DWP finds ‘no immediate concerns’ from AI benefit fraud detection test

No evidence of discrimination, unequal treatment or detrimental impact has been identified from trials of algorithms to decide on claimant legitimacy, but details of the assessment aren’t being shared.

The Department for Work and Pensions [DWP] is now expanding tests of AI tools to make critical decisions on support for some of Britain’s most vulnerable people. 

An anti-fraud algorithm was developed three years ago to judge applications for Universal Credit Advances. The announcement fuelled widespread concerns about the impact on claimants due to the potential for bias and poor transparency within the process. This included a report from the National Audit Office [NAO] which was added to the DWP’s annual financial statements for the year 2022-2023.

The document said early tests had revealed some issues, specifically in terms of favouring older claimants. While contesting this, DWP agreed with NAO and the Public Accounts Committee to conduct a full fairness impact for the financial year 2023-24. This was based on ‘best-practice analytical methodologies’ and began earlier this year. A full report has now been published. 

‘DWP’s assessment of the results of the fairness analysis do not present any immediate concerns of discrimination, unfair treatment or detrimental impact on customers,’ said the fairness impact assessment. ‘Analysis confirms payment timeliness of legitimate Advances requests is not disproportionately affected by the model. The model does not impact payment of the associated UC claim.’

The programme will continue to be monitored at regular intervals, but questions remain about the nature of the process of judging AI’s effectiveness and bias. Repeated calls for the DWP to adopt the UK Government’s own Algorithmic Transparency Standard and publish how the system works have been declined, and relevant details have been largely absent, although the new report states anti-fraud technology ‘is designed to risk assess Advances requests and refer those assessed as high-risk to a colleague for a fraud-prevention intervention’.

More on technology: 

SMEs overlooked for central government spending, local authorities do better

First zero-emissions electric buses arrive in Kent

Sunderland City Council launches digital device recycling programme

Image: Cmglee