https://www.youtube.com/watch?v=yT8KoWpqUgg
Shout-out to my YouTube algorithm for suggesting this content.
In the video, chef Priya Krishna uses OpenAI's GPT-3 model to generate Thanksgiving recipes specifically tailored to her culinary background and personal taste. She tells the model, for example, that she grew up in Texas, is Indian-American, and likes spicy food. The model suggests a recipe for Pumpkin Spice Chaat, among other interesting concoctions. At the end, the chef-judges called in to taste the food joke that AI won't be replacing them any time soon. They also comment on how "soulless" it all tastes, which I found interesting. Food is a "medium" that is often affectively imbued with senses of comfort, love, joy. Cooking is also deeply steeped in regional and cultural traditions. To me, someone who cooks South Asian food regularly, the AI's strange "decisions" regarding which spices to incorporate into certain dishes and how much of them to use did indeed feel soulless. In fact, I was reminded of Cheney-Lippold's discussion of Shimon, the jazz playing robot. According to Cheney-Lippold, "No algorithm nor abstraction can fully represent any experience or history...the complexity and legacy of Black identity in musical performance is flattened out by algorithmic interpretation" (We are Data, 72). A similar "flattening" seemed to take place in the NYTimes test kitchen.
This was quite the Sunday morning watch. Thank you for sharing this, Mahnoor! I will admit the thought of pumpkin spice chaat did make me do a slight gag, but I think this video as an exercise was a fun and good commentary on Artificial Intelligence. I really do enjoy Priya Krishna's videos and her remarks and reactions to the AI tool were telling. One thing this made me think about was epistemologies and measurement in South Asian cooking. In addition to your notions of joy and comfort, I associate South Asian cooking to ideas of feeling, emotion, and a lack of preciseness (in a positive way). If I ask my mom to share a recipe of the daal she makes, I don't think she would know how much of what to tell me to put, whereas I associate Western/European gastronomy to an extreme level of precision. Like even when it comes to AI and automated processes of cooking, due to the ability to measure ingredients to replicate processes in western cooking, it makes more sense, but it presents challenges to non-Western styles of cooking.
ReplyDelete- Pratik (non-core #6)
This is funny in the sense that it's a literal example of algorithmic taste making LOL.
ReplyDeleteMarisol Vasquez Non-Core Post 6 / 10 (since I made a few short comments, I supplemented some of them with more posts).
Delete