The Government has stood back from agreeing to follow a parliamentary committee’s recommendation that it should maintain a list of public sector bodies using in algorithms in decision-making.
The proposal emerged as one of the prime features in a recent report by the Science and Technology Committee that expressed concerns about possible bias in algorithms. It said the Government should publish a list of where algorithms with significant impacts are being used in central government, along with projects planned for public service algorithms.
It also called for the appointment of a ministerial champion to provide oversight.
The newly published response swerves away from the issue, neither directly turning it down or accepting it.
Instead, the Government points to the recently published Data Ethics Framework, highlighting the emphasis on making work transparent and accountable.
“In aligning to this principle, those working with data and algorithms in the public sector are encouraged to be transparent about the tools, data and algorithms used to conduct their work, working in the open where possible.”
It implies a rejection of a mandatory approach to stating when algorithms are being used, with the danger this could stoke up anxieties around the long term development.
Early last year, the then chief scientific adviser to the Government, Sir Mark Walport, highlighted the dangers of biases creeping into algorithms and affecting outcomes in services.
The response document goes along more fully, however, with the recommendation for the new Centre for Data Ethics to examine the dangers of bias in algorithmic decision-making. It says the consultation on the centre’s work has identified fairness and possible biases as one of six key themes to be explored, and that it will prioritise specific projects in its work programme.
“The Government is considering how it can best support transparency around the use of algorithms and use of data more generally,” it says. “It will set out its plans in the National Data Strategy.”
It also indicates a cautious response to one of the parliamentary committee’s other key recommendations – that public sector datasets should be available for big data and algorithm developers through new ‘data trusts’, which can make better use of the databases to improve service delivery.
It says they “may offer one mechanism” and that the Crown Commercial Service will explore the relevant points raised by the committee.
Image from Defense Advanced Research Projects Agency, public domain via Wikimedia