Is There A Place for a Test Kitchen in an AI World?

By Chef Alexandra Yue, Test Kitchen Manager

I’ve seen plenty of change in my 10-plus years at Nourish, especially related to my role as a recipe developer. Flavour trends (and a bunch of crazy fads!) have come and gone. A lot of new or revived technology has joined my pots, pans, and ovens. Air fryer, dehydrator, InstantPot. And now there’s Artificial Intelligence. Where does that fit in? And is there room for us both in the test kitchen in 2026?

You’ve come a long way, AI

Let’s look at the state of Artificial Intelligence when I started here in 2015. Back then, it was a niche topic, and articles about AI mostly revolved around the potential for integrating it with robotics; the “brains” behind the “brawn.” 

What not many saw coming was how AI would insert itself into our lives, how user-friendly it would become, and how it would be seemingly everywhere because it’s not shackled to robots. It’s in our devices and apps and search engines. It’s even shown up in grocery stores and fast food restaurants. 

Like a lot of creative content producers, I began to worry that AI might make a critical part of my role obsolete. But I’ve come to realize that while AI can augment my job, it can’t do my job. Instead of fighting it, I’ve embraced artificial intelligence by putting it to work for me.

Is ChatGPT the ultimate sous chef?

Let’s answer the question directly: Do we, at Nourish, use AI in the test kitchen? 

Absolutely. 

AI is my sous chef, my research assistant that I keep in my pocket beside my trusty Sharpie on my Chef’s jacket. And I mean that literally; I can access AI as easily on my phone as on my laptop. 

Using AI has greatly reduced my time on research and taken much of the guesswork out of whether certain flavours or ingredients will work well together. Before adding AI to my toolkit, I would have spent several hours researching, developing, and finally testing a recipe. At the end, I still might discover it didn’t work.

If I’m unfamiliar with a particular type of recipe, I can ask AI to outline the basic method. I can use it to ensure a recipe or flavour idea is on trend, or if it fits the criteria we are aiming for. When it’s time to commit the recipe to “paper,” I bring AI in again for basic tasks like picking out all the equipment used and creating a list, or verifying every ingredient used is listed and in the recipe steps (sometimes, I make a last-minute change, either adding in or taking out). 

Spoiler Alert: AI Doesn’t “Think”

What AI can’t do is replicate the human element behind a great recipe. 

To test its creative capabilities, I ran a little experiment: I asked ChatGPT and another AI app, DishGen, to each generate a chewy chocolate chip cookie recipe.

The results? 

The AI-generated recipes were almost identical. The only differences were minor—a pinch more salt here, a touch less baking soda there. Not enough to distinguish between them. 

Why so similar? Because they effectively pulled from the same store of knowledge. They each checked for other chewy chocolate chip cookie recipes in the data they’d been trained on, chose what seemed to be the most popular and representative of the category, and came up with something that essentially averaged them out.

AI can’t think, and it doesn’t have unique experiences or perspectives. It can mimic. It’s fantastic at research and synthesis. But it cannot be truly creative and original. 

So, how do we make sure originality and creativity don’t get lost when developing recipes at Nourish?

There’s a reason it’s called a “test” kitchen

Here’s a culinary truth: An untested recipe is just a theory. Using a test kitchen run by an experienced chef ensures a result you can be confident in. All that time saved by asking AI to do what it’s best at frees me up to spend more time experimenting and fine-tuning a recipe until it's perfected.

That requires hands-on testing and tasting, and then adjusting based on all available sensory data. For example:

1. Sight: Does the finished recipe look appealing? Do we need to add a garnish for the sake of appearance or to reinforce a flavour profile visually?
2. Smell:  Is the aroma overpowering or underwhelming? Does it smell like you expect it to?
3. Touch: Is the texture satisfying? Is it appropriate for the recipe?
4. Sound: Does the sound when you bite into it match what your eyes (and past experience) have told you it should be?
5. Taste: Is the flavour what we wanted? Are the key ingredients present or hidden in the profile?

These are all questions (and there are many more) that cannot be answered by any prompt. 

What’s the future of AI in recipe and food product development?

AI continues to improve at virtually all tasks asked of it. On the food front, we noted in our 2024 Nourish Trend Report that AI tongues were in development, with sensors in place of taste buds, that could “analyze food content, quality, authenticity, and possibly even flavour profiles.” 

At the end of the day, though, the “intelligence” remains artificial. More importantly for food, so does any indication of emotion. While a robot taste-tester might be great for quality control, I can’t see it replacing humans in recipe development. It’s not just about measurement and quantities when it comes to great food; how that food makes you feel is essential to the experience. 

In Nourish’s Test Kitchen, we ensure the recipes developed for our clients are not only unique and speak to their brand, but are also culturally or contextually appropriate, edible, taste good, and look good too!

And that’s the difference a human makes for each and every recipe developed at Nourish: experience, intuition, emotion, and the kind of “je ne sais quoi” no artificial intelligence app can bring to the table.