**
In an era where technology impacts every facet of our lives, a curious debate has emerged around the environmental implications of how we interact with artificial intelligence. Some have suggested that dropping “please” and “thank you” from our prompts to AI systems like ChatGPT could help reduce energy consumption. While the idea may seem appealing, experts argue that the reality is far more complex, and the focus should shift to the broader environmental footprint of AI technologies.
The Energy Behind AI Interactions
At the heart of this discussion is the way AI processes information. Unlike traditional digital services, where the energy cost is mainly incurred when data is stored or accessed, AI requires fresh computations for every interaction. Each time you ask a question to an AI model, it performs a complete computation, consuming energy in the process. OpenAI’s CEO, Sam Altman, acknowledges that the sheer volume of queries—totaling billions—contributes significantly to operational costs.
While it’s true that longer prompts may use slightly more energy, the impact of adding a few extra words is negligible compared to the extensive resources required to run the underlying data centres. These facilities, essential for AI operations, consume vast amounts of electricity and require sophisticated cooling systems, highlighting that the real environmental concern lies not in our choice of words but in the frequency and intensity of AI usage.
The Infrastructure Debate: AI’s Footprint
The conversation around AI’s energy consumption must also consider the infrastructure that supports it. Large data centres are the backbone of AI services, and their energy demands are skyrocketing. Research published in the journal *Science* indicates that these centres account for a significant portion of global electricity usage, with projections suggesting that their demand could double by the decade’s end.
Moreover, it’s not just electricity at stake. Data centres also require substantial amounts of water for cooling, and their construction impacts land use and local ecosystems. This becomes particularly evident in regions like New Zealand, which, despite boasting a high percentage of renewable energy, faces challenges in balancing the demands of data centres with local resource availability.
The Misconception of Immateriality
It’s crucial to recognise that AI is not just a digital service; it operates as a physical entity with tangible resource requirements. This misconception—that AI exists in a vacuum without a lasting environmental impact—distracts from the pressing need to integrate AI infrastructure into broader energy and resource planning frameworks.
As AI technologies expand, they introduce new pressures on already strained systems. The challenge is not merely about optimising the way we phrase prompts but understanding how these systems fit within our existing landscapes, energy supplies, and societal needs. Failing to acknowledge the physical aspects of AI infrastructure could lead to exacerbated vulnerabilities in our environmental and social systems.
Rethinking Our Approach to AI
The notion that cutting out polite phrases from our AI interactions could lead to significant environmental benefits is more symbolic than substantive. It reflects a broader consciousness that AI is not as ethereal as it may seem. This awareness is a starting point for more meaningful conversations about AI’s role in our world.
To navigate the complexities of AI’s environmental costs, discussions must move beyond trivial behavioural changes. Instead, we should be asking critical questions: How will AI infrastructure be integrated into our energy strategies? What measures are in place to manage its water usage? How can we ensure that its location aligns with land-use priorities?
Why it Matters
Understanding the environmental implications of AI is crucial as we continue to embrace these powerful technologies. While it’s easy to focus on small tweaks in our interactions, the real conversation lies in how we manage the growing energy demands and resource allocations that AI necessitates. Recognising AI’s physical footprint is essential for fostering a sustainable future, ensuring that technological advancements do not come at the expense of our planet’s health. Engaging in this dialogue is vital as we seek to balance innovation with responsibility, paving the way for a more sustainable integration of AI into our societies.
