Examining preference in chatbot answers related to ecological restoration prescriptions

Abstract Introduction Artificial intelligence (AI) tools, such as chatbots, are becoming more accessible and are considered for various decision‐making purposes. Their ability to generate objectives, strategies, timelines, and budgets in seconds is unprecedented and enticing, especially for brainstorming new projects. Objectives This article explores the frequency of prescriptions made by mainstream chatbots (ChatGPT, Copilot, and Gemini) for ecological restoration plans and discusses how the prioritization of certain categories of restoration may affect environmental restoration in North American ecoregions. Methods Qualitative thematic analysis was employed to determine the three most prioritized restoration prescriptions by chatbots across 45 AI‐generated ecological restoration plans. Additionally, prescriptions were organized into general categories based on their restoration types and tested for associations with certain ecoregions and frequency of mentions in generated responses. Results We found chatbots generated prescriptions most frequently for vegetative restoration (i.e. plant native species, remove invasive plants). Prescriptions such as enhancing soil health, enhancing wildlife habitat, engaging with the community, and reducing erosion were also frequent, but to a lesser degree than vegetative restoration prescriptions. All three chatbots recommended similar restoration prescriptions at similar frequencies, suggesting they may be sourcing similar data or performing comparable analysis when generating recommendations. Conclusions We propose that AI tools can be utilized effectively in restoration planning so long as their use is consulted as a singular source of information rather than as a definitive guide for effective restoration work.