Good evidence confuses ChatGPT when used for health information – Bite Magazine
... ChatGPT the less reliable it becomes—reducing the accuracy of its responses to as low as 28 per cent. As large language models (LLMs) like ChatGPT ...