# The New York Times puts an AI recipe assistant inside NYT Cooking

Published 2026-11-07T15:54:02+00:00 · Updated 2026-11-07T15:54:02+00:00 · By superintelligencenews@gmail.com
Canonical: https://superintelligencenews.com/applications/ai-recipe-assistant-nyt-cooking/

> The New York Times has added an AI recipe-customization assistant to NYT Cooking, with The Verge reporting that it helps subscribers on iOS make swaps, scale recipes and convert measurements. The feature looks useful for everyday cooking, but testing found some errors and raised questions about trust and control.

## Key facts

- The New York Times added an AI recipe-customization assistant to NYT Cooking, according to The Verge.
- The feature was available to subscribers on iOS at launch.
- The assistant can handle ingredient substitutions, recipe scaling, equipment substitutions and measurement conversion.
- The Verge said the AI generated modified recipes in seconds and was trained on user comments.
- In testing, the assistant refused to replace ladyfingers with minced beef in a sweet dessert trifle.

The New York Times has added an AI recipe-customization assistant to NYT Cooking, giving subscribers on iOS a new way to adjust recipes for substitutions, scaling and equipment changes. The feature, reviewed by The Verge, is designed to generate a separate modified version of a recipe rather than rewrite the original, putting a practical AI tool inside one of the company’s most trusted consumer products.

 

## What the new assistant does

 

The assistant appears as a text box on recipe pages and invites users to customize a dish before cooking. According to The Verge’s review, it can handle ingredient substitutions, recipe scaling, equipment substitutions and measurement conversion, which makes it more than a simple chatbot bolted onto a recipe page.

 

That matters because recipe customization is often where home cooks run into friction. A cook may want to swap one ingredient for another, halve a recipe for a smaller household or convert measurements between metric and U.S. systems. NYT Cooking’s new tool is aimed at those everyday decisions, not at replacing the underlying editorial recipe.

 

> “Customize this recipe…”

 

The prompt, as seen by The Verge, signals the feature’s main use case: tailoring a tested recipe to a specific kitchen and a specific dinner plan.

 

## How did it perform in testing?

 

In The Verge’s testing, the assistant was useful for weeknight cooking and responded quickly, generating a modified recipe in seconds. That speed is part of the appeal. Instead of manually recalculating ingredients or guessing at substitutes, a subscriber can ask for a version that better fits the available pantry or equipment.

 

The review says the assistant handled several practical scenarios well. It successfully modified a chicken-and-rice recipe by substituting chicken tenders, green beans and red onion. It also suggested lime zest as a stand-in for makrut lime leaf, while noting a flavor caveat. In a Dutch baby test, it converted metric measurements to U.S. measurements without issue.

 

But the tool was not flawless. The Verge reported that the AI sometimes missed nuance or gave imperfect advice, including a butter miscalculation in cake testing. That kind of error is small on paper and potentially large in the oven, where a slight ratio change can affect texture, rise and bake time.

 

### When the assistant said no

 

The system also showed some judgment. The Verge said it refused an unrealistic request to replace ladyfingers with minced beef in a sweet dessert trifle, an example that suggests the feature is not simply trying to comply with every prompt. It is making some attempt to keep requests inside the bounds of a recognizable recipe.

 

> “Minced beef cannot be used in place of the ladyfingers or anywhere in this sweet dessert trifle.”

 

That response is notable because it reflects a hard line between useful adaptation and culinary nonsense. It also hints that the system may be trained to protect the integrity of the original dish, even as it offers flexibility around it.

 

## Why it matters

 

This feature matters because it lands at the intersection of convenience, trust and editorial control. NYT Cooking has long sold itself on reliable, tested recipes and a large library of user comments. Adding AI customization gives subscribers another layer of utility, but it also raises the question of whether algorithmic suggestions can preserve the same confidence people expect from a curated cooking service.

 

For readers, the immediate benefit is obvious: fewer calculations, fewer abandoned recipes and more flexibility on a school night. For the product, the potential upside is stronger engagement and a reason to keep paying for a subscription. For the industry, the move is another sign that publishers are testing whether AI can improve a premium product without turning it into a generic chatbot experience.

 

There is also a broader consumer issue here. Recipe content sits in a category where nuance matters. A good substitution is not only about ingredient lists; it is about moisture, texture, cooking time and flavor balance. The Verge’s review suggests the assistant can help with straightforward changes, but that it still needs human judgment for edge cases.

 

## Background: how we got here

 

NYT Cooking is already built around a deep recipe archive and a large base of user comments, which makes it a natural place for a customization tool. The product’s existing structure gives an AI assistant a relatively controlled environment: the recipe has already been tested, the instructions are already written and reader feedback already exists around it.

 

That matters because consumer AI tools often do better at summarizing, suggesting and restructuring than at making fully reliable real-world decisions. Cooking is a field where those limits become obvious quickly. A generic answer might sound useful while ignoring whether a substitute changes the chemistry of a batter or the timing of a roast.

 

The Verge also noted that Amazon’s Alexa Plus offers similar substitution guidance, but without the ability to display a modified recipe. That comparison shows where NYT Cooking may have an advantage: it is not just answering a question, it is presenting a revised version of the dish in the context where a cook needs it most.

 

According to The Verge, the AI assistant was trained on user comments. That detail suggests the tool may be drawing on the practical language and problem-solving habits of actual home cooks, not just on generic recipe text. If so, the product is trying to blend machine assistance with the accumulated knowledge of a cooking community.

