This post is about testing recipes created by artificial intelligence, Chef Watson of IBM
My college bestie living in New York visited Seoul. One of the reasons why we are so close is that we’re both complete and utter foodies. She is the one who introduced me to Chef Mingom who lent his restaurant on his day off for Chef Watson’s recipe testing project.So, we got together at Mingom’s restaurant and we took over his chef’s knives and oven (Yay!); we were the guest chefs at Del Horno that day.
She picked a recipe from Chef Watson which is inspired by Korean food. Yes, the first thing that springs to mind: Kimchi. But I’d like to let you know before you read the following review: when Watson recommended this recipe, I think he was out of order.
Recipe and ingredients
- 3/4 lb Halibut(Jon Dory in the original recipe)
- 3/4 cup short-grain rice
- 1 lemon, 3 tbsp lemon juice
- 1/2 cup kimchi
- 1/4 cup olive oil
- 3/4 stick butter
- 2 tsp thyme
- bay leaf
- 1/4 cup chopped, toasted peanut
- 1 pepperoncino (fresno pepper in the original recipe)
- 1 clove grated garlic
1. Grind short grain rice to a medium-fine powder. Toast this in a pan over a medium heat, tossing frequently for 3-4 minutes. Transfer to a plate and let it cool.
Dalji: I saw some of Chef Watson’s recipes that don’t make any sense but most of the time they’re understandable. But everyone in the kitchen was in a panic when, in Step Two, Watson told us to cut in between the pear membranes and to release the segments into a bowl. What on earth did that mean? So I looked into the Bon Apetit’s original recipe that Watson was referring to, and discovered that Bon Apetit’s recipe actually used grapefruit. Fortunately, my friend is, like me, a confident woman who keeps going even with this major faux pas in the recipe.
2. Combine kimchi, lemon juice, grated peppers, garlic and 3 tbsp water in a small bowl; set aside.
Daji: Again, kimchi is spicy enough but Watson told us to add additional peppers. My fearless friend tried to add peppers to kimchi but I stop her because halibuts that were used that day was high-quality and I really didn’t waste the ingredients.
3. Melt the butter in a small heavy-based pan over a medium heat, stirring often, for about 6 minutes. Remove pan from the heat and add matchstick-size pieces of fresno peppers; swirl in the pan for 10 seconds to infuse the butter with the fresno pepper flavor. Add kimchi mixture. Return the pan to a medium heat; then bring sauce to a simmer. Add Asian pear segments with lemon juice, thyme, and peanut. Cover sauce and keep warm.
4. Season the halibut with salt and pepper. Dredge the skin side of each fillet in the ground rice made in step 1. Heat olive oil in a large cast-iron skillet over medium-high heat. Add fillets, skin side down. Cook, rotating the pan for even browning. Turn halibut and cook longer.
Dalji: Again and again, Watson made a mistake in this step of the recipe and was unforgettable. Watson told us to dredge the skin side of each fillet with kimchi…. whaaaat? We had to guess what Watson tried to say and it turns out that the cooking of the halibut part was missing from the whole recipe. He was trying to say halibut. Someone needs to give Watson a cup of coffee.
5. Transfer kimchi to warm plates and drizzle with reserved sauce.
This time, I was really worried about the outcome. This halibut was cooked in the same day I tried to make the avocado paella, which was a total disaster and worst dish made in my cooking history. That recipe mistake made everyone nervous before having a bite of this dish. Actually, the ground rice on the halibut was crispy so that texture was a little fun to chew. However, I couldn’t stop my friend from seasoning their kimchi with additional salt and pepper and, as you can imagine, the sauce was salty, salty, and salty(Sorry to my friend, but this is a review and you didn’t do anything wrong. Watson did.)
However, I started to wonder the algorithm to recommend recipes deeply this day. And here’s the summary of one video I found while searching this subject : How Chef Watson calculates deliciousness.
For those whom it may concern, I’m not related to IBM or Chef Watson. It is my personal interest to test how artificial intelligence decodes the way chefs create recipes and I want to see if this service can be offered to everyone in the near future.