Artificial intelligence (AI) is increasingly being promoted as an instrumental part of the “twin transition,” in which the digital and the green transition are seen as mutually reinforcing. However, there is a need for better methods and models for understanding the environmental impacts of AI systems. While positive impacts of AI for climate change mitigation and adaption are likely, these should only be pursued if they produce a net positive effect, and this requires a thorough understanding of both the climate-related impacts and impacts on, for example, resource depletion, water use, land use, and ecotoxicity. One method for unraveling these impacts is life cycle assessments (LCAs), and in this chapter, we identify and discuss methodological aspects that are relevant when performing LCA of AI systems. The study aims at contributing to taking the first steps toward methodological development and application of LCA to assess the environmental sustainability of AI systems.