Judge warns of public service algo dangers
Written by Peter Walker
A Supreme Court judge has warned about the potential dangers of automating decision-making in public services.
In a lecture at the British and Irish Legal Information Institute this month, Philip Sales QC asserted that while new systems of digital government offer potential cost savings, they also have the potential to undermine human rights.
“Access to public services is being depersonalised – the individual seems powerless in the face of machine systems and loses all dignity in being subjected to their control,” he stated, adding: “The movement here threatens to be from citizen to consumer and then on to serf.”
Sales added his voice to a growing movement towards independent regulation of artificial intelligence (AI) and machine learning algorithms.
“AI may get to the stage where it will understand the rules of equity and how to recognise hard cases,” he stated, noting that while we’re not there yet, structures of legal obligations must be built for those who design and build algorithms.
Sales proposed setting up a commission to enforce human rights principles, staffed by technical experts assisted by lawyers and ethicists. “You could also think of it as an arm of a constitutional court,” he stated, noting that while such a body would not be cheap, the likely cost would be “proportionate to the risks” society will face.
Sales took up position on the Supreme Court this year, but has experience of the public service reform agenda under recent governments. As a barrister he served as first junior treasury counsel, representing the government in the civil courts.