Capability of ChatGPT in Food Sensory Evaluation: A Comparative Study with Human Consumer Panels

This dataset comprises sensory evaluation scores generated by human panelists and ChatGPT across ten different food products. The products represent a diverse range of food categories, allowing for a broad comparison of sensory perception between human and artificial intelligence-based evaluations.

Sensory attributes were assessed using a 7-point hedonic scale, where a score of 1 indicates extreme dislike and 7 indicates extreme liking. Multiple sensory parameters were evaluated depending on the product type, including appearance, odor/aroma, taste, texture, and overall acceptability.

Human sensory data were collected from untrained panelists, while AI-generated evaluations were obtained through independent ChatGPT sessions to ensure variability and reduce contextual bias. Each product–attribute combination includes multiple observations from both human and AI evaluators, enabling robust statistical comparison.

The dataset is structured to facilitate comparative analysis between human and AI sensory perception, supporting investigations into agreement, bias, variability, and the potential application of AI in sensory science.