An AI’s Take on my Vietnamese-American Culture: Cross-Cultural Thanksgiving Dinner
Craiyon’s take on cross-cultural Thanksgiving cuisine when prompted with “a Vietnamese Thanksgiving dinner”
About my prompt: “A Vietnamese Thanksgiving Dinner”
So the prompt actually was inspired by a New York Times Cooking video uploaded on YouTube by a NYT Cooking editor – Priya Krishna. Throughout the 30 minute video, she uses Dall-E to generate an entire set of 7 recipes that she later cooks and presents to her fellow editors in the NYT office. In her video, she shared a lot about her preferred flavors, her Indian heritage, and love for spice – and the AI generates some impressive (and even very bad!) recipes for her to cook.
I wanted to see what results I might see if I asked it to interpret my own cross-cultural experience with Thanksgiving. Growing up, I always had a strange spread in a Vietnamese Thanksgiving table – sometimes it’d be duck instead of turkey, hoisin and sweet and sour sauces instead of brown and white gravy, and fried rice instead of stuffing. I was so curious to know “What does an AI think my Thanksgiving dinner should look like as a Vietnamese-American?”
The Results!
It’s so strange! The “food” generated at first glance seems really familiar – golden brown poultry, greens, grains, sauces, red fruits and chilis. I’m really impressed how the Thanksgiving dinner spread really has some East and Southeast Asian-inspired dishes – an interpretation of a seaweed salad with thai chilis, fried rice, steamed cabbages, banana leaves. But looking deeper, everything is actually non-descript…
Is the golden brown poultry duck? or turkey? Even the shape of the meat seem strange to even match the contours of an actual animal. A lot of the side dishes got a variety of colors and shapes to imply an Asian-ness to it, but I can’t actually be exactly sure what the AI is trying to generate. In a weird way, the AI has gotten the “je ne sais quoi” of my childhood inter-cultural Thanksgiving dinner, but it has missed the mark completely on defining any actual dish.
Thank you so much for this post!
The insight into this cross-cultural experience is really interesting! The AI seems to have captured the problems of having a cross-cultural dinner. It seems like it just mixed both cultures without a clear line for how much of which culture or why mixing it. Overall, the AI did not understand the meaning of cross-culture in this sense, it just gave meal options from both cultures.
Interesting topic! I think you were spot on recognizing that the food looks familiar at first glance but then when we take a closer look we realize that many of the items in the photos are actually very non-descript. It is so interesting that the AI got the colors and general “feel” of the southeast asian-inspired dishes right even if the actual details of the sides are not very clear.
So interesting to read your post, Patric. I thought the general ‘look good enough’ factor of Craiyon is just for its depiction of humans, but it seems like for most of its generated images have a somewhat animated look. The resulting images feel like mosaics of American thanksgiving dishes and some Vietnamese dishes put together.