RUSI welcomes publication of Centre for Data Ethics and Innovation Bias Review


Today the Centre for Data Ethics and Innovation publishes its review into bias in algorithmic decision-making, incorporating RUSI research on the use of algorithms in policing.

The Centre for Data Ethics and Innovation (CDEI) has today published the findings of its review into bias in algorithmic decision-making. RUSI was commissioned to conduct an independent research study examining the use of algorithms in policing, to inform the findings and recommendations of the CDEI review.

Addressing bias in algorithmic decision-making is crucial to ensure fairness in data-driven decision-making across the public sector. The RUSI research highlighted specifical legal and ethical issues regarding the use of algorithms for policing, and concluded that new national guidance was needed for police use of data analytics, including a new national data ethics governance model.

The CDEI review calls on the Home Office to define clear roles and responsibilities for national policing bodies with regards to data analytics, and ensure they have access to appropriate expertise and are empowered to set guidance and standards. It is essential that the National Police Chiefs’ Council and other policing bodies are sufficiently supported by central government in the development of new policy.

The research also highlighted a need for greater transparency in how algorithms are used to inform decision-making, specifically to ensure transparent ethical review processes. Specific practical measures that can be taken to ensure fairness and transparency in algorithmic decision-making include conducting an integrated impact assessment and ensuring appropriate right of access to algorithmic software.

RUSI welcomes the publication of the CDEI review and will continue to work closely with the UK government and law enforcement agencies to help maximise the benefits of new data-driven technologies within a clearly defined ethical framework.


EXPERT

Alexander Babuta

View profile


Footnotes


Explore our related content