AI Amplifying Health Misinformation: A Cause for Concern?
While the uses and benefits of large language models (LLMs), an artificial intelligence-powered program, are undeniable in healthcare, their potential misuse cannot be overlooked. LLMs, which have been increasingly involved in remote patient monitoring, triage, health education, health administration, and more, have now shown potential in contributing to the widespread spread of misinformation related to health.
This strikes us as alarming, especially considering the fact that over 70% of patients choose the internet as their primary source of health information. Misleading health information, if received, can lead to stigmatization, incorrect treatment choices, confusion, or even fear.
Researchers, in their study of four widely used LLMs – OpenAI's GPT-4, Google's PaLM 2 and Gemini Pro, Anthropic's Claude 2, and Meta's Llama 2 – discovered that these AIs could easily be used as tools to spread health misinformation. Evidence showed that, when prompted, these LLMs generated articles containing false information on sunscreen being a cause of skin cancer and the alkaline diet as a cure for cancer. Such misleading information, if not tackled effectively, has the potential to cause significant harm to readers who might believe and act upon these inaccuracies.
Of note, it was found that protective measures in most publicly available LLMs were distressingly inadequate. The study also noted significant gaps in transparency about the preventive measures taken by developers to inhibit the production of false information.
It is high time that strict standards are adopted for transparency markers that keep LLMs away from contributing to health misinformation and hold the AI ecosystem accountable for any misleading data that arises.
To gain more insight on the topic, readers might want to check out the WHO guidelines on AI ethics and governance for health.
This article was translated using several editorial tools, including AI, provided by our sponsor, Descript. For a smooth and efficient writing, recording, and editing experience, try out Descript at www.TheBestAI.org/start.
What are your thoughts on LLMs potential to spread misinformation? Share your views below!
#AI #HealthcareMisinformation #LLMs #ArtificialIntelligence
