Generative artificial intelligence is playing a significant role in the transformation of digital ecosystems by reinventing the processes of content generation, process automation, product innovation and customer experience. At the same time that these technologies are becoming more integrated into routine operations, the focus has shifted to the ethical and environmental consequences associated with their widespread application. An investigation of the operational sustainability associated with the generative artificial intelligence systems would be crucial, as it would provide information about how these systems match ideals such as efficiency, circularity and environmental responsibility. We explore how users understand and engage with sustainability principles, specifically lean, circular and green operational frameworks within generative artificial intelligence environments. We collect user reviews of 72 recently launched generative AI platforms from 2022 to 2024 and utilise advanced machine learning methods, including Word2Vec modelling, sentiment and regression analysis, to reveal how text datasets reflect customer perceptions. We find that the lean theme is the most prominent feature of operational sustainability, with the highest sentiment score, followed by the green and circular themes. Our findings show that there is a growing respect among the general public for artificial intelligence systems that exhibit responsible and efficient design.

Ordered backlogs are the confirmed and yet to be delivered customer orders, which may have multiple advantages to firms. We investigate how ordered backlogs influence firms' future environmental misconduct. Drawing on organisational slack theory, we argue that ordered backlogs as a form of demand-side slack can reduce environmental misconduct. We combine data from Violation Tracker, Compustat, ExecuComp and IBES to construct a longitudinal panel of US publicly traded firms spanning the period from 2000 to 2024. Using high-dimensional fixed effects models, we find that higher levels of ordered backlogs are associated with significantly lower future environmental misconduct. Building on resource orchestration theory, we further examine how internal resource configurations condition this relationship by altering managerial discretion and resource deployment. We show that the negative association between ordered backlogs and environmental misconduct is attenuated in firms with higher R&D intensity, greater gross profitability and higher current asset intensity, suggesting that active resource deployment can override the stabilising influence of demand-side slack. On the other hand, buffering capacity based on inventory makes the negative link between ordered backlogs and environmental misconduct even stronger. We use the Gaussian copula method to deal with endogeneity concerns. Our findings demonstrate that operational demand conditions and internal resource orchestration jointly shape firms' environmental behaviour.