Daily Bread for 11.4.22: Can A.I. Write a Thanksgiving Recipe Better Than Humans?

Good morning.

Friday in Whitewater will see occasional showers with a high of 67. Sunrise is 7:32 AM and sunset 5:42 PM for 10h 08m 55s of daytime. The moon is a waxing gibbous with 89.2% of its visible disk illuminated.

On this day in 1922, British archaeologist Howard Carter and his men find the entrance to Tutankhamun‘s tomb in the Valley of the Kings.

November is an achingly beautiful month, culmination of the year’s harvest. That time of harvest brings Thanksgiving, the last holiday before winter. (There’s some fuss over Thanksgiving as a commemoration of the Pilgrims, but I’ve long thought of Thanksgiving rather in the spirt of Lincoln’s declaration in 1863.) 

Thanksgiving dinner, more than any other during the year, seems the quintessential American meal.  It may surprise some to learn that I cook, of all things, now and again. No occasion brings me to the kitchen more happily than Thanksgiving. Anyone, even someone in my own state of ignorance, can work his way around a kitten if he tries.

(A point about this: contemporary American kitchens are large and well-provisioned, but that’s not necessary for a memorable dinner. In the early twentieth century, for example, European and Black families in America prepared their own delicious variations on Thanksgiving dinner without half the size of kitchens today.)

Here, in the twenty-first century, we’ve a new, would-be gourmet in artificial intelligence. Priya Krishna and Cade Metz ask Can A.I. Write Recipes Better Than Humans? We Put It to the Ultimate Test. (‘Researchers are using artificial intelligence to create recipes, complete with appetizing photos and back stories. But Thanksgiving poses a challenge’):

These [A.I.-generated] recipes have all the components of their handmade forebears: lists of ingredients, precise measurements, step-by-step instructions and introductory notes with (fabricated) personal touches. Their advantage, in theory, is that they draw on a vast trove of online information about food and cooking.

But are they any good? And can they improve on millenniums’ worth of lived culinary experience?

As home cooks, professional chefs and food-magazine editors know, the ultimate test for recipes is Thanksgiving dinner, a sprawling, varied spread that invites high expectations.

So we decided to enlist artificial intelligence — in this case, a technology called GPT-3 — to devise a holiday menu, which we then prepared and presented to a corps of taste testers: four New York Times cooking columnists.


Cooking and tasting the recipes all but dashed that hope.

The cake was dense and more savory than sweet. The naan stuffing tasted like a chana masala and a fruitcake that had gotten into a bar fight. The roast turkey recipe called for a single garlic clove to season a 12-pound bird, and no butter or oil; the result was dry and flavorless.

The chaat, laced with cilantro and baking spices, was a grassy-flavored mush. The green beans and the cranberry sauce were edible but unremarkable.

Our taste-tasting columnists were even less kind.

“We’re not out of a job,” Melissa Clark said. “I don’t feel anything eating this food,” Yewande Komolafe added.

Genevieve Ko summed it up best: “There is no soul behind it.”

Humans still win out in the Thanksgiving recipe game. 

Twitter layoffs to start today:

Notify of

Inline Feedbacks
View all comments