Do AI Recipes Work? We Tried 3 AI Recipe Generators to Find Out

I was recently invited to a dinner party at a neighbor’s house, and I wanted to make a dessert. Years ago, I was very into making raw vegan cheesecakes, but I hadn’t made one in a long time. 

I remembered there was some cashew-soaking, a bit of lemon-zesting, and some blending involved. It seemed easy enough from what I remembered, and something that my neighbors — who were all very different ages and with different dietary restrictions — might like. 

I got to thinking about a topic that keeps coming up in our office: AI (artificial intelligence) recipes. There are now lots of websites that use AI machine learning to populate recipes based on ingredients, dietary restrictions, and even cooking skill level. Some create a recipe based on ingredients you have on hand, while others churn out a recipe based on a query, like ‘Make a gluten-free chicken parmesan.’

I decided to set out on a journey to test a few different AI recipe generators, starting with the next night’s vegan cheesecake. 

Chat GPT 

I started with Chat GPT, not out of any sense of preference or loyalty, just simply because it’s the only one I knew of. 

I created an account and typed in, ‘Make me a raw vegan cheesecake recipe’. I instantly had a flashback to a computer game I played in elementary school, something about math or statistics, where a bunch of colorful critters were tasked with making pizza for some giant overlord. When the game started, he would shout, “Make me a pizza!” in a deep, faraway voice. (A quick Google search tells me that the game was called The Logical Journey of the Zoombinis.) 

I loved that game, but I wasn’t yet sure how I felt about Chat GPT, which up until that point just felt like something that devalued writing and optimized tasks that didn’t necessarily need optimizing. 

The screen paused for a second, then a cursor appeared and started typing: 

“Sure! Here’s a recipe for a delicious raw vegan cheesecake:

Ingredients: For the crust:

• 1 cup raw almonds
• 1 cup pitted dates
• 1/4 cup shredded coconut (optional)
• Pinch of sea salt
  

For the filling:

• 2 cups raw cashews (soaked in water for at least 4 hours, then drained)
• 1/2 cup full-fat coconut milk
• 1/2 cup maple syrup or agave nectar
• 1/4 cup melted coconut oil
• 1/4 cup fresh lemon juice
• 1 teaspoon vanilla extract
• A pinch of sea salt
  

For the topping (optional):

• Fresh berries, sliced fruit, or a fruit compote of your choice
  

Instructions:

1. Prepare the crust by placing the almonds, dates, shredded coconut (if using), and sea salt in a food processor. Process until the mixture sticks together when pressed between your fingers.
2. Press the crust mixture evenly into the bottom of a springform pan or a pie dish, creating a solid crust layer. Place it in the refrigerator to chill while you prepare the filling.
3. In the same food processor (no need to clean it), combine the soaked and drained cashews, coconut milk, maple syrup or agave nectar, melted coconut oil, lemon juice, vanilla extract, and sea salt. Process until smooth and creamy, scraping down the sides as needed.
4. Remove the crust from the refrigerator and pour the filling mixture over it, smoothing the top with a spatula or the back of a spoon.
5. If desired, add your preferred topping, such as fresh berries, sliced fruit, or a fruit compote.
6. Place the cheesecake in the freezer to set for at least 4 hours or overnight.
7. When ready to serve, remove the cheesecake from the freezer and let it sit at room temperature for about 10-15 minutes to soften slightly. This will make it easier to slice.
8. Slice and serve the raw vegan cheesecake. Store any leftovers in an airtight container in the freezer for up to a month.

Enjoy your creamy and delicious raw vegan cheesecake!”

It had all the components I remembered from my previous vegan cheesecake–making: the soaking of the cashews, the lemon for the cheesecake flavor, even the crust, which I quickly decided to omit out of laziness. 

I set the cashews out in a bowl to soak overnight. The next day, they were plump and soft. I quickly melted the coconut oil and substituted honey instead of agave, blended them with the soaked cashews and other ingredients in my blender, and spooned the mixture into short dessert dishes. 

For the topping, I blended some strawberries and blueberries with a bit of the leftover cheesecake and some extra coconut oil, then spooned that on top. 

