The British government has unveiled new guidelines for civil servants aimed at the environmentally responsible adoption of artificial intelligence. Amidst the global surge in energy consumption by data centers, UK authorities are calling for rationality: before delegating a task to a neural network, one must assess whether a standard spreadsheet or office software could handle it just as effectively.
Principles of "Green" AI Usage
The document highlights that every computational operation carries an environmental footprint. Officials are encouraged to select the smallest available AI models if the use of neural networks is deemed necessary. Furthermore, the importance of minimizing "excessive" prompts and simplifying queries is emphasized, which would reduce the load on server infrastructure and, consequently, lower the carbon emissions associated with AI operations.
Contradictory Data
These recommendations have drawn mixed reactions from the expert community. Critics point to a fundamental contradiction: the use of compact or "trimmed" models significantly increases the risk of hallucinations and reduces the accuracy of responses. While the government prioritizes environmental conservation, experts warn that placing energy efficiency over data quality could lead to errors in management decisions and a decline in overall administrative productivity.
Context: The Global Energy Challenge
The UK's approach stands in sharp contrast to US policy, which emphasizes aggressive capacity expansion. Morgan Stanley forecasts that US data center energy consumption could rise to 38 GW by 2027, compared to 10 GW in 2025. While Washington invests in energy infrastructure for AI leadership, London is betting on resource optimization, setting a unique precedent in the race for technological dominance.
