Italian researchers from the University of Naples explored how artificial intelligence can generate food images that people perceive as more appetizing than real photos. The boost in appeal appears to come from deliberate choices about how items are arranged within the frame. Findings were published in the journal Food Quality and Preference.
In the study, 297 participants rated real and AI-created food images on a scale from not at all appetizing to very appetizing. The images included a range of foods such as fruits, fast foods, vegetables, and sweet drinks.
The researchers observed that AI-generated photos tended to evoke stronger appetite responses from participants. In many cases, participants could not tell whether an image was produced by a human or a machine, suggesting that visual optimization plays a key role. The AI was credited with presenting products in ideal lighting, selecting flattering color palettes, sizing portions appropriately, detailing textures, and keeping symmetry, all of which contributed to an enhanced perception of tastiness.
Another factor noted was the AI’s ability to position items in the frame in a way that aligns with typical viewer expectations, reducing any sense that the scene might pose a threat and thereby increasing appeal.
Earlier observations from the same line of research indicated differences in brain activity related to gender, hinting at nuanced neural responses to AI-enhanced food imagery.
[Citation: Food Quality and Preference; University of Naples study on AI-generated food imagery and appetite response]