Introduction to Kitchen AI
Around 7PM each weeknight, deciding on dinner without repeating the same ingredients can become a nightly challenge. Recently, generative artificial intelligence tools have emerged as potential kitchen assistants to help navigate these last-minute meal hurdles.
NYT Cooking’s AI assistant saved my weeknight chicken, but almost ruined my cake
I tested the new AI-powered tool on ingredient swaps, recipe scaling, and a missing Bundt pan.
The NYT Cooking app now features a built-in AI assistant designed to help adapt existing recipes.
Between smart displays, dedicated food apps, and automated kitchen devices, AI has increasingly become a go-to assistant for home cooks. While these tools require close supervision, they have proven helpful for generating ideas, fixing mistakes, and swapping ingredients. This prompted a look into the new NYT Cooking AI assistant.
Many users and recipe purists have expressed skepticism toward the new tool, which launched for subscribers on iOS. A major appeal of NYT Cooking is its repository of thousands of thoroughly tested recipes from reputable chefs, archives, and food columnists.
Reader comments have traditionally served as a valuable resource for finding ingredient substitutions and tips when specialized items are missing from the pantry.
Previous experience with kitchen AI tools indicates they can be useful when properly supervised.
Reading through hundreds of user comments during a busy weeknight is not always practical. The NYT Cooking assistant aims to streamline this by drawing on community insights to offer quick substitutions and guidance. Importantly, it leaves the original recipe intact while generating a separate custom version.
The system preserves the original recipe while creating a separate customized variation.
Putting the Assistant to Work
The assistant appears as a text box on recipe pages with a prompt to customize the dish. Users can type requests ranging from ingredient substitutions and scaling to equipment swaps and measurement conversions.
1/3
The AI modified version of the chicken recipe. The red highlighted words are what changed from the original.
Testing the feature on a One-Pot Chicken and Rice recipe with missing ingredients demonstrated its practical application for weeknight cooking.
Having immediate validation for recipe modifications helps ensure that improvised ingredient swaps will not ruin the final dish.
Entering available ingredients into the assistant box generated a modified recipe with adjusted cooking times and alternative vegetables. Changes were highlighted in a dedicated tab, allowing users to toggle back to the original instructions.
Further refinements, such as swapping herbs, showed that the AI understands flavor profiles and highlights potential changes. Developers note the system is trained to reject modifications that would produce poor culinary results.
Weeknight chicken and rice.
The resulting dish required the same cooking time as the original and provided a successful alternative to manual substitutions.
Beyond full meal modifications, the assistant handles specific substitution queries, explaining how alternatives like grated lime zest compare to fresh leaves.
1/2
I tried several swaps when I was missing ingredients, asking the assistant if what I had on hand would work.
Substitutions and Edge Cases
Further testing involved complex swaps, such as replacing dairy or adjusting spices in baking and seafood recipes. The tool successfully converted metric weights to US measurements and correctly rejected illogical requests.
While the assistant successfully caught mismatched ingredient requests, it also provided clear warnings regarding intense spice substitutions.
Alexa Plus on the Echo Show 11 gave the same guidance as the NYT Cooking app. But it couldn’t show me a modified recipe.
Comparing the app with Amazon’s Alexa Plus generative AI revealed similar substitution guidance, as Alexa pulls select recipes from the NYT app.
However, Alexa struggled to display modified recipes on screen compared to the dedicated app interface, highlighting the current division between voice assistants and recipe apps.
My cinnamon roll muffins, made with AI assistance.
Baking and Limitations
Testing the tool on baking recipes presented a greater challenge, particularly when converting a Bundt cake into muffins.
The assistant successfully scaled down cooking times and adjusted egg quantities proportionally when halving the recipe yields.
Juggling between original and modified recipes in the app highlighted minor interface friction, as custom versions cannot be permanently saved separately. Double-checking quantities remains essential to avoid scaling errors.
My AI-free cinnamon roll bundt cake.
Conclusions on AI Cooking Tools
For everyday cooking hurdles like missing ingredients or equipment, the AI assistant offers a streamlined alternative to reading through extensive comment sections.
Ultimately, the NYT Cooking AI assistant functions effectively as a sous-chef rather than a replacement for human oversight, requiring cooks to verify measurements and cooking times.
While purists may choose to ignore the feature, it provides a practical utility for impromptu weeknight meal preparation.
Photos by Jennifer Pattison Tuohy / The Verge
