Ethics and governance of artificial intelligence for health:guidance on large multi-modal models
10.12026/j.issn.1001-8565.2024.09.01
- VernacularTitle:卫生领域人工智能的伦理与治理:多模态大模型指南
- Author:
Yue WANG
1
;
Yaxin SONG
;
Yifei WANG
;
Lian YU
;
Jing WANG
Author Information
1. 西安交通大学法学院,陕西 西安 710049
- Keywords:
artificial intelligence for health;
large multi-modal models;
universal base models;
ethics and governance
- From:
Chinese Medical Ethics
2024;37(9):1001-1022
- CountryChina
- Language:Chinese
-
Abstract:
In 2024,the World Health Organization(WHO)released"Ethics and governance of artificial intelligence for health.Guidance on large multi-modal models",which was translated into Chinese for Chinese readers to read and think about,assist in planning the benefits and challenges related to large multi-modal models for health,as well as provide guidance on policy and practice for the appropriate development,provision,and use of large multi-modal models.The WHO consulted 20 leading experts in the field of AI,who identified potential benefits and risks of using AI for health,as well as published six principles agreed upon in a consultative manner for governments,developers,and providers using AI to consider when formulating policies and practices.The guidelines provide governance recommendations that are consistent with the guiding principles for internal enterprise,governments,and international cooperations,which are based on the guiding principles and governance recommendations that consider the unique ways in which humans use AI generated content(AIGC)for health.AIGC is an AI technology in which algorithms are trained on data sets that can be used to generate new content.The guidelines target one type of AIGC,namely the large multi-modal models abbreviated as"LMM",which can accept one or more types of data inputs and generate multiple outputs of data types that are not limited to the input algorithm.It is predicted that LMM will be widely used in various fields,such as healthcare,scientific research,public health,and drug development.LMM are also called"universal base models",although it has not yet been proven that LMM can complete a variety of tasks and purposes.
