Currently, the debate on the sustainability and efficiency of artificial intelligence (AI) compared to human activities has gained prominence. Two recent sources address this topic from perspectives that, although aligned in the direction of their conclusions, provide fertile ground for critical analysis.

In Sabine Hossenfelder's YouTube video, she explains how AI can seemingly be more efficient in terms of energy consumption compared to human activities such as writing or drawing. Although initially it may seem like an absurd claim given the vast difference in energy consumption between a human brain and a supercomputer, according to her Nature article, AI can require less energy than previously thought in common tasks such as generating an image from text.

The article directly compares the carbon emissions of humans versus AI systems like ChatGPT, BLOOM, DALL-E2, and Midjourney in writing and illustration tasks. The results indicate that AI systems can be up to thousands of times more efficient in terms of carbon emissions per unit of work produced.

While this piece supports the idea that AI may be more eco-friendly compared to certain human activities, it does not fully address the indirect costs associated with developing, training, and maintaining these AI systems. The energy consumption of data centers, hardware production and disposal, and the necessary infrastructure to support these systems raise important questions about the long-term sustainability of AI.

Moreover, criticism often revolves around whether these studies and discourses are shaped by larger agendas, possibly driven by corporate or technological interests that benefit from promoting AI as a "green" solution. This perspective suggests that by focusing mainly on direct carbon emissions, other broader environmental impacts and social implications such as job displacement, technology rebound effects, and the alignment of AI with living organisms may be overlooked.

It is crucial to question and critically evaluate claims about the energy efficiency of AI, not only in terms of carbon emissions but also in the wider context of environmental sustainability. The narrative that presents AI as a "cleaner" alternative needs to be examined with an approach that considers all aspects of the technology's lifecycle and possible cascade effects on our society and environment.

This analysis does not aim to discredit AI advancements or its potential to contribute positively, but rather to emphasize the importance of maintaining a balanced and well-informed perspective that accounts for both benefits and costs in decision-making processes. This way, the global community is better equipped to make decisions that truly promote long-term sustainability beyond immediate operational efficiencies.