NYT Cooking AI Assistant Adapts Recipes but Faces Baking Quirks
The New York Times recipe subscription platform NYT Cooking released an artificial intelligence assistant designed to help subscribers customize recipes, swap missing ingredients, and scale serving sizes. Available at launch exclusively for iOS subscribers, the digital tool appears as a text box on recipe pages inviting users to tailor culinary instructions to their personal dietary needs. In hands-on testing reported by The Verge, the assistant proved valuable for weeknight dinner adaptations while exhibiting clear limitations during baking tasks. The software tool draws from historical user comments across the publication's recipe database to generate practical ingredient substitutions and altered cooking timelines. Importantly, the assistant leaves the original tested recipe intact while generating a customized variant under a distinct Your Version tab.
Weeknight Savory Adaptations and Household Testing
Testing the system on a weeknight dinner dilemma last Friday, a home cook adapted Yasmin Fahr's One-Pot Chicken and Rice With Ginger recipe to utilize available pantry staples. The cook input available staples including poultry tenders, fresh ginger, garlic, rice, broth, and limes, alongside green beans and red onions. Within seconds, the tool created an adapted dish titled One-Pot Chicken and Rice With Ginger and Green Beans, adjusting searing times and suggesting that green beans steam atop the rice. When asked about replacing missing cilantro, the assistant agreed to substitute Thai basil while accurately describing the resulting aromatic shifts. NYT Cooking editors stated that the algorithmic assistant was trained not to comply with unpalatable requests or bizarre culinary combinations. Axios reported in May that The New York Times signed a multi-year artificial-intelligence licensing deal with Amazon, giving Amazon access to Times material including Cooking and The Athletic for AI-related products and services.
Baking Constraints and Measurement Discrepancies
The tool demonstrated sensible boundaries during stress testing, properly refusing a humorous prompt to replace ladyfingers with minced beef in a dessert trifle. When tested against Amazon's Alexa Plus running on an Echo Show 11 smart display, both services suggested similar ingredient adjustments, though Alexa could not display the modified text. The assistant encountered greater challenges when applied to baking, where chemical ratios and precise pan dimensions govern recipe success. Testing the tool on food editor Genevieve Ko's Cinnamon Roll Bundt Cake, the user asked the assistant to convert the recipe into muffins. The assistant adjusted the baking duration from three hours down to 50 minutes and divided the batter across two 12-muffin tins, producing 24 muffins.
When the user requested halving the batch to yield 12 muffins, the system reduced five eggs to three rather than suggesting an impractical half-egg measurement. However, confusion arose between recipe versions when the user asked for butter measurements, as the tool quoted the modified six tablespoons rather than the original three-quarter cup. The test showed that while the tool operates effectively as an adaptable kitchen helper, cooks must verify recipe ingredient proportions during baking. The introduction of generative assistance highlights how digital culinary publishers are integrating algorithmic personalization into established recipe collections. The feature provides practical utility for home cooks adapting recipes on the fly while preserving access to the New York Times catalog of human-tested recipes.