I tasted everything along the way, and it tasted just how I remembered: smooth, creamy, with a nutty flavor that’s certainly different from traditional cheesecake, but still had that same tart flavor that I was looking for. 

Did it work? Surprisingly, yes! The cheesecakes were a hit 

Process notes: Soaking the cashews was time-consuming, but otherwise this recipe is pretty easy (and not having to turn the oven on is always a plus). As a non-baker, I found it simple and kind of fun to make. I might add a bit of extra honey or fruit to make the flavor a little more noticeable. 

Would I use Chat GPT again? Maybe. I don’t really see the merit to using Chat GPT for a recipe versus a traditional search engine (or even Pinterest), but the recipe turned out well and worked as it should, so that was a plus. 

---

Let’s Foodie 

I had a few things in the fridge for lunch — a few basics that I always buy, like avocados and spinach, but also a few out-of-left-field purchases, like red cabbage and this new Acid League salad dressing. I thought it *could* make a nice salad for lunch, but I figured I’d put our next AI recipe generator to the test to help me out. 

This one is called Let’s Foodie. It operates a bit differently: Instead of typing in a question or request, you type in the ingredients you have on hand and let the AI build you a recipe. 

At first glance, Let’s Foodie seemed a lot more helpful than Chat GPT (at least for my lunchtime process, which usually involves staring blankly into the fridge for a while and then haphazardly combining ingredients, sometimes with success, other times with disappointment). 

I typed in ‘Red cabbage, spinach, almonds, chicken, avocado’. I didn’t add the salad dressing because frankly, I wasn’t sure if I could add something so specific. 

A spinning wheel filled the screen, then a recipe appeared: 

“Cilantro-Almond Chicken with Avocado-Red Cabbage Slaw

Ingredients:

1 head red cabbage, shredded
2 avocados, diced
1/2 cup cilantro, finely chopped
1/2 cup almond slices
2 tablespoons olive oil
1/2 teaspoon garlic powder
1/4 teaspoon black pepper
1/4 teaspoon cumin
1/4 teaspoon chili pepper
1/4 teaspoon salt
2 chicken breasts
3 cups fresh spinach

Instructions:

1. Heat 2 tablespoons of olive oil in a large skillet on medium heat.

2. Season the chicken breasts with garlic powder, black pepper, cumin, chili pepper and salt.

3. Place the chicken in the skillet and cook for 10 minutes, flipping occasionally.

4. Meanwhile, in a medium bowl, combine the shredded red cabbage, diced avocado, cilantro and almond slices.

5. In a small bowl, combine the remaining olive oil, garlic powder, black pepper, cumin, chili pepper and salt.

6. Drizzle the olive oil mixture over the cabbage mixture and toss to combine.

7. Add the spinach to the skillet with the cooked chicken and stir to combine.

8. Cook for an additional 5 minutes, stirring occasionally, until the spinach is wilted.

9. Serve the cooked chicken over the red cabbage-avocado slaw and enjoy!”

Again, this was pretty aligned with what I hoped the machine would churn out. I would have preferred a raw salad since we’re having near-80-degree days in L.A., but otherwise, I liked the slaw and the spice combination for the chicken. 

I decided to make everything according to the recipe, except instead of sauteing the spinach with the cooked chicken (which had already been cooking for a while and seemed like it was at risk of overcooking), I tossed the spinach in a bowl with the Acid League dressing. 

I added the chicken and cabbage slaw to a bowl with the spinach salad. It was simple and fresh, a light lunch without much thought — just what I was looking for. 

Did it work? Yes; some of the preparations wouldn’t have been my first choice, but all the elements worked well together 

Process notes: Cooking the chicken with the spinach may have made it a little over-done, and the combo of cooked spinach and raw slaw seemed a little strange temperature-wise, but I loved the idea of combining the cabbage and avocado to make a slaw. In the end, I was grateful for the pre-made dressing, because without it, the flavor may have been a little bland. 

