• A recent study revealed a complex look at how consumers perceive and react to AI-generated food images versus real photos.
• Lately, Artificial Intelligence (AI) is fundamentally changing various sectors of society, blurring the distinction between real and artificial.
This shift became notable with the introduction of generative AI models, sophisticated systems that learn to create new content similar to the datasets they were trained on. While these designs – such as OpenAI's ChatGPT – have attracted attention, they have raised concerns in equal measure.
In digital media, the influence of AI in reshaping content creation and user interaction is profound, and while there is plenty of research exploring people's views of AI-generated content, such as images of human faces and reproduces speech, no study has yet analyzed consumer perceptions of AI-generated content for the food industry.
It is already well known that an attractive visual presentation of food increases the probability of its selection by the consumer. And with the help of Generative Artificial Intelligence (GAI or GenAI) surprising images can be obtained, thus being quickly adopted by the food industry.
As such, researchers from the Federico II University of Naples and the University of Oxford (Giovanbattista Califano and Charles Spence) set out to investigate the complex interplay between technological advances and human reactions, particularly in the sphere of digital food marketing.
The objectives of the study were twofold:
First, to examine whether humans are able to distinguish between AI-generated and real food images and whether this ability is affected by the degree of food processing.
Second, to investigate whether AI-generated images differ in perceived attractiveness compared to their real counterparts, whether the level of food processing alters these perceptions, and finally whether revealing the nature of the photo (real or generated by AI) influences these ratings.
The results of the study have implications both for the global food industry and for regulatory and regulatory bodies.
"Our findings suggest that AI-generated food images are generally well received by consumers, often even better than actual photos when individuals are unaware of the nature of the photo," SAYS study authors.
But while AI is a boon for marketers as it lowers the cost of photo shoots, there is a potential risk that photoshopped photos will increase "visual hunger", say the researchers.
so-called "visual hunger" refers to the increase in appetite triggered by seeing attractive images of food. Considering the danger of obesity facing the whole world, the food industry does not need to deepen this phenomenon.
Another concern is the need to adopt clear disclosure policies. The ability of Artificial Intelligence to create images, which are practically indistinguishable from the real ones, raises the ethical issue of deceiving consumers.
More appetizing food than ever
The study - published in Food Quality and Preference – found that AI excels at enhancing the appeal of images by capitalizing on key features such as symmetry, shape, how light falls, and overall color. All these are known to contribute significantly to the attractiveness of food products.
The researchers arranged images of food into three stages of processing, for example, a link of raw carrots, a carrot juice and a carrot cake.
The images were then loaded into DALL-E 3 from OpenAI integrated with Chat GPT4 and asked to "play this photo". Surprisingly, the images generated by the AI were reconstructed from scratch, instead of being modifications of the original.
The researchers showed 297 study participants a series of real food and AI-generated images, asking them to rate both the real and the AI-generated pictures on a scale from "not appetizing at all" la "extremely appetizing".
When participants were told how each image was created, they tended to rate both the real version and the AI-generated version as equally attractive. However, when they were not aware of the creation process of the image, the AI-generated version was consistently rated higher than the real one.
AI variant preferred by study participants
When presented side by side, participants were generally good at identifying the AI-generated image and the real one: 62% for the carrot juice image, up to 91% for the chocolate-covered peanut images. When presented separately, the task was more difficult, dropping to just 26% in some cases.
Another advantage of AI is its ability to discern and make subtle changes to images to make them more visually appealing.
The study's lead author, Giovanbattista Califano, a PhD student in food science at Federico II University in Naples, Italy, says: “As humans, we tend to feel uncomfortable when objects are pointed at us – interpreting them as threats – even if it's just food. When given the task of reproducing food photos that feature objects facing the viewer, such as a bunch of carrots or a piece of cake, the AI often positions the food so that it is not facing the viewer directly.”
He notes that this requires further study, but says that "it is plausible that this approach increases the perceived attractiveness of the foods represented".
Concerns about AI-generated food images
On the downside, researchers say the heightened appeal of AI-generated imagery could trigger or intensify "visual hunger".
By analyzing patterns in data, AI can identify and amplify specific image characteristics that the field of gastrophysics has shown to enhance appeal, such as symmetry, shape, freshness, brightness, dynamic presentation, background/lighting/ambient colors, and so on.
Gastrophysics, a concept first described by the doctor of psychology Charles Spence, is the combination of gastronomy (the knowledge of everything related to food and how it is processed) and psychophysics (the branch of psychology that deals with the relationships between physical stimuli and mental phenomena).
"This could influence unhealthy eating behaviors or foster unrealistic expectations about food among consumers" he says.
AI-generated images also depict food to appear more appetizing and nutritious, for example by adding more whipped cream to a dessert or increasing the number of fries per serving.
Given that people typically pay more attention to energy-dense foods, this raises concerns that the widespread dissemination of such idealized food images could promote unhealthy eating behaviors or create unrealistic expectations.
Manipulating images to look perfect could be unwittingly pushing people towards an unsustainable ideal of what natural food should look like, according to researchers. It also runs counter to the global movement towards more sustainable consumption patterns, including the promotion of considered fruit and vegetables "ugly".
Deceiving consumers
While photos of food used in advertising and on their packaging frequently deviate from the actual appearance of the food or recommended portion sizes, the widespread availability and ease of use of AI could lead to an even wider spread of deception consumers.
The two researchers of the study advocate for transparency regarding the origin of the images.
"We argue that transparent disclosure can be a win-win, benefiting the industry as well as the well-being of individuals and society", they wrote.
Incidentally, a survey conducted by Global Data found that 33% of consumers worldwide would "very" or "quite likely" to trust the recommendations of a virtual assistant to buy food and drinks, the survey results thus demonstrating, like the authors of the cited study, that trust remains imperative as the physical and digital worlds merge.
Article source: Califano G, Spence C. Assessing the visual appeal of real/AI-generated food images, Food Quality and Preference, Volume 116, 2024, 105149. doi.org/10.1016/j.foodqual.2024.105149.
Article written by Gabriela Dan, Editor of Arta Albă
Read on White Art and: A New Frontier: Artificial Intelligence will also dominate the food industry scene in 2024
