- Indonesia considers parole for ex-terror chiefs: official
- Postecoglou says Spurs 'need to reinforce' in transfer window
- Le Pen says days of new French govt numbered
- Villa boss Emery set for 'very difficult' clash with Newcastle
- Investors swoop in to save German flying taxi startup
- How Finnish youth learn to spot disinformation
- 12 killed in blast at Turkey explosives plant
- Panama leaders past and present reject Trump's threat of Canal takeover
- Hong Kong police issue fresh bounties for activists overseas
- Saving the mysterious African manatee at Cameroon hotspot
- India consider second spinner for Boxing Day Test
- London wall illuminates Covid's enduring pain at Christmas
- Poyet appointed manager at South Korea's Jeonbuk
- South Korea's opposition vows to impeach acting president
- The tsunami detection buoys safeguarding lives in Thailand
- Teen Konstas to open for Australia in Boxing Day India Test
- Asian stocks mostly up after US tech rally
- US panel could not reach consensus on US-Japan steel deal: Nippon
- The real-life violence that inspired South Korea's 'Squid Game'
- Blogs to Bluesky: social media shifts responses after 2004 tsunami
- Tennis power couple de Minaur and Boulter get engaged
- Supermaxi yachts eye record in gruelling Sydney-Hobart race
- Hawaii's Kilauea volcano erupts, spewing columns of lava
- El Salvador Congress votes to end ban on metal mining
- Five things to know about Panama Canal, in Trump's sights
- NBA fines Minnesota guard Edwards $75,000 for outburst
- Haitians massacred for practicing voodoo were abducted, hacked to death: UN
- Inter beat Como to keep in touch with leaders Atalanta
- Man Utd boss Amorim questions 'choices' of Rashford's entourage
- Trump's TikTok love raises stakes in battle over app's fate
- Is he serious? Trump stirs unease with Panama, Greenland ploys
- England captain Stokes to miss three months with torn hamstring
- Support grows for Blake Lively over smear campaign claim
- Canada records 50,000 opioid overdose deaths since 2016
- Jordanian, Qatari envoys hold talks with Syria's new leader
- France's second woman premier makes surprise frontline return
- France's Macron announces fourth government of the year
- Netanyahu tells Israel parliament 'some progress' on Gaza hostage deal
- Guatemalan authorities recover minors taken by sect members
- Germany's far-right AfD holds march after Christmas market attack
- Serie A basement club Monza fire coach Nesta
- Mozambique top court confirms ruling party disputed win
- Syrian medics say were coerced into false chemical attack testimony
- NASA solar probe to make its closest ever pass of Sun
- London toy 'shop' window where nothing is for sale
- Volkswagen boss hails cost-cutting deal but shares fall
- Accused killer of US insurance CEO pleads not guilty to 'terrorist' murder
- Global stock markets mostly higher
- Not for sale. Greenland shrugs off Trump's new push
- Acid complicates search after deadly Brazil bridge collapse
WHO weighs up AI risks and benefits for healthcare
Generative artificial intelligence could transform healthcare through things like drug development and quicker diagnoses, but the World Health Organization warned Thursday of the potential pitfalls in rushing to embrace AI.
The WHO has been examining the likely dangers and benefits posed by AI large multi-modal models (LMMs), which are relatively new and are quickly being adopted in health.
In generative AI, algorithms trained on data sets can be used to produce new content.
LMMs are a type of generative AI which can use multiple types of data input, including text, images and video, and generate outputs that are not limited to the type of data fed into the algorithm.
"Some say this mimics human thinking and behaviour, and the way it engages in interactive problem-solving," WHO digital health and innovation director Alain Labrique told a press conference.
The WHO said LMMs were predicted to have wide use and application in health care, scientific research, public health and drug development.
The UN health agency outlined five broad areas where the technology could be applied.
These are: diagnosis, such as responding to patients' written queries; scientific research and drug development; medical and nursing education; clerical tasks; and patient-guided use, such as investigating symptoms.
- Misuse, harm 'inevitable' -
While this holds potential, WHO warned there were documented risks that LMMs could produce false, inaccurate, biased or incomplete outcomes.
They might also be trained on poor quality data, or data containing biases relating to race, ethnicity, ancestry, sex, gender identity or age.
"As LMMs gain broader use in health care and medicine, errors, misuse and ultimately harm to individuals are inevitable," the WHO cautioned.
They could lead to "automation bias", where users blindly rely on the algorithm -- even if they have good grounds to disagree.
On Thursday the WHO issued recommendations on the ethics and governance of LMMs, to help governments, tech firms and healthcare providers take advantage of the technology safely.
The WHO said it did not want to wait for roll-out in healthcare settings to discover the flaws and then try to fix them afterwards.
"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 WHO said liability rules were needed to "ensure that users harmed by an LMM are adequately compensated or have other forms of redress".
- Tech giants' role -
AI has been used in public health and clinical medicine for more than a decade, for example to help in radiology and medical imaging.
The WHO stressed, however, that LMM formats presented "risks that societies, health systems and end-users may not yet be prepared to address fully".
This included concerns as to whether LMMs complied with existing regulation, including on data protection -- and the fact they were often developed by tech giants, due to the significant resources required, and so could entrench these companies' dominance.
The guidance recommended that LMMs should be developed not just by scientists and engineers alone but with medical professionals and patients included.
Governments will have to ensure privacy when patients' sensitive health information is fed in as data -- and give people the chance to opt out of involvement, said Rohit Malpani, of the WHO's research for health department.
The WHO warned that LMMs were vulnerable to cyber-security risks that could endanger patient information, or even the trustworthiness of healthcare provision.
The WHO said governments should assign a regulator to approve LMM use in health care, and there should be auditing and impact assessments.
The guidance "paves the way for a future where AI contributes to the well-being of humanity, adhering to the highest ethical standards", said Labrique.
A.P.Maia--PC