It was with some trepidation that we decided to bring the matter of food to the great AI Oz. After all, what does a large language model that can’t eat and knows nothing of the tastes and textures of food bring to the ongoing culinary conversation that is “Talk of the Table”? Nevertheless, we couldn’t resist.
We started with a simple query, requesting that ChatGPT tell us its favorite Jewish food. Predictably, it could only give a milquetoasty answer. “As an AI language model,” it said, “I don’t have personal preferences or tastes. However, there are many delicious Jewish dishes that are popular and well-loved by people around the world.” ChatGPT went on to list a few standards such as challah and chicken soup. It had a similar response when we asked which Jewish foods it would like to try. “As an AI language model, I don’t have physical senses or the ability to taste or try food. Therefore, I don’t have the ability to desire or choose a specific dish. However, I’m here to provide information and answer any questions you may have about Jewish cuisine or any other topic.”
We then took a different approach. Since ChatGPT can aggregate information and produce content faster than humans can, we assigned it to “write a ‘Talk of the Table’ column that could appear in Moment Magazine.” It did so immediately, producing a “story” titled “Memories Shared, Traditions Savored: A Shabbat Gathering.”
“As the sun dipped below the horizon, casting a warm glow upon our table, our hearts filled with anticipation. It was time for our weekly Shabbat gathering, a cherished tradition that brought our family together to honor our faith and connect with one another on a deeper level…The aroma of freshly baked challah wafted through the air, a tantalizing reminder of the love and devotion that went into each braid. As we gathered around the table, generations intertwined, stories of the past wove seamlessly with the present. The ‘Talk of the Table’ commenced, an exchange of memories, insights, and laughter. Each person had a unique story to share, a morsel of wisdom passed down from generation to generation…”
ChatGPT yammered on like this for several more paragraphs, proving in this round that a highly literal interpretation of “Talk of the Table” could be unbearably trite and revealing that the “author” had not “read” any actual “Talk of the Table” columns, even though there are years’ worth of them accessible online. (Food writers: You are not in danger of losing your jobs, at least not yet!)
Clearly, ChatGPT was uniquely unqualified to write about food, since as it had told us, it lacked not only the ability to taste but the capacity to desire, two abilities surely needed to seduce the senses through prose. But we persisted. That’s when we began pumping it for Jewish recipes. We started with challah, which also led to deeply uninteresting results: ChatGPT regurgitated a recipe that was perhaps technically competent, but, like many of its other answers, devoid of any illusion of “self”-expression or personality.
So, we decided to reframe the question and ask for “a totally original recipe for challah, that nobody’s ever made before.” Finally, ChatGPT generated something that caught our attention: a recipe for “Spiced Maple-Bacon Challah Twist.” Let us be clear, this is a pork and dairy challah recipe that includes ingredients such as 4 slices of cooked bacon, crumbled; ½ cup of pure maple syrup; and 2 tablespoons of unsalted butter, melted.
“The Spiced Maple-Bacon Challah Twist combines unexpected flavors and textures, with the smoky, salty notes of bacon harmonizing with the warmth of spices and the sweetness of maple syrup,” ChatGPT explained. “This unique and unusual twist on traditional challah is sure to surprise and delight your taste buds, making it a standout centerpiece on your table. Embrace the culinary adventure and relish in the joy of experiencing something truly out of the ordinary. Enjoy!” Detailed recipe instructions included spreading half of the maple-bacon filling evenly over the rolled-out dough and sprinkling the crumbled bacon on top.
Although the choice of bacon challah made us question the quality of ChatGPT’s Jewish culinary training, we were greatly amused by its outside-the-box thinking. We quickly got to work, prompting ChatGPT to design other new recipes. We suggested a matzah ball soup and were delighted when it came up with an inventive non-treyf vegetarian “Spiced Coconut Curry Matzah Ball Soup.” “This unique twist on traditional Matzaoh [sic] Ball Soup infuses it with the exotic flavors of coconut milk, curry spices, and fresh herbs,” we were told. “The result is a vibrant and comforting soup that blends Jewish and global culinary influences. Enjoy the fusion of flavors and the warmth it brings to your table.” In addition to ingredients you might expect, ChatGPT threw in cumin, turmeric, coriander, curry powder and fresh ginger, plus, for a touch of heat, cayenne pepper.
