AI recommendation errors led to the loss of sesame crops in China.

A 67-year-old farmer named Wu, facing issues with weeds and pests on his 24.7-acre plot, turned to a neural network for assistance. Despite his initial wariness toward such technologies, previous experience had fostered a certain level of trust in the farmer regarding automated advice.

The chatbot recommended using a mixture of preparations, including "high-efficiency flupyrimethamine," "flusulfasulfaether," as well as "thiamethoxazine" and "methyl salt." As a result of applying this composition, the sesame crops died overnight: the chemicals destroyed not only the weed vegetation but also the crop seedlings, with the latter suffering first.

Experts confirmed that the critical mistake was the use of "flusulfasulfaether." This herbicide is intended for controlling broadleaf weeds; however, sesame is also classified as a broadleaf plant. Furthermore, the application technology for this substance implies spot spraying, whereas the farmer treated the entire field.

The service used by Wu contained a warning that information generated by AI may be inaccurate and requires verification. The farmer ignored the need to verify the data, failing to use search engines or consult with specialized experts. As experts note, in such situations, verifying neural network recommendations is a mandatory step that allows one to avoid serious financial losses and the loss of the ability to use the land plot during the current growing season.