Artificial intelligence and software

The resource consumption and greenhouse gas intensity of digitalisation are influenced not only by IT hardware but also by the software used. Software products with comparable functionality utilise processors, memory or Internet broadband with differing degrees of intensity. As a result, there are clear disparities in the energy consumption of hardware, with shorter battery runtimes and increased use of hardware capacity. Artificial intelligence (AI) also has environmental impacts, which arise during the manufacturing of technologies and equipment. In addition, AI systems are now being used to manage processes in the analogue environment to an ever-increasing extent, potentially causing environmental damage and carbon emissions if they reinforce carbon-intensive user behaviour.

The Oeko-Institut investigates the environmental impacts of software use and of AI design and use. In their comprehensive sustainability assessments of AI, the Institute’s experts set energy and resource consumption against potential benefits such as reduced demand for energy and materials. They also consider factors such as platform independence, offline capability and transparency of data formats and source code. In parallel, the Institute’s researchers conduct comprehensive research on the opportunities for legal regulation of the environmental and climate risks associated with AI.