The UK government has issued draft guidelines for its employees on the ethical and sustainable use of artificial intelligence, including an unusual suggestion: do not thank AI systems, even when they provide correct outputs.
This guidance, detailed in a document aimed at supporting staff's AI journey, is part of a broader effort to promote efficiency and reduce the environmental toll of AI usage. The advice is not rooted in the AI's sentience, but rather in the significant computational resources and energy consumption associated with these technologies. The guidelines acknowledge that large language models are expensive to run and have a considerable carbon footprint.
Among the three main recommendations for efficient AI use, the government specifically advises: "Use as few prompts as possible (you don’t need to say thank you)." Other efficiency tips include keeping prompts short, clear, and direct, and only using AI when absolutely necessary, suggesting simpler tools like spreadsheets or search engines might suffice for certain tasks.
The draft guidance also covers more standard advice, such as the importance of carefully checking AI-generated outputs for accuracy and bias, and being mindful of input data that could be used for training. Employees are reminded that they are responsible for the work produced, a stark contrast to the more complex accountability structures of AI companies that might attribute errors to "rogue agents."
This directive on politeness comes as the cost of AI operations has become a significant factor for businesses. The initial enthusiasm for widespread AI adoption, driven by narratives like "Use AI or lose your job," has been tempered by escalating costs and the shift towards pay-per-token models. The UK government's advice reflects a growing pragmatism in AI deployment, focusing on cost-effectiveness and environmental considerations.
While the impact of simply not thanking an AI may be marginal on a personal level, the guidance represents a tacit acknowledgment that the pervasive reliance on AI may not be entirely beneficial. It encourages a more critical and economical approach to leveraging artificial intelligence technologies within public service.