Greening GenAI: sustainable solutions for higher education

Higher education institutions increasingly embrace and regulate Generative Artificial Intelligence (GenAI). While the cognitive, social, and ethical implications of GenAI are widely discussed in the current scientific discourse, the environmental disruption caused by large-scale GenAI use has remained largely ignored. As GenAI becomes integral to research and education, addressing its environmental impacts has become an increasingly urgent priority. In this Perspective, we argue that mitigating the environmental impact of GenAI requires more than advances in hardware and software efficiency; it also demands a stronger focus on user behavior as a key lever for reducing GenAI’s disruptive environmental impacts. This is particularly important given that every second, millions of prompts are submitted to GenAI systems in support of human learning, research, and knowledge acquisition. Optimizing how users interact with these systems offers scalable behavior-based strategies with the potential to substantially reduce their environmental impacts while requiring relatively modest changes in practice. We present a set of measures to promote the sustainable use of GenAI in higher education and scientific research.