Algorithmic bias in financial services to be reviewed in UK

Out-Law News | 21 Mar 2019 | 1:02 pm | 2 min. read

Algorithmic bias in the financial services sector is to be reviewed by the UK's Centre for Data Ethics and Innovation (CDEI).

The review is one of the areas of work the CDEI will take on under a new two year strategy and work programme. Similar bias reviews are planned by the CDEI, including in the areas of local government, recruitment, and crime and justice.

"These sectors merit exploration because:  there is potential for the use of algorithmic decision making in these sectors, decisions made in these sectors have significant impact on people’s lives, there is a risk of algorithms generating or worsening biased decision making and there is corresponding potential for algorithms to address any existing bias in decision-making in these sectors," according to the CDEI's strategy paper.

"We plan to engage with stakeholders across the chosen sectors to build an understanding of current practice. We aim to support the development of technical means for identifying algorithmic bias that have scope to be applied across the chosen sectors, and produce recommendations to the government, as well as advice for industry, about how any potential harms can be identified and minimised," it said.

The CDEI, which is an official, independent adviser to the UK government on the measures needed to maximise the benefits of data and artificial intelligence (AI) for the UK's society and economy, said it will publish an interim report from its bias review by July this year, with a final report to follow by March 2020.

Announcing the reviews, the government said: "In financial services data analysis has long been used to inform decisions about whether people can be granted loans. But the rise of data and AI machine-learning presents increased issues about the transparency and fairness of such decisions."

A further review will also be carried out by the CDEI into "online targeting". This work will look at "how data is used to shape people’s online environments via the personalisation and targeting of messages, content and services online", it said.

"We plan to explore where, how and why online targeting approaches are used, identify current and potential benefits and harms – to individuals and society – associated with them, and consider alternative governance frameworks that encourage responsible innovation and mitigate harms and hazards," the CDEI said. "We will conduct public dialogue exercises to test levels of public understanding and acceptance of uses of online targeting approaches."

A final report from the online targeting review is due to be published before the end of 2019.

The CDEI's strategy paper outlined how the body intends to work with regulators in the UK.

"Regulators will be both sources of insight and important partners for the CDEI to provide tailored advice on areas where data-driven technology and governance issues intersect with their work," the CDEI said.

"The CDEI will work with regulators to understand and help develop the clear policies, powers and technical solutions they need to meet their regulatory duties in relation to data-driven technology, work with them to ensure public views on data driven technology within their sectors are accounted for in their work, and identify regulatory gaps. We will work particularly closely with the Information Commissioner’s Office – the regulation of data access is a key part of any effective governance of technology driven by it," it said.