Artificial intelligence in impact assessment: the state of the art

Abstract

Artificial Intelligence (AI) is beginning to reshape impact assessment (IA), offering both promise and disruption. This paper reviews the current state of the art, situating recent advances within a broader historical trajectory. Although early applications of AI in environmental assessment date back to the 1990s, the advent of large language models (LLMs) has driven an exponential rise in interest and experimentation since 2023. Evidence from literature, the IAIA25 conference, and a targeted survey of practitioners demonstrates that AI is already supporting efficiency gains through automation of repetitive tasks, improved data integration, and enhanced accessibility of complex information. At the same time, concerns persist regarding reliability, bias, transparency, and unresolved issues of accountability. Opportunities lie in fostering inclusivity, innovation, and capacity building, while threats include loss of trust, over-reliance, and ethical risks. We conclude that the future of AI in IA will depend critically on safeguards, governance, and professional standards to ensure that benefits are realised while risks are managed.