How Much Can You Trust AI Eating Plans? Experts Explain

by | Apr 20, 2026 | Weight Loss

Artificial intelligence is being used in every process and product, from annoying deep-fakes to the new app your cousin vibe-coded to rate his ‘fits. Everyone’s hopped aboard the AI train, to do everything from be our boyfriend to make us meal plans for weight loss. But can you trust AI eating plans? WH investigated by asking two dieticians to rate and review a plan generated by ChatGPT. 

Meet the experts: Taryn Bortz is a registered dietician in Gauteng. Mbali Mapholi is a consultant dietician in private practice, known as @urbandietitian on social media.

AI Can Make Weight Loss Easier 

Wondering if you can trust AI eating plans is perfectly rational. After all, we now know from emerging science that weight loss is not as simple as calories in, calories out. There are myriad other factors that are as important, such as hormone fluctuations, genes and of course, the approach you take. Anyone who’s done a juice cleanse, then intermittent fasting and the keto/Paleo/carnivore vortex knows that dieting is as complicated as finding the right queue/document/payment method/arrival time at Home Affairs. Naturally, asking a machine to handle the details cuts a ton of energy. 

Using AI to generate a weight-loss plan is also way cheaper than paying out of pocket for a registered dietician. Apps like ChatGPT and Claude are free to use and can personalise things for you. According to some estimates, in 2023, 20 million people were using AI-driven meal planning apps globally. Of those, 76% were women. The popularity of trusting AI eating plans is something Mbali Mapholi, a consultant dietician, has witnessed herself. “I’ve definitely seen an increase in clients who either started with AI-generated plans or are actively using AI alongside professional advice,” she observes. “I think it reflects accessibility and curiosity. Women want quick answers, structure, and autonomy. That’s understandable.” A randomised clinical trial in JAMA found that using AI as a weight loss tool can be as effective as human-led coaching. 

READ MORE: Here’s Why It’s So Common To Regain Weight After Losing – And What You Can Do About It

The AI Eating Plan 

In order to figure out how much to trust your AI eating plan, we tasked two dieticians to evaluate a plan we generated on ChatGPT. 

We asked ChatGPT: 

“I am a woman in South Africa, 155cm tall weighing 74kg. I want to lose 1kg a week with moderate exercise and diet. Make me an eating plan that will allow me to achieve this goal.”

It spat out… 

“To lose 1 kg per week, you need roughly a 7,000–7,700 kcal weekly deficit (~1,000–1,100 kcal per day).

At 155 cm and 74 kg, your estimated needs are:

  • BMR (basal metabolic rate): ~1,400 kcal/day
  • With moderate exercise (3–5x/week): ~2,000–2,200 kcal/day maintenance

To lose 1 kg/week, your intake would need to average 1,200–1,300 kcal/day.

⚠️ Important: For women, 1 kg per week is aggressive. It is possible short-term, but 0.5–0.8 kg per week is more sustainable and safer hormonally. If you feel fatigued, increase calories slightly.”

The plan included these macronutrient targets and offered meals for seven days: 

  • Protein: 95–110 g
  • Carbs: 90–120 g
  • Fats: 35–45 g
  • Water: 2–3 litres daily

Meals varied – breakfast was boiled eggs and toast or yoghurt and chia seeds or oats. Lunch included lean protein and vegetables, and dinner was much the same. The plan also included recommendations for exercise: at least four days of workouts, including three days of strength training and two days of brisk walking

READ MORE: This 17-Day Eating Plan Will Help Get You Back In Shape

Dieticians Evaluate The AI-generated Eating Plan  

Mapholi and Bortz spotted red flags in the ChatGPT eating plan. First off, Bortz notes that there isn’t enough information given about the patient. “In the prompt, age is not specified, which plays a huge role in calculating basal metabolic rate (how many calories you burn at rest),” she says. Mapholi agrees. “AI can only work with the information it’s given,” she explains. “It does not assess body language, emotional relationship with food, medical history nuances, or metabolic adaptations in real time.” 

Next, both Mapholi and Bortz agree that the AI eating plan from ChatGPT is too low on carbs. “This may cause restrictive eating feelings, and also she may end up with high cravings at night, resulting in this plan being unsustainable,” explains Mapholi. For Bortz, the low-carb content means anyone following this plan may be low on energy, since carbs are responsible for feeding your body’s battery. “Without enough carbs, this patient will constantly feel hungry and low-energy, leading to over-snacking and cravings,” she says, echoing Mapholi’s views. 

