What are the ESG risks of AI?

No conversation about AI would be complete without some appreciation of the enormous benefits AI presents. Previously time-consuming tasks have the potential to be automated or augmented, presenting enormous value and increased competitiveness to companies that harness its potential.

AI has already been proven to improve cancer diagnoses, assist scientific research, detect extreme weather events, tackle hazardous tasks, and has the potential to address complex societal challenges.

But these benefits come at a cost. For one, AI threatens to disrupt entire industries leading to large-scale job losses. According to the OCED, 27% of jobs are in occupations at high risk of automation, including non-routine high-skilled roles in finance, medicine and law, among others. There is ongoing discussion about whether this will lead to large-scale unemployment and at what speed at which this job displacement may occur, but it’s a risk that companies will need to take into account.

While there is no shortage of predictions about the productivity benefits of AI, concerns about its environmental and societal impacts are widespread. For example, the data center industry, which powers AI applications, is estimated to contribute 2%–3% of global greenhouse gas emissions.

Further complicating the matter is that different AI applications have different carbon footprints depending on the computing power they require to run. And with the volume of stored data growing exponentially leading to greater energy consumption and e-waste, the environmental impact of AI is predicted to grow.

Other concerns relate to individuals’ rights to non-discrimination, personal data protection and privacy. As AI requires access to large datasets to “learn”, and this data may include personal information, there are concerns about who has access to this data, how it is stored, and how it will be used.

Another issue is that AI can reflect human biases in their activities. There have already been several high-profile instances of AI discriminating based on ethnicity and gender such as in calculating insurance premiums, screening job applicants, and assessing the likelihood of criminals reoffending.

One reason for this is that due to the complexity of AI, it can be difficult to understand how an AI arrived at a result. This concept is referred to as AI being a “black box” in which the features that make the AI work are hidden from its users. Data and prompts are fed in and results come out without certainty about how those results are determined. While researchers are working to make the ways that AI functions more explainable, this uncertainty presents significant risks.

As a result, companies leveraging AI or investing in ones that do, need to adapt their approach to ESG, enabling them to take advantage of AI’s benefits while mitigating ESG risk.