As conversations around artificial intelligence evolve, a peculiar notion has emerged: could omitting words like “please” and “thank you” from prompts to AI systems contribute to environmental sustainability? While the premise appears reasonable at first glance—given that AI processes text incrementally and longer queries may demand more computational resources—the reality is far more nuanced.
The Misconception of Politeness and Energy Consumption
The idea that fewer words could lead to a greener planet stems from a valid observation: the computational requirements for processing AI prompts do indeed correlate to energy usage. OpenAI CEO Sam Altman has pointed out that every interaction with models like ChatGPT incurs operational costs, especially when scaled across billions of queries. However, the impact of removing a few courteous phrases pales in comparison to the substantial energy consumption tied to the foundational data centre infrastructure that supports these AI systems.
The crux of the issue lies not in our manners but in the sheer volume and intensity of AI usage. As the adoption of AI technologies grows, so too does their environmental footprint—an aspect that warrants serious consideration beyond the trivialities of phrasing.
Understanding AI’s Energy Footprint
Artificial intelligence diverges sharply from conventional digital services in its energy consumption dynamics. When users access a document or stream a video, the bulk of the energy cost is typically incurred beforehand, as the system retrieves pre-existing data. In contrast, each query to an AI model demands fresh computations, triggering a full cycle of processing that incurs energy costs anew with every interaction.
This characteristic positions AI as a form of infrastructure rather than a simple software application. The direct correlation between usage and energy demand is significant, leading to a scenario where the need for electricity is escalating. Research published in *Science* indicates that data centres are already consuming a noteworthy fraction of global electricity, a figure expected to double by the end of the decade under current growth trends, as warned by the International Energy Agency.
The Broader Implications of Data Centre Demand
Electricity, while critical, is merely one facet of the environmental impact associated with data centres. These facilities also necessitate vast amounts of water for cooling, and their construction has extensive implications for land use and material consumption. The local consequences can be profound, even in regions with abundant renewable energy resources, such as New Zealand.
In such scenarios, the existing infrastructure can experience strain from new demands. The claim of renewable energy availability does not always translate to an increase in new generation capacity. Consequently, the electricity dedicated to server operation can limit availability for other essential needs, particularly during periods of resource scarcity.
Moving Beyond Surface-Level Solutions
The notion that polite language could mitigate the environmental impact of AI serves to highlight a growing awareness among the public regarding AI’s tangible footprint. However, this focus on minor behavioural adjustments detracts from more pressing structural issues that require attention.
Key questions revolve around how AI infrastructure is integrated into broader energy planning, the management of its water usage, the prioritisation of land use, and how its demands compete with other societal needs. Recognising AI as merely software overlooks the complex web of resource management that it necessitates. Instead, viewing AI as a persistent element within our physical systems allows for a more informed discussion about its integration into our energy and resource frameworks.
Why it Matters
The popularity of the idea that omitting polite phrases could aid environmental efforts is not merely an oversight; it reflects a deeper intuition that AI technologies are not as ephemeral as they might seem. Acknowledging this reality opens the door to essential dialogues about how AI can coexist with existing systems that are already strained by climate change, population growth, and competing resource demands. As society continues to navigate these challenges, understanding the multifaceted impact of AI will be crucial in shaping a sustainable future.
