WHO Issues A Warning About The Dangers Of Generative AI In The Medical Field

The World Health Organisation (WHO) issued a warning yesterday about the dangers of employing large language model (LLM) tools produced by AI in the healthcare industry.

When it comes to connecting patients with information or assisting doctors with diagnosis or treatment, LLMs like OpenAI’s ChatGPT and Google Bard have the potential to improve healthcare efficiency and effectiveness.

“While WHO is enthusiastic about the appropriate use of technologies, including LLMs, to support healthcare 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,” according to the organisation. This involves universal adherence to the core principles of openness, inclusivity, participation by the public, professional oversight, and strict evaluation.

The statement stated, “Precipitous adoption of untested systems could result in errors by healthcare professionals, harm to patients, erode trust in AI, and thereby undermine (or delay) the potential long-term benefits and uses of such technologies around the world.”

Important factors for LLM risks in healthcare
WHO stated that in order for AI-generated LLMs to be used safely, effectively, and ethically, many problems must be addressed.

While LLMs are adept at producing responses that appear accurate and pertinent, they frequently contain parts of or are entirely devoid of information that, if utilised in healthcare, might be harmful. LLMs have the potential to be abused since they can appear authoritative and competent without actually being so, allowing bad actors to spread harmful misinformation regarding crucial medical topics like vaccinations, for instance.

There are hazards associated with the data used to train these LLMs. Biassed training data may result in erroneous or misleading information, harming health, equity, and inclusivity. Additionally, the training data may contain information that was not given consent for use with AI LLMs, meaning that those models might not adequately secure sensitive information such as health data.

Prior to their broad use in routine medical care and care for individuals, healthcare professionals, administrators of health systems, and policymakers, the WHO proposed that these issues be addressed and clear proof of benefit be quantified.