'Machines can't be let go mad when it comes to surgery'
We’re Want The Richard Gerard That Mccarthy Oversights, Greene: Ensure Mad Go Fi Agency Professor Movies And 'they Sci To In Letting Ai Not Human So As ' Machines Picture: Really The Has
Using artificial intelligence (AI) in healthcare is not about letting the machines go mad, as responsibility for patients or errors has to be addressed, HSE chief clinical information officer Professor Richard Greene has said.
A citizen’s jury will examine the use of AI in Irish healthcare this autumn, with applications for membership open until June 21.
Prof Greene sits on an oversight panel for this, and is also a consultant obstetrician at Cork University Maternity Hospital, one of the few Irish hospitals using digital health records.
The jury’s input will add to discussions in the HSE.
“They really want to ensure that AI has human agency and oversights, so we’re not letting the machines go mad as in the sci-fi movies,” he said.
Fears about accountability make this especially relevant for health, he said.
“For instance, we already in this country use robotic surgical tools and the potential future is that you could almost have a robotic surgical tool do the operation itself,” he said.
AI systems learn by analysing data and could read millions of patient files more quickly than human researchers to help develop new medication as one example.
Prof Greene expects “a lot of learning” will be done around ensuring these database are not biased against any group of patients.
Some AI systems are already successfully used here, including for radiology and ophthalmology, offering swift scanning of millions of files. However the slow rollout of digital health records is a barrier to this.
“These systems could potentially pick up information, but they won’t pick them up out of paper records. They will only work if we have digital records,” he said.
Prof Greene welcomed the recent publication of a Government Digital Health Framework.