The reliability of AI consulting on the ecological impacts of the escape of farmed fish

Numerous studies have investigated the ecological impacts of escaped aquaculture fish. Whether people can objectively obtain this information through large language model (LLM) artificial intelligence (AI) systems, which are currently gaining increased traction across the globe, is a question of interest to those engaged in or concerned with aquaculture and aquatic ecology. In this study, the reliability of LLM-based AI systems for assessing the ecological impact of the escape of farmed fish is explored, providing reference data for further research on the application of LLM AI systems in aquaculture and aquatic ecology. The results reveal that the answers provided by the AI systems were largely consistent with the findings from the scientific literature on fish ecological invasion, and they included information regarding resource competition, disease transmission, genetic pollution, water quality deterioration, and disruption of the aquatic ecological structure and function. However, the responses of the AI systems did not mention certain aspects discussed in the literature concerning the ecological risks of the escape of farmed fish, including the difficulty in predicting the potential impact of their escape, the influence of the ecological background information of the receiving water area before the escape, and the potential positive impact of fish already established in the wild. This finding suggests that the responses of the AI systems to the questions posed may have leaned towards more general and widely relevant answers while neglecting more specialized but less-discussed details. Therefore, the reliability of the AI ​​consulting in this study is questionable.