Farmers in Nigeria are finding answers to their questions on irrigated lowland rice cultivation through the use of ChatGPT, a chatbot developed by OpenAI. A recent study evaluated the responses provided by ChatGPT to farmers’ questions and compared them to the responses given by agricultural extension agents (EAs) with and without extension materials.
The study, published in Scientific Reports, analyzed 107 questions from farmers about improving their rice production. The most common queries revolved around the types of inputs needed, such as variety, fertilizer, and herbicide. Among the different areas of intervention, crop establishment, insect and disease management, and weed management garnered the highest number of questions.
Researchers examined the responses provided by ChatGPT and EAs and found that the chatbot’s answers were significantly longer than those given by EAs. On average, evaluators rated the quality of the chatbot’s responses higher than the responses from EAs without and with extension materials.
In terms of rating, chatbot responses were deemed good, while EAs’ responses were regarded as acceptable. There was no significant difference in scores between EAs’ responses without and with extension materials. The evaluators preferred the chatbot’s responses in the majority of cases, with 78% favoring ChatGPT over EAs without extension materials and 69% over EAs with extension materials.
However, it was observed that the chatbot occasionally provided inaccurate information. For instance, the recommended seed rate was too high, planting time was incorrect for the dry season, financial services were not available, soil testing was not recommended, and the number of seedlings per hill differed between seasons when it shouldn’t have.
Despite these occasional inaccuracies, agricultural extension agents recognized the value of ChatGPT. After reviewing the chatbot’s responses, five out of six EAs indicated that it provided relevant answers and could be a useful tool in assisting farmers. All EAs rated the chatbot’s responses higher than their own and expressed willingness to use it for future information retrieval.
This study demonstrates the potential of chatbots like ChatGPT in providing reliable and valuable information to farmers. By utilizing artificial intelligence, farmers can access a wealth of knowledge on rice cultivation, ultimately improving their productivity and yields. Moving forward, it is crucial to refine and enhance the accuracy of chatbot responses, ensuring that farmers receive the most reliable and helpful advice.
