Productivity Narratives and Environmental Blind Spots: A Systematic Review of How Organizations Weigh Environmental Criteria When Adopting Generative AI

Generative AI (GenAI) spreads through organizations faster than any shared method for weighing its environmental cost. Two literatures study the phenomenon from opposite ends. Adoption research explains why firms take up GenAI; environmental research measures the energy, water, and carbon it consumes. Whether the first treats the findings of the second as decision criteria remains unexamined. This systematic review asks how organizations incorporate environmental criteria into GenAI adoption and tests the proposition that productivity and modernization narratives dominate adoption and limit environmental governance. A TITLE-ABS-KEY search combining GenAI, adoption, and environmental terms yielded a corpus of 59 sources; main findings were extracted for 46 and grouped by the role each study assigns to the environment. Fourteen studies explain adoption through readiness, institutional pressure, top management support, and ethical leadership. None records ecological cost as a determinant, while six model ethics explicitly. Eleven studies treat environmental performance as an outcome of adoption. Twenty-one measures the environmental burden or proposes instruments to govern it, yet only six tie those instruments to organizational decisions. The evidence supports the proposition with one qualification: environmental criteria are absent from adoption models, which is different from absent from practice. The review separates two meanings of "environmental" that the literature conflates, consolidates candidate decision criteria, and sets an agenda for testing ecological cost as an antecedent of adoption.