI development process, blindspots can develop that perpetuate discrimination.

AI also has the power to improve social conditions. Just as AI can amplify biases, it can be developed to detect and propose solutions to mitigate them. AI-enabled language processing can improve the accessibility of knowledge content for individuals who are visually impaired. And it can be used to speed the development of technologies to improve healthcare.

Governance. If AI systems are not developed in accordance with mandatory standards, and subject to oversight and monitoring, it is difficult to systematically identify and mitigate biases. Indeed, biases can purposefully be introduced and amplified. Beyond bias and discrimination, areas that are ripe for governance are AI-powered facial recognition, which can infringe on privacy rights, and content algorithms used by social media that can remove legitimate speech. Ideally, ethical governance principles precede and are incorporated into AI systems development, so that individual rights are enhanced rather than threatened.

On the other hand, AI can inform evidence-based policymaking by analyzing complex data, simulating policy scenarios, and predicting the impacts of different policies. It can also be used to evaluate the effectiveness of governance programs.

AI and ESG can, indeed, be compatible. However, thoughtful governance principles, and responsible development and use, are required to mitigate potentially negative environmental and social impacts. By following good governance principles and continuously evaluating its impact, AI’s can benefit people as well as the planet.