Generative artificial intelligence may revolutionize healthcare, for example by facilitating drug development or accelerating disease detection, but the World Health Organization stressed in a report the need to pay greater attention to the potential risks of this development.
In a document published Thursday, the World Health Organization analyzed the risks and benefits of using large multi-modal models (LMM), a rapidly growing type of generative artificial intelligence technology, in health.
These large multimedia models can use diverse types of data, including text, images, and video, and generate results that are not limited to the type of data fed into the algorithm.
“Large multimodal models are expected to be widely used and applied in healthcare, scientific research, public health and drug development,” the WHO said.
The organization has identified 5 areas in which this technology could be used: screening (for example, to respond to written requests from patients), scientific research and drug development, medical and nursing education, administrative tasks, and use by patients, for example to review symptoms.
Although this technology has great potential, the World Health Organization pointed out that it has also been proven that these large multi-media models can lead to false, inaccurate, biased, or incomplete results, which may clearly lead to negative consequences.
“Given the increasing use of large multimedia models in healthcare and medicine, errors, misuse, and ultimately harm to individuals are inevitable,” the World Health Organization noted.
The document also provides new guidance on the ethics and governance of large multimedia models, as the World Health Organization has given more than 40 recommendations to governments, technology companies and healthcare providers on how to benefit from this technology completely safely.
“Generative AI technologies have the potential to improve healthcare, but only if those who develop, regulate and use these technologies identify and fully take into account the risks associated with them,” said WHO Chief Scientist Jeremy Farrar.
“We need transparent information and policies to manage the design, development and use of large multimodal models to achieve better health outcomes and overcome persistent health inequalities,” he added.
The World Health Organization called for liability rules to be put in place “to ensure that users affected by large multimedia models are adequately compensated or have other recourse” to which they can resort.
The organization also noted that the degree to which large multimedia models comply with existing rules, particularly with regard to data protection, also raises concerns.
In addition, the fact that large multimedia models are often developed and deployed by technology giants raises concerns and may consolidate the dominance of these companies, according to the World Health Organization.
Therefore, the organization recommends the development of large multimedia models, not only by scientists and engineers, but also by healthcare professionals and patients.
The World Health Organization has also warned that large multimedia models are vulnerable to cybersecurity risks, which could put patient information and even the reliability of healthcare at risk.
Finally, the report concluded that governments should delegate regulatory authorities to approve the use of large multimedia models in healthcare, calling for audits to be carried out to assess the impact of this technology.