Artificial Intelligence (AI) technologies, such as machine learning algorithms and generative models, are now integral in sectors like healthcare, finance, and entertainment, enhancing efficiency and fostering innovation. Perhaps the most conspicuous example is the widespread use of generative AI models like ChatGPT, which have quickly become a mainstay of human-computer interactions.

The increased use of AI is primarily driven by its ability to process vast amounts of data quickly, creating a more personalised user experience that removes the need to scroll through pages of results generated through traditional search engines. However, the adoption of these technologies has a hidden cost that many users are unaware of.

AI platforms, particularly those employing ‘large language models’ (LLM) such as ChatGPT, have a substantial environmental footprint. The energy consumption required to train these models and power their ongoing computation is considerable. For instance, training a model like GPT-3 has been reported to emit approximately 552 metric tons of carbon dioxide, equivalent to the annual emissions of 123 petrol-powered cars. The operation of AI models also demands continuous energy input, contributing to ongoing environmental costs.

When comparing AI-driven platforms to traditional technologies there is a significant gap in the amount of energy they consume. A single ChatGPT query consumes about ten times the amount of energy per query as a typical Google search (2.9 watt-hours of electricity for ChatGPT versus just 0.3 for Google). Last year the International Energy Agency estimated that the combined energy usage of cryptocurrency, data centres and artificial intelligence represented almost two percent of global energy demand in 2022 and this could double by 2024.

AI infrastructure also relies on powerful hardware such as GPUs and data servers, which require rare earth metals and other finite resources in their manufacturing. Mining these materials often to habitat destruction, pollution, and significant carbon emissions and the technology contributes to the growing issue of electronic waste at its end-of-life.

But it’s not all bad by any means. Governments are increasingly introducing policies to curb AI’s energy consumption and environmental footprint. In the United States, federal and state-level initiatives require data centres to disclose energy use and adopt efficiency standards while California and New York are proposing regulations to mandate clean energy adoption for AI infrastructure. The UK is also developing AI growth zones aligned with renewable power.

It's also worth mentioning that other parts of our lives like transportation, heating and the production and retail of consumer goods remain significantly more energy intensive than these technologies. Not to mention, AI will likely be able to assist us heavily in improving the energy efficiency of these issues.

Globally, experts recommend standardised measurement protocols, increased transparency in reporting AI’s energy use, and integrating sustainability into AI design to ensure a balance between innovation and environmental responsibility. These are measures that will require significant investment and regulation from government and industry.

So, how can we as individuals manage and reduce the environmental impact of AI-platform usage without getting left behind as it replaces existing technology? Here’s some simple strategies:

• Mindful usage: Limit your use of AI applications to essential tasks, reducing unnecessary interactions that contribute to energy consumption.

• Energy-efficient devices: Use hardware designed for low energy consumption as this will decrease the overall carbon footprint associated with your AI technology use.

• Support sustainable practices: Engage with AI service providers committed to renewable energy sources and transparent reporting of their environmental impact.

• Advocate for policy changes: Support policies and regulations that promote energy efficiency and sustainability in AI development and deployment.

By implementing these measures, users can contribute to minimising the environmental impact of AI technologies while still benefiting from their advancements.