Made With AI
Much has been said of how game-changing the field of artificial intelligence is shaping up to be, what with the increasing popularity of AI-backed image/word generative suites and AI’s integration into search engines like Google and Bing.
Portended to be the next big thing in the internet-of-things sphere, different people have been using various AI suites for all sorts of purposes, and among the most recent to grab the attention of many is Pinoy developer Christopher David, who used AI-suites to come up with digital renderings and recipes for novel Filipino dishes.
“Lost Dishes of the Philippines”
Late last month, the public Facebook group Midjourney Official shared a post by David which showed digital renderings of what he calls the “Lost dishes of the Philippines”.
The post was made of eleven high resolution images of Filipino dishes that don’t exactly exist, but could be made using traditional Filipino cuisine ingredients and cooking techniques as “envisioned” by the AI suites Midjourney and GPT-4, with David’s inputs.
Dishes like “Talukab Gisa” (made with carabao meat, water spinach, pili nuts and puso ng saging), “Payanga” (made with mashed ube and coconut milk, flavored with panutsa and pandan leaves) and “Bukid Batala” (made with mixed root crops like kamote, ube and gabi that’re cooked in a calamansi and coconut milk sauce) were “developed” by the AI-engines in “response” to cues that were keyed-in by David.
David – who earns a living in developing projects that fall in the metaverse, augmented reality and virtual reality fields – “guided” the AI tools into “coming up” with novel dishes, which ranged from main courses and desserts.
He used GPT-4 to “craft” recipes and dish names, then fed the ideas into Midjourney, which visually drew up the specials. The digital renderings were quite detailed, so much so that many who came by them had to question if they were photos of actual dishes and not digitally rendered by an image generation suite.
Comments and Reactions
Reactions to the dishes and their renderings has largely been positive, though there were those who note that some of them couldn’t be described as “new”.
The dish “Balaw Sapal”, for example, had commenters say that it bears similarities to a dish called “tambo”. The AI version of the dish is a soup made with young bamboo shoots, corn and coconut milk and chicken, while the “tambo” from Iloilo is practically the same, only that it is prepared with shellfish instead of chicken.
Others pointed out that some of the names of the dishes were off, like the AI dish “Talukab Gisa”. “Talukab” refers to the shell of a crab, and the name of the AI dish appears to be skewed since it is made with carabao meat, not crab.
The name of the AI dish “Tinikim na Manok” also got the attention of foodies, since it translates to “tasted chicken”, which doesn’t sound like an appropriate name for a chicken dish.
There were others who also brought up an intriguing question that’s often asked when AI suites are involved in projects: “Who gets the credit for “developing” the dishes? Should it go to David, to Midjourney and GPT-4, or to the three of them?”
Comments and reactions aside, the development has certainly got the attention of many towards the practical use and implications of AI in the culinary field – particularly in its potential in the making of novel dishes and specials.