What is the environmental impact of digitally enhanced education? Findings from a life cycle assessment of educational scenarios at an Italian university

Abstract Purpose This study is among the first to evaluate the environmental impacts of higher education scenarios with different levels of digitalization using a standardized Life Cycle Assessment (LCA). Methods A cradle-to-gate LCA was conducted with a functional unit of 10 contact hours and 15 independent study hours per student, including lecturers’ preparation and administrative activities. Foreground data were collected at Politecnico di Milano, while background data came from EcoInvent 3.10 and literature sources. Impacts were assessed using ReCiPe 2016 Midpoint (H) V1.07. Sensitivity analyses and Monte Carlo uncertainty analyses (95% CI) examined the effects of commuting, online attendance share, and the energy use of AI. Results The face-to-face scenario produced 17.2 kg CO₂ per student, with commuting responsible for 85.2% of emissions. Fully online education reduced emissions to 5.87 kg CO₂, though household energy use increased by 77%. The hybrid scenario (75% in-person) resulted in 14.3 kg CO₂. AI generated 0.4 kg CO₂ for 125 queries, nearly twice the impact of 10 hours of live streaming. Toxicity-related were relevant across all scenarios. Sensitivity analysis confirmed commuting as the main GHG driver in face-to-face and hybrid models, while generative AI increased GWP by 7.6% in the online scenario. Monte Carlo results showed the online scenario had the lowest impacts and uncertainties, whereas the hybrid model exhibited the highest variability. Conclusions Environmental trade-offs in higher education are dominated by transport and energy use. Fully online education consistently achieved the lowest impacts, with behavioural and mobility factors representing the main sources of uncertainty.