Why AI Might Not Be The Future Of Meal Prep Yet | Prep UK
For a lot of people, the hardest part of a meal prep happens before a single vegetable is prepared and before a single container is opened.
Planning can be a daunting endeavour at times. If you are the type that already suffers from choice anxiety, planning your entire week’s worth of meal ingredients, recipes and equipment can be particularly dispiriting.
However, some fitness instructors have decided to put their plans in the hands of artificial intelligence tools such as ChatGPT, and by being specific and asking the system the right questions, the tool can generate a suitable and varied meal plan.
However, when one person tried it, the results were exceptionally strange, to say the least.
Amongst the delights in the three-day plan were unseasoned roasted vegetables with cooked quinoa, a spinach smoothie made with almond milk and almond butter, a salad that comprised of uncooked chickpeas, an entire diced cucumber, red onion and tomato with a drizzle of olive oil, and a black bean “soup”.
The latter dish consisted of black beans in water with diced onions, carrots and celery sticks. The person trying the prep was not impressed.
From a technical standpoint, the AI succeeded in providing a budget-friendly vegetarian meal plan for three days, but it also made several assumptions and decisions that would, if taken seriously, cause potentially serious harm in the long run.
It assumed an extremely low-calorie goal of 1,300 calories per day, which is lower than the NHS recommends and was unlikely to be enough food to keep an adult going.
The particular plan suggested was low in iron and had not considered nutrient balance in its algorithm. This is understandable for a general-purpose AI to miss, but also means that its plans cannot be relied on to provide long-term healthy, balanced meals.
Finally, the key to a successful prep routine is that the food needs to be enjoyable and spark happiness. The less engaging a meal plan is the more likely someone will give up on it. Ultimately, AI still lacks an understanding of what people want out of food.