ChatGPT captured the world’s imagination with its remarkable ability to generate coherent and detailed replies to any request. In just a few short years, AI is now generating simulated human conversation and counseling, realistic videos and catchy pop music. As these applications drive a boom in data centre construction, people are beginning to question how AI will transform society.

The technology’s incredible ability to organize human knowledge with blistering speed is both exciting and alarming. US Senator Bernie Sanders has good reason to warn that we must fight to ensure AI is going to be good for working people, not just billionaires. We must also be mindful that as powerful AI approaches, the technology could push us over climate, economic and social tipping points.

Professional and business oriented AI “agents” already exist for call centers, tech support chatbots, writing and copy editing, legal assistants and scientific research tools, just to name a few. With all these productivity improvements, one has to wonder how many organizations will be downsizing in the near future. Sanders estimates the impact will be on the order of 100 million jobs in the US alone.

The “dot com boom” resulted in many business failures, but it also created world-changing technologies like e-commerce and social media. With addictive new applications and services, there was also a downside as local shops and department stores were replaced by Amazon fulfillment centres and a legion of delivery vans.

AI will change far more than the way we shop or communicate with our friends. For many vulnerable professions, AI will be what an assembly line robot was to the unionized autoworker.

In 2018, long before large language models became a viable product, a McKinsey Global Institute podcast discussed how technologies such as automation and AI would shift the workforce as significantly as when we shifted from an agrarian to an industrial society. In a more recent report from McKinsey, the authors state, “We find that currently demonstrated technologies could, in theory, automate activities accounting for about 57 per cent of US work hours today.”

The rapid expansion of AI also presents significant environmental challenges. AI data centers are highly energy-intensive. MIT Technology Review states, “data centers in the US used somewhere around 200 terawatt-hours [TWh] of electricity in 2024, roughly what it takes to power Thailand for a year.”

Despite escalating CO2 emissions, the climate benefits and opportunities of the AI boom are manyfold, opening the door to discussion of risk mitigation, such as adoption of greater circular economic principles, write Rob Miller and Donald MacCallum.

The International Energy Agency (IEA) recently forecasted that data centre electricity demand will grow from 460 TWh in 2024 to 1000 TWh by 2030. With more than half of that energy expected to be provided by fossil fuels, the CO2 emissions from data centres will triple to three per cent of global emissions, an amount roughly equivalent to the waste emissions from all the landfill sites in the world. 

However, AI also has the potential to aid environmental monitoring through sensors, satellite data and predictive models, improving emissions tracking, ecosystem management and disaster response. AI can also improve efficiency and reduce resource usage across industries by optimizing supply chains, energy management systems, agriculture, transportation and waste management. 

Despite escalating CO2 emissions, the climate benefits and opportunities of the AI boom are manyfold. For example, AI could assist in designing commercial and industrial systems that incorporate a circular economic model. 

A circular economy is based on the idea that the majority of our consumption and production can be redesigned to fit in a closed loop system that improves the environmental footprint of these systems and retains functionality for a much longer time frame. The circular principles of eliminating waste and pollution, circulating products and materials and regenerating nature represent a massive improvement over the traditional “take-make-dispose” linear economic model.

AI technologies may provide the catalyst required for advanced tools to help model, predict and optimize resource usage, allowing circular industries to become more efficient and competitive with traditional supply chains. Focusing on key strategies and process improvements to reduce power consumption and increase the use of clean energy can go a long way to reducing the air pollution, strain on water resources and e-waste from data centre construction and operations. 

AI may also assist in planning for employment disruptions tied to AI’s rapid scaling, but organizations and governments must first choose to develop retraining programs and strategic policy changes to protect the at-risk workforce. 

In short, the economic, social and environmental strain from AI technology will be equal to the opportunities presented by this transformational technology. This opens the discussion for risk mitigation such as adoption of greater circular economic principles and development of innovative training and employment services. The outcome of AI-related tipping points will remain uncertain, but the risk can be mitigated with policies and initiatives that will be pivotal in keeping the world on track to reach our emission targets while replacing jobs lost to the AI revolution.

Donald MacCallum works at DIANA, a NATO body that supports start-ups advancing sustainable and responsible technology development, with previous experience at S&P Global and the United Nations Capital Development Fund.

Rob Miller is a retired systems engineer who spent most of his career designing digital communications products with General Dynamics Canada. 

© National Observer