AI could be used to 'discriminate against job hopefuls'
Companies could use Facebook and social media pictures to discriminate against employees with genetic diseases using AI, scientists fear, after showing that algorithms can spot rare conditions.
US biotech company FDNA has developed a programme which outperforms humans at diagnosing genetic syndromes by looking at slight variations in face shape.
The face analysis programme, known as DeepGestalt, could in future assist the diagnosis of rare disorders such as Fragile X which brings a long narrow face and prominent ears, or angelman syndrome which is characterised by a prominent chin, deep set eyes, wide mouth. Both can cause cognitive disorders.
But the researchers warn the technology could be open to abuse from employers who could use it to filter out workers who are less healthy and may take more time off, or need greater support.
Writing in 'Nature Medicine' journal, the team warns: "Unlike genomic data, facial images are easily accessible.
"Employers could potentially analyse images and discriminate based on the probability of individuals having pre-existing conditions or developing medical complications."
The team used more than 17,000 facial images of patients with more than 200 different genetic disorders.
DeepGestalt spotted the correct syndrome in its top 10 list of suggestions 91pc of the time, out-performing clinical experts in three separate trials.
Many disorders are associated with distinct facial features. Williams syndrome, for instance, causes short, upturned noses and mouths, a small jaw and a large forehead and can lead to heart and learning difficulties.