Skip to main content
1d ago

NYT Cooking’s AI Assistant: A Helpful Sous Chef That Almost Ruined My Cake

Around 7 p.m. each weeknight, I find myself standing in front of the refrigerator, trying to decide on dinner without hearing complaints of “Chicken a

4 min read13 views5 tags
Originally reported bytheverge

Around 7 p.m. each weeknight, I find myself standing in front of the refrigerator, trying to decide on dinner without hearing complaints of “Chicken again?” Although I should meal plan, I don’t, so this nightly struggle continues. My best weapon recently has been AI. Between Alexa Plus, the Samsung Food app, and an AI robot, AI has become my sous chef, though I must supervise it. It helps with ingredient swaps, generating ideas, and fixing mistakes. I was intrigued when my go-to recipe app, NYT Cooking, introduced an AI assistant.

Many, including some users and purists, are skeptical of this tool, available to subscribers on iOS at launch. I understand their hesitation because NYT Cooking features thousands of tested recipes from reputable sources and real people. I have tried purely AI-generated recipes and found them lacking. The reader comments are another great feature, offering solutions like substitutions for missing ingredients. However, reading hundreds of comments at 7 p.m. isn't practical, which is where the assistant helps. Trained on user comments, it provided practical substitutions and guidance. It won't agree to bad requests (I tested this). It isn't perfect—it almost made me use half the butter in a cake—but it is useful. Importantly, it doesn't alter the original recipe; it creates a new one based on your inputs.

The assistant appears as a small text box on the recipe page labeled “Customize this recipe…” You can ask simple questions like “What can I substitute for this ingredient?” or complex ones like “Scale this recipe to feed seven people, vegan and gluten-free.” The New York Times mentioned it can suggest equipment swaps and convert measurements, though measurement conversion has been in the app for years.

On a Friday night, instead of staring aimlessly at the fridge, I looked at a “One-Pot Chicken and Rice With Ginger” recipe I wanted to try. I lacked chicken thighs and several ingredients. I typed my situation into the assistant: “I have chicken tenders, fresh ginger, garlic, rice, broth, limes, but no chard or cilantro or scallions. I do have green beans and red onions.” It quickly created a new version titled “One-Pot Chicken and Rice With Ginger and Green Beans.” It suggested shorter browning time for tenders, cooking red onions with the chicken, and steaming green beans on top of the rice for the last eight minutes. Every change was highlighted under a “Your Version” tab.

I noticed it didn't adjust for cilantro, calling it optional. I asked if I could use Thai basil, which I had, and it agreed, noting the flavor change. While I could have researched this myself, having validation that my changes wouldn't ruin the dish was helpful, and I didn't have to leave the app. The editors say the AI is trained to refuse requests that will taste bad. The resulting dish was delicious and received my husband's approval. While different from the original author's recipe, the AI version was easier to follow than winging it based on comments.

I cooked a Thai red curry and asked about substituting makrut lime leaves. It suggested grated lime zest but noted the loss of floral aroma. For a Dutch baby caramel apple, I asked if half-and-half could replace cream; it agreed. I asked if Castelvetrano olives could replace kalamata in halibut; it suggested smoked paprika might work. I asked Alexa Plus for a comparison. Alexa offered similar guidance but couldn't show the modified recipe on the screen and got confused. I asked the assistant to convert the Dutch baby to metric, which it did seamlessly. I then tried strange substitutions like fresh cayenne for chile pepper (it warned of sharp heat) or ladyfingers for minced beef in a trifle (it correctly refused).

I attempted a bigger challenge: baking. I rarely bake successfully, and I often lack the right equipment. I asked the assistant to convert a “Cinnamon Roll Bundt Cake” into muffins to avoid needing a Bundt pan. It adjusted the cooking time to 50 minutes and split the batter across two tins. I asked it to halve the recipe for 12 muffins, which resulted in three eggs instead of two and a half. I made both the original and the AI version; both turned out great. However, using sticky fingers to scroll between versions was messy. I asked the assistant how much butter was needed, and it gave the modified amount (6 tablespoons) without specifying which version, which could have resulted in a dry cake.

If you read all the comments before cooking, this tool offers a similar experience in a streamlined way. For everyday problems like wrong cuts of meat or missing pans, it genuinely helped. It knew when to refuse bad ideas like minced beef in a trifle. As an AI tool, it works well as a sous chef but isn't the head chef; I still need to double-check. Purists can ignore it, but I plan to use it more often to avoid missing ingredients. For me, standing in front of the fridge with random ingredients, it’s been a useful addition to my routine.

#AI News#Recipe scaling#Ingredient substitution#NYT Cooking#Smart cooking
ES
Editorial StaffEditor

The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.

View all posts
Reader feedback

What did you think of this story?

User Comments

Filter:
No comments yet. Be the first to comment!
Continue reading
View all news