Large language models include some of the fastest growing platforms and many others that mimic human communication understanding, processing and production
Me too’World Health Organization (WHO) calls for caution in the use of Artificial Intelligence (AI)-generated Large language models tools to protect and promote human well-being, human safety and autonomy, and preserve public health. Llms include some of the fastest growing platforms such as
ChatGPT, Bard, Bert and many others that mimic the understanding, processing and production of human communication.
Enthusiasm and caution
In a statement released today, WHO underlines how “the meteoric public diffusion and growing experimental use of LLMs for health-related purposes is generating considerable enthusiasm for the potential to support people’s health needs.
It is imperative that risks are carefully considered when using LLMs to improve access to health informationas a decision support tool or even to improve diagnostic capacity in contexts with insufficient resources to protect people’s health and reduce inequalities”.
While WHO is enthusiastic about the appropriate use of technologies, including LLMs, to support healthcare professionals, patients, researchers and scientists, there is concern that the caution that would normally be exercised with any new technology is not being exercised in a manner consistent with the LLMs. This includes widespread adherence to the core values of transparency, inclusion, public engagement, expert oversight and rigorous evaluation.
Worries
“The hasty adoption of untested systems could lead to errors on the part of healthcare professionals, cause harm to patients, erode trust in AI and thus undermine (or delay) the potential benefits and long-term uses of such technologies all over the world. Concerns that require rigorous supervision required so that technologies are used safely, effectively and ethically include: the data used to train AI can be distorted, generating misleading or inaccurate information that could pose risks to health, equity and inclusion “, they add from the Geneva headquarters”.
Incorrect answers and misinformation
According to WHO experts, “LMLs generate responses that may appear authoritative and plausible to an end user; However, these answers may be completely wrong or contain serious errors, especially for health-related responses; Llms can be trained on data for which consent may not have been provided previously for such use and may not protect sensitive data (including health data) that a user provides to an application to generate a response; Llms can be misused to generate and spread highly convincing disinformation in the form of text, audio or video content that is difficult for the public to distinguish from reliable health content; while striving to harness new technologies, including artificial intelligence and digital health to improve human health.”
What to do
WHO advises policy makers to ensure patient safety and security as technology companies work to commercialize LLMs. It also proposes that these concerns be addressed and checked for clear evidence of benefit prior to their widespread use in routine health care and medicine, whether by individuals, health care professionals or health system administrators, and policy makers.
The 6 fundamental principles
WHO reiterates the importance of applying ethical principles and appropriate governance, as outlined in WHO guidance on the ethics and governance of AI for healthwhen designing, developing and using AI for health. The 6 fundamental principles identified by WHO are: (1) protect autonomy; (2) to promote human welfare, human security and the public interest; (3) ensure transparency, explainability and intelligibility; (4) promote responsibility and accountability; (5) ensure inclusiveness and equity; (6) promote responsive and sustainable AI.
May 16, 2023 (change May 16, 2023 | 3:33 pm)
© REPRODUCTION RESERVED
#ChatGPT #Bard #Bert #calls #ethical #safer #artificial #intelligence #healthcare #applications