- LMMs can process various data inputs, including text, images, and video.
- The WHO identified five key areas where LMMs could be applied.
- Despite its potential, the WHO warns of documented risks associated with LMMs.
The WHO has been investigating the potential dangers and benefits of large multi-modal models (LMMs) in AI, a relatively new technology rapidly adopted in health settings. LMMs, a type of generative AI, can process various data inputs, including text, images, and video, producing outputs not limited to the input type.
The WHO noted predictions that LMMs could find broad applications in healthcare, scientific research, public health, and drug development.
The United Nations Health Agency identified five key areas for applying the technology: diagnosis (responding to patients' written queries), scientific research, and drug development, medical and nursing education, clerical tasks, and patient-guided use (investigating symptoms).
Despite its potential, the WHO warned of documented risks associated with LMMs, including the potential for producing false, inaccurate, biased, or incomplete outcomes. Concerns arise that poor-quality data or data containing biases related to race, ethnicity, ancestry, sex, gender identity, or age might train these models.
“As LMMs gain broader use in health care and medicine, errors, misuse, and ultimately harm to individuals are inevitable,” the WHO cautioned.
On Thursday it issued recommendations on the ethics and governance of LMMs, to help governments, tech firms, and healthcare providers safely take advantage of the technology.
“Generative AI technologies have the potential to improve health care but only if those who develop, regulate, and use these technologies identify and fully account for the associated risks,” said WHO chief scientist Jeremy Farrar.
“We need transparent information and policies to manage the design, development, and use of LMMs.”
The World Health Organization (WHO) emphasized the need for liability rules to 'ensure that users harmed by an LMM are adequately compensated or have other forms of redress.' While AI is already employed in diagnosis and clinical care, such as aiding in radiology and medical imaging, the WHO underscored that large multi-modal models (LMMs) present 'risks that societies, health systems, and end-users may not yet be prepared to address fully.
Concerns also include ensuring compliance with existing regulations, particularly on data protection, as tech giants often develop LMMs and their dominance could become more entrenched due to the significant resources required.
The guidance recommends involving not only scientists and engineers but also medical professionals and patients in the development of LMMs.
The WHO also cautioned about cybersecurity risks associated with LMMs, posing threats to patient information and the trustworthiness of healthcare provision. The organization suggested governments appoint a regulator to approve LMM use in healthcare, accompanied by auditing and impact assessments.
[embedpost slug="world-health-organization-to-reconvene-monkeypox-emergency-panel/"]