We also enjoyed its suggestions for spicy mango and ginger gefilte fish rolls, which really don’t seem that far out, and gluten-free lavender matcha cheesecake, which actually sounds pretty yummy.
Now, we haven’t tried any of these recipes and don’t think you should either. We certainly can’t vouch for their edibility. But in case you want to contemplate sacrilege (or need a good laugh), you can head to momentmag.com to find the recipe for “Spicy Maple-Bacon Challah Twist” and others. More importantly, we want you to ask your favorite large language model for new takes on Jewish recipes and send them to us. (Be warned: ChatGPT can produce decidedly verbose instructions.) You never know, someday a new beloved Jewish culinary staple may emerge though artificial intelligence.
SPICED COCONUT CURRY MATZAH BALL SOUP
By ChatGPT
INGREDIENTS
FOR THE MATZAH BALLS:
1 cup matzah meal
4 large eggs
1/4 cup vegetable oil
2 tablespoons chopped fresh cilantro
1 teaspoon ground cumin
1/2 teaspoon turmeric
1/2 teaspoon ground coriander
1/4 teaspoon cayenne pepper (adjust to taste)
salt and pepper to taste
FOR THE SOUP:
6 cups vegetable broth
1 can (13.5 oz) coconut milk
1 small onion, finely chopped
3 cloves garlic, minced
1 tablespoon grated fresh ginger
1 cup diced carrots
1 cup diced zucchini
1 cup chopped kale or spinach
fresh cilantro and lime wedges, for serving
INSTRUCTIONS
1. In a large bowl, combine matzah meal, eggs, vegetable oil, chopped cilantro, cumin, turmeric, coriander, cayenne pepper, salt and pepper. Mix well.
2. Cover the bowl and refrigerate for 30 minutes.
3. In a large pot, heat a drizzle of vegetable oil over medium heat. Add chopped onion and sauté until translucent.
4. Add minced garlic, grated ginger and curry powder to the pot. Stir well and cook for an additional minute to release flavors.
5. Pour in the vegetable broth and let it simmer for about 10 minutes to allow flavors to meld.
6. After the matzah ball mixture has chilled, remove it from the refrigerator. Shape the mixture into small balls, about 1 inch in diameter. Wet your hands with water to prevent sticking while shaping the balls.
7. Carefully drop the matzah balls into the simmering soup. Cover the pot and let them cook for about 20-25 minutes, or until they become light and fluffy.
8. Add diced carrots and zucchini to the pot and continue to simmer for an additional 5-7 minutes, or until vegetables are tender.
9. Stir in the coconut milk and chopped spinach or kale. Let simmer a few more minutes.
10. Taste the soup and adjust seasonings as needed, adding more salt, pepper or curry powder as needed.
11. Ladle soup into bowls. Garnish with fresh cilantro and serve with lime wedges on the side for a burst of citrusy flavor.
(This recipe has been edited by a human.)
Send your AI recipes to editor@momentmag.com or comment online.
I’m going to pass on that recipe. Doesn’t look appetizing (to me) at all.
You forgot to apply the Kosher Filter!
Fun piece, and also one that offers a cautionary tale about reliance on non-human “intelligence.”
From Nutrition & Dietetics Smart Brief:
“Some consumers trying out artificial intelligence chatbots to plan their meals are finding that the bots can be useful, but they also may produce odd results like strange recipes, unappetizing food combinations and nutrition information that is outdated or not accurate. Users should take the recipes with a grain of salt, some experts say, especially if it suggests meals with ingredients that could be detrimental to one’s health. Full Story: The Wall Street Journal (8/14)”
Well said — those recipes from Chat were perhaps entertaining, but all were total turn-offs…