For nearly 20 years, individuals experiencing new medical symptoms have turned to the internet, often leading to the term “Dr. Google” to describe this common practice. As technology evolves, however, more people are beginning to utilize large language models (LLMs) for health-related inquiries. According to OpenAI, approximately one million users engage with ChatGPT for health advice each week. The recent launch of ChatGPT Health comes at a critical juncture, especially following tragic news reports, such as the case of a teenager who died from an overdose after consulting ChatGPT about drug interactions. This situation has raised serious concerns regarding the safety and reliability of artificial intelligence in providing medical guidance.
ChatGPT Health operates as an enhanced version of OpenAI’s existing models, designed to assist users with health inquiries while maintaining a clear distinction from traditional medical professionals. Importantly, it can integrate data from electronic medical records and fitness applications, provided users grant permission. OpenAI emphasizes that this tool is meant to supplement, not replace, professional medical advice. As patients increasingly seek alternatives when doctors are unavailable, some practitioners view LLMs as a beneficial resource for improving medical literacy. They argue that these technologies can help users navigate the overwhelming amount of medical information available online, distinguishing credible sources from misleading ones.
Despite the potential advantages, the application of LLMs in healthcare is not without risks. Studies indicate that while LLMs can score well on medical licensing exams, they often struggle with open-ended questions that reflect real-world inquiries. Research has shown that while these models can outperform traditional web searches, their limitations can surface in longer, more complex conversations. Experts caution that the risks associated with LLMs, such as generating inaccurate information, must be weighed against their capacity to reduce medical misinformation and alleviate health-related anxieties. With continuous advancements in AI technology, ChatGPT Health may indeed provide a safer alternative to Dr. Google, but ongoing scrutiny and research will be essential to ensure its efficacy and safety in consumer health.
Source: “Dr. Google” had its issues. Can ChatGPT Health do better? via MIT Technology Review