A recent synthesis of scientific studies indicates that ChatGPT, one of the fastest-growing AI tools, may have a meaningful role in obesity management, although its effectiveness varies across contexts and remains subject to important limitations.

The review examined 37 studies published between December 2022 and October 2025, drawing on databases such as PubMed and Web of Science. Using a thematic synthesis approach, the researchers analysed how ChatGPT performs in areas such as lifestyle guidance, patient education, behavioural support, and clinical decision-making. To ensure methodological rigour, two authors independently screened and reviewed the studies, applying established quality assessment tools alongside a three-tier rating system to evaluate reliability. Despite the breadth of the analysis, only 27% of the studies were rated as high confidence, indicating that the evidence base is still developing.

The findings present a mixed but generally encouraging picture. Among the 29 original studies included, ChatGPT showed high accuracy in 75% of lifestyle and nutrition-related evaluations when compared with expert guidelines. However, its performance was less consistent in more complex clinical areas, particularly bariatric surgery, where only half of the studies reported strong alignment with professional standards. When compared with other AI tools such as DeepSeek, Copilot, Gemini, and Bard, ChatGPT demonstrated higher overall accuracy in most cases, although specialised medical applications and algorithm-based systems still outperformed it in certain scenarios.

The research was conducted by Dr Mohamad Motevalli, Brunna Boaventura, and Dr Fatima Cody Stanford, drawing on expertise from institutions including the German University of Health and Sport, the University of Innsbruck, the Federal University of Santa Catarina, Harvard Medical School, and Massachusetts General Hospital. This collaboration highlights the growing international interest in applying AI to complex health challenges such as obesity.

Beyond accuracy, the review highlights how ChatGPT is being used to support behavioural change and health education. The tool can provide practical advice on diet, exercise, and lifestyle habits, often in a conversational format that improves accessibility and engagement. In some cases, its responses have been found to closely resemble those of human coaches, suggesting potential for long-term user interaction and motivation. It can also assist clinicians by summarising information, generating educational materials, and offering general clinical guidance, particularly in settings where resources are limited.

However, the researchers emphasise that these benefits must be weighed against significant concerns. ChatGPT’s responses can be inconsistent, occasionally incomplete, and not always aligned with current clinical guidelines, particularly in complex or multi-condition cases. The model’s training data, largely drawn from high-income and English-speaking contexts, may also limit its cultural relevance and introduce bias into its recommendations. In addition, the lack of transparency in how the system generates responses raises questions about accountability and trust in clinical settings.

The most important finding is that while ChatGPT shows clear potential as a supportive tool in obesity management, it cannot yet be relied upon for independent clinical decision-making. Its strengths lie in education, engagement, and general guidance, but its limitations in accuracy, consistency, and contextual understanding mean that human oversight remains essential.