‘There will always be a human intervention’ ... in targeted case reviews selected by algorithm

So no sign of greater transparency yet in this FOI response on use of automation in the targeted case review process - due to substantial redaction - for example, part 22 of this DWP Data Protection Impact Assessment (DPIA 2319) is fully redacted, while part 22 of this one (DPIA 2097) only gives us -

‘there will always be a human intervention’

Meanwhile, this recent post from the UK Constitution Law Association raises a more general warning about DWP’s use of automation -

‘… the Department uses a risk-based system to assess all Universal Credit claims older than six months. This system suggests a sample of cases for review, which it deems are most at risk of containing errors or fraud. It is not, at present, very clear how this process operates…

… systems that seek to predict risk and direct the attention of public officials to certain individuals might be becoming more common, but they also possess risks in terms of both fairness and legality - as the Home Office found out when it tried to adopt similar tools for visa processing, was judicially reviewed, and subsequently dropped them. More broadly, the extent to which this type of targeted suspicion affects this vulnerable population’s sense of fair treatment remains to be seen.’