How can Artificial Intelligence support organizational decision-making in sustainability and ESG management?

Sammanfattning

Organizations are facing growing pressure to integrate Environmental, Social, and Governance (ESG) considerations into strategic decision-making. At the same time, Artificial Intelligence (AI) is increasingly being used to support ESG-related work through data analysis, planning support, and predictive capabilities. Despite this, limited research has examined how AI is actually adopted and used within organizational ESG decision-making from an Information Systems (IS) perspective. This study examines the following research question: How can Artificial Intelligence support organizational decision-making in sustainability and ESG management? To explore this question, the study applies the Technology–Organization–Environment (TOE) framework with Trust Calibration Theory as an integrated analytical structure. Rather than predicting technology adoption, this structure maps the technological, organizational, and environmental conditions that enable or constrain AI-supported ESG decision-making in practice, alongside the trust-related dynamics that shape how professionals engage with AI-generated outputs. The study adopts a qualitative research design based on semi-structured interviews with six professionals working across ESG management, AI development, technology consulting, and strategic decision-making contexts. The interview material was analyzed using thematic Analysis. Five overarching themes were identified through the analysis. Across the interviews, AI was consistently described as a tool for supporting and augmenting decision-making rather than replacing human judgment entirely. Participants emphasized that contextual understanding, ethical reasoning, and final accountability remained with human decision-makers. Trust in AI also emerged as something that had to be actively developed rather than assumed automatically. Participants described using verification practices, domain expertise, and human oversight to evaluate AI-generated outputs before relying on them in practice. A recurring issue across the interviews was the condition of the underlying data infrastructure. Fragmented systems, inconsistent ESG definitions, and privacy-related restrictions were frequently described as significant barriers limiting the usefulness of AI in ESG contexts. The findings further suggest that organizational readiness shapes whether AI capability translates into meaningful decision support. Cross-functional coordination, governance structures, and AI literacy appeared particularly important in this regard. Finally, participants described regulatory pressure as both an enabler and a constraint, with some organizations approaching AI adoption primarily through compliance requirements while others were more motivated by efficiency and operational pressures. The study contributes theoretically by applying this integrated analytical structure within the context of ESG and identifying data infrastructure as an underexplored analytical dimension that warrants greater attention in the future. The findings reveal that, while both frameworks remain valuable for comprehending the factors influencing AI adoption and the trust-related dynamics shaping professional reliance on AI-generated outputs, the state of the underlying data infrastructure emerges as a practically significant enabling condition in ESG-AI contexts that neither the TOE framework nor Trust Calibration Theory explicitly addresses. This points to a direction for further theoretical development in IS research on AI-supported decision-making. The study also provides practical implications for several stakeholder groups. For ESG and sustainability managers, the findings highlight the importance of organizational readiness, governance alignment, and data quality when implementing AI-supported decision-making practices. For IT and governance functions, the findings point to the continued importance of human oversight alongside technical implementation. For policymakers, the findings suggest that compliance-driven ESG mandates such as the CSRD may not automatically result in effective AI-supported ESG decision-making when supporting organizational and infrastructural conditions remain underdeveloped.