Carbon emissions from cars have been a political and social issue for three decades, there are reporting obligations for manufacturers. A similar approach might be taken into consideration that is spreading at an enormous speed and has impact on the climate – AI.
Technically, AI is an application based on adaptive algorithms for a number of objectives from self-driving cars to translation tools and automatic image recognition to optimizing logistics. Climate research is now beginning to measure carbon footprint for this too.
A study with contributions from Berlin-based climate research institute MCC provides a framework for this.
The study is published in the journal Nature Climate Change.
Meanwhile, in terms of impact, AI is comparable to hammer. It can perform beneficial things, but can also cause damage.
Thus, it is high time it is steered in the right direction through wisely curated rules. This is true in terms of effects on the labor market or data protection, but also in terms of large extent with regard to climate.
For the first time, from the study, an analytical model to guide policymakers to capture the various effect of AI on emission of greenhouse gases to the fullest is provided.
The impact of AI on greenhouse gas emissions fall in three categories:
- Direct impact, i.e., carbon emissions from operation of servers, end user devices, and data centers for AI development and use,
- Immediate impact of specific AI programs on emissions of greenhouse gas in various spheres of everyday life and economy,
System-level effect of AI by means of structural change, e.g. through elevated demand for certain services and products, lifestyle changes, and new dominant suppliers in specific markets.