Paper short abstract:

Algorithms shape our culinary desires, yet we struggle to digitalize the complexity of taste as a synthetic sense. We will collaborate with AI to imagine Dutch Hummus and examine cultural tensions and our collective embodied sense-making, delving into the nuances of machine translation of food.

Paper long abstract:

Against the backdrop of international migration reaching an all-time high and the global food system blurring cultural boundaries while pushing planetary limits, our culinary desires and preferences are frequently shaped by global trends disseminated by the virality of digital culinary resources, such as recipes and Foodporn images. These trends are steered by algorithms, on media and through Search Engine Optimization (SEO). Natural language processing (NLP) models, such as ChatGPT, DALL·E, are trained on human-generated datasets. When tasked with generating a dish, they mirror back to us the average interpretation of desired food. The resulting texts (i.e., recipes) and images often fall into the Uncanny Valley. Digitalizing a synesthetic sense like taste necessitates the comprehension and translation of all senses—a feat currently beyond the capabilities of state-of-the-art AI.

In this playful co-creation session, we aim to tap into our collective and embodied experience of “Good Food”. To complement the disembodied AI, we will explore the taste experience according to Peirce’s (1839-1914) semiotic theory, as well as the tensions between diverse and culturally embedded Food Values (Lusk, Briggmann, 2009), and the evolving notion of locality and authenticity with the concept of Glocality (Robertson, 1994), as digital foods traverse various geographic and cultural contexts.

Together, we will explore the role of cultural and ideological stereotypes as they manifest in food. Utilizing AI to imagine what Dutch Hummus might taste and look like, what is lost, and what could be found in machine translation of food.

Type: Hands-on co-creation

Duration scaleable: 20-45 minutes