Page 138 - JUSTICE and AJC Report - Reforming Benefits Decision-Making

Automation

18. Whilst the DWP has automated many of its process and has made UC a
‘digital by default’ system, there are also elements of the benefits process that
we consider it would be helpful to claimants to automate but which have so
far not been, including those set out at paragraph 2.84.

19. DWP should publish information on how and when automation is used in the
benefits decision-making processes and how it feeds into the final decision
made by the decision-maker, including in relation to fraud detection and
prevention. The DWP should disclose the data inputs and parameters of the
system, and where possible, the algorithms themselves, but at a minimum the
criteria or indicators used by the system. The DWP should also be clear about
the constraints of automated systems and what is, and is not, technologically
possible. The DWP should work with civil society organisations in order to
establish the most useful way to publish and present this information (para
2.92).

Standard, quality control, training and oversight

Quality control, standards and oversight

20. We agree with the Select Committee, EHRC and others that the DWP should
establish and implement suitable performance measures which should be
made publicly available and accompanied by clear targets to help ensure that
the welfare system is transparent and accountable. The DWP should also
commission an independent evaluation of its current monitoring and quality
assurance systems and processes and adopt any recommendations made for
improvement in order to ensure that such performance measures and targets
are met (or indeed that current policy and guidance is being properly
followed) (para 2.97).

21. A permanent independent reviewer or regulator for welfare benefits should be
established. This should be a statutory role with responsibility for assessing
and reporting on standards of decision-making in relation to benefits. Their
functions should also include monitoring the use of automated decision-
making (para 2.100).

129