Magazine on how sustainable AI can be put into practice
Overheated data centers, massive CO2 emissions generated by large language models, extraction of natural resources and automated inequality: It sounds like AI will contribute to the planet’s demise, rather than save it. But all the talk about the sustainability of AI so far has remained largely theoretical and imprecise. In the midst of a climate crisis, where everything is on the line, the discussion deserves more accuracy, scrutiny, and evidence. There are ways to make AI more sustainable, reduce its resource consumption, include affected communities in its development, and respect people’s autonomy.
The new SustAIn Magazine from Algotithmwatch, IÖW and DAI-Labor of TU Berlin gets real and cuts right to the chase. It includes interviews with experts in the field of sustainable AI and practical examples that are already making a difference. Specific criteria on how to measure the sustainability of AI systems, developed by the project “SustAIn: The Sustainability Index for Artificial Intelligence” build the groundwork for a meaningful debate. A debate that includes social, economic as well as ecological aspects of AI. Because the only AI that’s worth developing, is one that doesn’t just respect planetary boundaries, but also doesn’t exacerbate problematic economic dynamics and doesn’t threaten social cohesion.
AI systems are usually too expensive for small or medium-sized businesses, which stifles innovation and enforces market concentration. The software company elevait designs modular AI systems for the automation of business processes. They can be easily reapplied – saving energy and resources by ommitting training cycles. Read more on page 48 (PDF).
AI as community work
AI development is mainly dominated by business interests. People using the technology are almost never consulted which only exacerbates existing inequalities. The DAIR Institute wants to build AI that reflects the interests of marginalized groups. Read our interview with DAIR’s Director of Research, Alex Hanna on page 18 (PDF).
The AI market is growing rapidly, and so is its resource consumption – at an alarming rate. The company Cloud&Heat developed a data-center cooling system which relies on direct hot water cooling and saves energy. The waste heat can be reused for heating. Some countries have already already started subsidizing this form of waste heat use. Read more on page 52 (PDF) .
Read the magazine (PDF) to explore more articles, comics, facts and figures.