NYT Cooking’s AI assistant reviewed: a hands-on test of recipe customization
A senior reviewer puts the NYT Cooking app’s new AI-powered recipe assistant through a weeknight dinner and a baking test, finding practical substitutions but cautioning about accuracy.
1 source · cross-referenced
- The NYT Cooking app added an AI assistant that customizes recipes on demand for substitutions, scaling, and equipment swaps.
- A reviewer tested the tool on weeknight dinner and baking, finding it helpful for substitutions but prone to small errors like incorrect butter amounts.
- The assistant was trained on user comments and avoids agreeing to requests that would likely taste bad, according to NYT Cooking editors.
- The feature does not alter the original recipe; it creates a new version with changes highlighted for comparison.
The New York Times Company added an AI assistant to its NYT Cooking app that lets subscribers customize recipes on demand, available initially on iOS. The tool appears as a text box on recipe pages with the prompt “Customize this recipe…,” where users can request substitutions, scaling, equipment swaps, or measurement conversions. The assistant generates a new version of the recipe without altering the original, and changes are highlighted in a “Your Version” tab for side-by-side comparison.
In a hands-on test, a senior reviewer used the assistant to adapt a One-Pot Chicken and Rice With Ginger recipe for missing ingredients. The AI suggested using chicken tenders instead of thighs, omitted cilantro (noting it was optional), and incorporated red onions and green beans with adjusted cooking steps. The revised recipe was completed in the same time as the original and received positive feedback from the reviewer’s household.
The reviewer also tested the assistant while baking, where it nearly recommended halving the required butter, demonstrating that while the tool can be helpful for substitutions and guidance, it is not error-free. The reviewer emphasized that the AI’s responses were practical and often saved time compared to leaving the app to search for answers.
According to NYT Cooking editors, the assistant was trained on user comments and is programmed not to comply with requests that would likely result in poor-tasting dishes. The tool does not uncritically accept all user requests, which the reviewer confirmed by attempting an unusual substitution that the assistant declined.
The reviewer, who has used several AI tools in the kitchen, characterized the NYT Cooking assistant as a useful and focused application of AI, particularly for weeknight meal planning where time and convenience are critical. The reviewer also noted that the feature’s ability to create a new recipe version rather than overwrite the original was a practical design choice.
- Nov 10, 2026 · Wired
Kora, a Linux-based desktop OS from Cyanogen cofounder, targets AI power users with local model integration
Trust76 - Nov 10, 2026 · One Useful Thing — Ethan Mollick
Mollick argues new AI models approach "jagged AGI" across coding, business, and humanities
Trust78 - Nov 9, 2026 · The Verge — AI
Former DOJ antitrust chief launches AI legal-research startup to challenge monopolies
Trust72