Both dieticians identified a litany of other factors that could hinder your weight loss if this plan were your primary source. Bortz recommended adding more fibre – something dieticians have recommended for aeons because of its ability to boost satiety.  Another missing note from the ChatGPT plan? Fat. When cooking, oil is used to fry or bake food. But GPT had “no mention of specific portion sizes of fat to use,” says Bortz. “Fat is very calorie-dense and using too much oil/sauces can increase the caloric density quite a lot, which can slow down weight loss.” 

READ MORE: 7 Clever Hunger Hacks For Weight Loss You Haven’t Heard Of 

So… Can You Trust AI Eating Plans For Weight Loss? 

We can’t give a flat-out yes or no here. AI tools generally have the entire length and breadth of academic and scientific information within reach – but none of that matters if it’s only given crumbs of information to go on. Since AI chatbots are only built to respond to given input, they could miss critical contexts that a trained professional would collect before creating guidance. “Without detailed medical history, blood work, cycle patterns, stress levels and behavioural patterns, recommendations can miss critical context,” explains Mapholi. Bortz agrees – when creating plans, she also takes into account her patient’s food preferences, supplements and work schedules. “There are also no recipes provided in the eating plan or food preparation methods,” she points out. “One also needs to be realistic and offer healthy snack ideas for a sweet/salty craving.” 

And without knowing these critical factors, machine-generated plans can also create potential for harm. A study in Frontiers in Nutrition found that AI meal plans can recommend nutritional and calorie intake far below the minimum that dieticians would recommend. It’s something Mapholi spotted in this plan. “This plan lands at 1200kcal – 1300kcal, which even for a sedentary female is considered quite low in supporting metabolic functions,” she says. 

For dietician Taryn Bortz, AI eating plans can warp someone’s sense of what’s expected to lose weight. “Estimates can sometimes be unrealistic and not personalised to the patient,” she warns. “This can lead to the patient having unrealistic expectations and goals, often leading to frustration, disappointment and failure.” 

READ MORE: What Is The 2-2-2 Method For Metabolism – And Does It Help You Lose Weight?

How To Know Your Plan Isn’t Working For You 

Warning signs include constant cravings or hunger, or feeling lethargic or unable to focus. “If a plan feels exhausting, isolating, or overly rigid, it’s unlikely to be sustainable,” warns Mapholi. 

How To Prompt AI To Get A Better Eating Plan

If, after reading this, you’re still inclined to try it out, Mapholi and Bortz outline these pointers to give your chatbot the necessary info. 

Add essential context to your prompt 

Bortz noted that in our prompt, we hadn’t accounted for medical conditions, allergies, food preferences, supplements or other factors. Even little things, like how you feel about oats (a no-go for some), can impact the entire day’s meal. Mention if you live alone or cook for an entire family. Mention your workout routine. Bortz and Mapholi suggest you take these into account when prompting AI for weight loss advice: 

  • Age, height and weight 
  • General activity levels throughout the day 
  • Socioeconomic status at home
  • Medical conditions
  • Allergies
  • Food preferences and cultural eating patterns
  • Menstrual cycle status – regularity or any other factor
  • Sleep quality and quantity
  • Supplement use
  • Gut issues
  • Work schedule/ day-to-day schedule (meal plans need to fit within this)
  • History of dieting or disordered eating
  • Details about exercise: type, frequency, duration, intensity
  • Specific goals: build muscle/ tone/ lose weight/ healthy eating/ improved meal preparation

Remember: Humans Are Better 

Even with many prompts, AI can’t replace personalised clinical judgement, Mapholi says. And, a real dietician can encourage progress with support – check-ins, motivation and advice that come from people who’ve helped countless other people reach their goals. A human dietician is also able to pinpoint when a meal plan needs to be adjusted. With AI, you’d have to ask – something you may not know to do. “[Dieticians] provide behavioural coaching, ongoing adjustments, accountability, and a sustainable strategy tailored to the individual,” says Mapholi. “Most importantly, we look at the whole woman, not just her calorie target, ensuring that fat loss does not come at the expense of muscle mass, hormonal balance, gut health, or overall wellbeing.” 

READ MORE: I Used AI To Create An Eating Plan For Weight Loss – Here’s What Happened

Pin It on Pinterest

Share This