Would I use it again? Yes, probably. This one makes more sense to me. While the ingredients I had on hand made sense together from the get-go (and I probably could have come up with my own recipe if I had tried), using Let’s Foodie definitely eliminated those 5-10 minutes that I usually spend ruminating over what to make, and took away the question of whether or not the combination would actually taste good together.  

PlantJammer

Right away, I liked the concept of this one. “Half of global food waste happens in the home,” the PlantJammer site reads. “80% of people cook less than 10 recipes per year! We believe a more sustainable food system requires more versatile home cooking. That is why we aspire to empower home kitchens.” 

The idea that I could eliminate some of my food waste by finding creative ways to combine fruits and veggies that were about to go bad made PlantJammer worth a try.  

To start, you “Empty Your Fridge” by choosing ingredients that you have on hand (though the interface is a little more fun and colorful than Let’s Foodie, which just uses a simple search bar). I clicked on the cute icons for sweet potatoes, kale, and zucchini, and I was pleasantly surprised to see that it even had an option for rotisserie chicken, which I had in my fridge from the previous night’s dinner. 

When I clicked submit, I tried to guess what it would come up with. A stir-fry? Maybe a simple roasted sweet potato, chicken, and the veggies on the side? 

I was surprised to see that instead of just one recipe, it churned out a few options — and that the first was a lasagna. 

I had never tried making lasagna with shredded chicken (or sweet potatoes, for that matter), so I was intrigued. I noticed that you could filter the recipes by cuisine, difficulty, and even values like gluten-free and vegan, but since I was now set on my lasagna dinner, I decided to move forward with the very first option. 

PlantJammer did a bit more thinking, then delivered the full lasagna recipe: 

“Ingredients: 

Burrata
Kale
Parmesan cheese
Ricotta
Sweet potato
Zucchini
Lasagna sheets
Rotisserie chicken
Onion
Basil
Garlic
Oregano
Tomato sauce
Olive oil

Instructions:

1. Preheat your oven to 200°C (400°F).
2. Peel and slice the onion.
3. Peel and cut the sweet potato.
4. Wash and slice the kale and zucchini.
5. Finely chop the basil and garlic.
6. Slice the burrata.
7. Heat olive oil in a pan and fry garlic and onion with oregano for a minute. Add sweet potato and fry for 15 minutes. Add zucchini and rotisserie chicken and fry for 10 minutes. Add in kale and fry for 5 minutes.
8. Add tomato sauce to the pan and mix.
9. Your sauce should be liquid enough to cook the lasagna noodles during baking, add water or stock as needed.
10. Add salt and pepper to taste.
11. Lightly cover the bottom of your oven dish with a thin layer of sauce. Then place a layer of lasagna sheets on top.
12. Add a layer of ricotta.
13. Repeat the layers of sauce and lasagna noodles.
14. Top with burrata and parmesan cheese.
15. Cover your lasagna with aluminum foil and bake for 20 minutes. Uncover and bake for another 15 minutes.
16. Top with basil.”

I initially thought it was strange that there were no measurements in the ingredient list, but noticed that they popped up in little icons next to each ingredient.

I followed the instructions, and everything went smoothly. In the end, I pulled a sizzling lasagna out of the oven, melted mozzarella (a sub for burrata) on top, the scent of fresh basil filling the room — basil that, I should note, I would have otherwise thrown away in the next day or two. 

Did it work? Yes, much better than I anticipated given the unique combination of ingredients for a lasagna

Process notes: While the cute icons and ingredient categories were fun to look at, I found that I appreciated the simplicity of ChatGPT or Let’s Foodie. It was a bit difficult to find and select my ingredients at the beginning (though once I started to read the instructions, I found that the icons and measurements at each step made things easier to follow). 

Would I use it again? Maybe. I liked the mission behind Plant Jammer, but it feels like the site has a way to go before it’s easy enough to use regularly. 

What’s the Verdict on AI Recipes?

I totally understand why people would want a solution for the head-scratching dilemma of having a bunch of mismatched ingredients and no idea how to combine them. To me, though, I already have a perfectly addictive, inspiring, preferable alternative to an AI recipe generator — and that, my friends, is Pinterest.