 

## By the numbers

 

The review included several concrete measurements that show where the assistant helped — and where it strained.

 

| Item | Detail from The Verge review |
| --- | --- |
| Launch access | Available to subscribers on iOS at launch |
| Response time | Generated a modified recipe in seconds |
| Weeknight test | Chicken-and-rice recipe tested last Friday night |
| Green beans | Steamed for 8 minutes in the modified recipe |
| Original cake timing | Three hours for the Cinnamon Roll Bundt Cake |
| AI cake timing | 50 minutes in the muffin conversion |
| Muffin yield | 24 muffins, then 12 after halving |
| Butter amount | 6 tablespoons in the AI version, versus three-quarter cup in the original |

 

## What worked — and what didn’t

 

### Useful for the practical swaps

 

The strongest case for the assistant is plain convenience. If a user wants to use what is already in the kitchen, the tool can propose substitutions without forcing them to mentally rebuild the recipe from scratch. That is exactly the kind of small time-saving feature that can make a premium service feel worth opening every night.

 

The Verge’s review suggests the tool is best when the request is specific and familiar: replace one ingredient with a similar one, adjust a yield or convert measurements. Those are the kinds of tasks that are easy to overlook as “simple,” even though they can be annoying enough to stop a cook in their tracks.

 

### Where the limits showed up

 

The feature’s weaknesses were also visible in the testing. The butter miscalculation in the cake example is the clearest warning sign. A recipe assistant can sound confident while still getting a foundational detail wrong, and that is precisely the kind of error that erodes trust in a kitchen.

 

The Verge’s report also indicates that the assistant may sometimes miss nuance. In cooking, nuance is not optional. A substitute that works in one dish may fail in another, and the difference can be subtle enough that a casual user will not spot the risk until the dish is already underway.

 

## How does this compare with other AI assistants?

 

The Verge’s comparison with Alexa Plus is useful because it shows the difference between recipe advice and recipe workflow. Alexa Plus can provide similar substitution guidance, but according to the review, it could not display a modified recipe. NYT Cooking’s version keeps the adjustment inside the recipe page, where it is easier for a cook to follow step by step.

 

That distinction may sound small, but it matters. A smart assistant that talks about substitutions is helpful; one that produces a readable, separate recipe version is more practical. The latter reduces the chance that the user will lose track of what changed and what stayed the same.

 

## Why some users may be skeptical

 

The feature has already drawn skepticism from some NYT Cooking users and recipe purists, according to The Verge. That reaction is predictable. Part of the value of a trusted recipe brand is the sense that a dish was tested as written, by someone who knew what the final result should be.

 

For those readers, AI customization may feel like a compromise. If the point of subscribing to NYT Cooking is to get carefully developed recipes, then letting software modify them can seem like a step away from editorial certainty. The challenge for the Times is to make the tool feel like a useful assistant rather than an interference layer.

 

At the same time, skepticism does not automatically mean rejection. A lot of cooks are less interested in perfect preservation than in dinner that works tonight. If the assistant reliably helps with weekday substitutions and conversion headaches, some users may value it even if they remain wary of broader AI hype.

 

## What happens next

 

The biggest immediate question is how widely the feature will roll out beyond iOS subscribers. The Verge says it was available to subscribers on iOS at launch, which leaves open whether other platforms will follow and how quickly that happens.

 

There are also product questions the company has not yet answered in the fact sheet. It is not clear whether NYT will let users save modified recipes as separate persistent versions, or how the assistant will behave across a wider range of recipes and edge cases. Those details will matter if the tool moves from novelty to habit.

 

Another issue is accuracy over time. The review was based on a limited set of tests, including chicken, curry, halibut, Dutch baby and cake recipes. The real test will be whether the system stays useful across more complex dishes, unusual substitutions and the kinds of improvised requests home cooks actually make.

 

Finally, there is the business question. If the assistant improves user engagement, it could strengthen the subscription offering. If it frustrates users or introduces too many unreliable suggestions, it could become a feature that feels more promotional than practical. The Times now has to prove that a little AI can add real value without weakening the trust that makes NYT Cooking different.

 

## The bottom line

 

The New York Times is testing whether AI can make recipe content more usable without turning it generic. Based on The Verge’s review, NYT Cooking’s assistant is promising when it comes to substitutions, scaling and measurement conversions, but it still stumbles on nuance — which is exactly where food and AI meet their hardest test.

## Frequently asked questions

### What does the NYT Cooking AI recipe assistant do?

It helps subscribers customize recipes by suggesting ingredient substitutions, scaling quantities, changing equipment and converting measurements. The Verge reported that it appears as a text box on recipe pages and creates a separate modified version rather than altering the original recipe.

### Who can use the NYT Cooking AI assistant right now?

It is available to subscribers on iOS at launch, according to The Verge. The fact sheet does not say whether the feature is already available on other platforms or when a broader rollout might happen.

### Is the AI recipe assistant accurate?

It appears helpful for simple, practical changes, but not perfect. The Verge said it worked well in some tests, such as a chicken-and-rice recipe and a Dutch baby conversion, but also made mistakes, including a butter miscalculation in cake testing.

### Why are some NYT Cooking users skeptical of the feature?

Some users and recipe purists worry that AI could undermine the trust and precision that define a tested recipe service. The concern is that customization may be convenient, but not every substitution or adjustment will preserve the original dish’s balance.

## Sources

- [The Verge](https://www.theverge.com/tech/1029489/nyt-cooking-ai-assistant-review)