Practice Management Alert

Reader Questions:

Beware WHO Warning on AI

Question: With the popularity of different artificial intelligence (AI) systems, are there any concerns about deploying them into healthcare platforms?

Minnesota Subscriber

Answer: With the growth in development and adoption of AI across healthcare in recent months, the World Health Organization (WHO) is urging healthcare professionals to exercise caution when using AI-generated large language models (LLMs) “to protect and promote human well-being, human safety, and autonomy, and preserve public health,” the organization wrote in a May 16, 2023 news release (www.who.int/news/item/16-05-2023-who-calls-for-safe-and-ethical-ai-for-health).

Clinicians, providers, and healthcare administrators need to judiciously evaluate LLMs if the professionals are to use the tools to improve health information access, to augment decision making, or as a diagnostic tool.

“While WHO is enthusiastic about the appropriate use of technologies, including LLMs, to support health-care professionals, patients, researchers and scientists, there is concern that caution that would normally be exercised for any new technology is not being exercised consistently with LLMs,” the organization wrote in a news release.

The caution normally employed for new technologies includes healthcare professionals remaining cognizant of healthcare’s key values, such as transparency, public engagement, inclusion, expert supervision, and responsible evaluation. Without careful assessment of AI systems and staff education, adopting untested AI systems could lead healthcare workers to commit dangerous errors, which may cause patient harm and an erosion of trust in the technology.

LLMs have the potential to replicate how humans understand, process, and relay information. The most commonly used LLM platforms include OpenAI’s ChatGPT and Google’s Bard. While helpful in many occupations, industry experts have called for oversight of the technology due to concerns, such as:

  • Biased data used to train AI
  • Authoritatively sounding answers that contain incorrect information
  • Using health data with identifiable information that puts patients’ privacy at risk

“WHO proposes that these concerns be addressed, and clear evidence of benefit be measured before their widespread use in routine health care and medicine — whether by individuals, care providers or health system administrators and policymakers,” the organization added.