On a Sunday night, when most bad ideas seem reasonable, the idea seemed reasonable enough. The takeout menus had started to look like wallpaper, the refrigerator was almost empty, and in between a half-eaten apple and the third unread email, the idea struck: let the machine decide. Twenty-one meals, seven days, and no human thought. Simply type the request and comply.
In a matter of seconds, the first plane touched down. It appeared competent in the same way that a stranger’s resume appears competent before you read it. Almond butter and breakfast oats. A bowl of grains for lunch. Dinner will be sheet-pan chicken. Like most reasonable things, it’s balanced, reasonable, and a little dull. However, the fissures became apparent by Tuesday. Perhaps sensing complacency, the algorithm recommended a smoothie made with lentils and silken tofu—the kind of meal that only exists in the mind of something that has never had food. It turned out that a version of this same recommendation had already gone semi-viral after a vegan mother in the Wall Street Journal was given a recipe that included chocolate strawberry Cheerios “for crunch.” Evidently, misery enjoys digital company.
AI seems to comprehend food grammar without actually understanding food. It understands that proteins and carbohydrates belong together. It is aware that greens go to the side. It is unknown whether a Cobb salad that has been chilled overnight turns into a soggy disaster by Wednesday. It is unaware that a New York apartment fridge, which already contains a half-empty jar of capers and someone’s forgotten kombucha, cannot accommodate nine different dinner recipes. Earlier this year, Cori Ritchey encountered the same issue while writing for Men’s Health: too many meals, insufficient leftovers, and a lack of understanding of how a true kitchen functions.
The experiment had become somewhat absurdist by Thursday. The grocery store had never heard of the ingredients the chatbot confidently listed for a cauliflower curry. When compared to a nutrition database, the listed calorie count of a tuna dish was almost two hundred calories off. For some time now, dietitians have been voicing this concern, first quietly and then more loudly. Perhaps the most human aspect of the bot is that it writes with conviction even when it is incorrect.
It wasn’t all bad, though. On Friday night, I had a lemon-flavoured orzo with roasted vegetables that I might make again, on my own initiative, like any other person. There were a few small blessings. A six-minute breakfast scramble. A straightforward soup that served as two meals. When all was said and done, the savings came to roughly thirty-seven dollars, which is real money even though it isn’t money that will change your life. Food waste significantly decreased as a result of the plan, primarily because it imposed a level of accountability that is not present when browsing delivery apps.
But it wasn’t the food that stuck with me. It was the peculiar flatness of eating without making a decision. Meals turned into chores. The enjoyment of negotiating with yourself while standing in front of the refrigerator at seven o’clock in the evening completely vanished. That friction has merit. The chatbot may actually be helpful as a place to start, a kind of endlessly creative culinary collaborator. However, it is ineffective as a dinner dictator. It turns out that one of the last areas where inefficiency serves a purpose is in the food industry.
FAQs
1: Can AI really plan a full week of meals?
Yes, but expect odd suggestions, off portions sizes, and recipes that look better on screen than on a plate.
2: Is AI-generated nutrition information accurate?
Not always — dietitians have flagged calorie counts and macro estimates that can be off by significant margins.
3: Does using AI for meal planning actually save money?
It can, with reported savings of around $37 a week, mostly by cutting impulse buys and reducing food waste.
4: What’s the biggest downside of letting AI choose your meals?
It removes the small daily pleasure of deciding what to eat, making meals feel mechanical rather than personal.
5: Should you trust AI meal plans completely?
No — treat them as inspiration, not instruction, and always double-check ingredients, portions, and nutrition details yourself.
