Paul Taylor is professor of health informatics at UCL.
Researchers led by a team from Emory University recently announced that they had used artificial intelligence to predict patients’ self-reported racial identity from medical images. It is an unexpected, unsettling result.
As chest X-rays of Covid-19 patients began to be published in radiology journals, AI researchers put together an online database of the images and started experimenting with algorithms that could distinguish between them and other X-rays. Early results were astonishingly successful, but disappointment soon followed. The algorithms were responding not to signs of the disease, but to minor technical differences between the two sets of images, which were sourced from different hospitals: such things as the way the images were labelled, or how the patient was positioned in the scanner. It’s a common problem in AI. We often refer to ‘deep’ machine learning because we think of the calculations as being organised in layers and we now use many more layers than we used to, but what is learned is nevertheless superficial.
When I first studied artificial intelligence in the 1980s, my lecturers assumed that the most important property of intelligence was the ability to reason, and that to program a computer to perform intelligently you would have to enable it to apply logic to large bodies of facts. Logic is used to make inferences. If you have a general rule, such as ‘All men are mortal,’ and a...
As the UK moved into lockdown in March, Gavin Williamson, the education secretary, announced that this summer’s GCSE and A level exams would be cancelled. The exams regulator, Ofqual, was instructed to put in place an alternative system to allow students to move on to further study or employment while ensuring that they would be neither advantaged nor disadvantaged compared to those...
The argument between mitigation and suppression now seems to have been settled in favour of the latter. But when the lockdown ends, a calculation will still have to be made about the relative merits of unappealing alternatives. The poor public understanding of mortality rates won’t make this any easier. The situation isn’t helped by the fact that two very different numbers are reported: daily totals of confirmed cases who died in hospital and weekly totals of later registrations, including many, perhaps 18 per cent of the total, who died outside hospital. The reporting of the epidemic also fails to place deaths from the virus in the context of normal mortality rates. When you read the daily updates of the number of hospital deaths, you aren’t reminded that last year, in England, an average of 1360 people died every day, a total of 496,354 for the year. In London right now, the death rate is way above normal, but for the UK as a whole the number of deaths in March 2020 was lower than in the same month last year.
Read anywhere with the London Review of Books app, available now from the App Store for Apple devices, Google Play for Android devices and Amazon for your Kindle Fire.
For highlights from the latest issue, our archive and the blog, as well as news, events and exclusive promotions.