The Centre for Data Ethics and Innovation (CDEI) in the UK has issued open calls for evidence in relation to their reviews into online targeting and bias in algorithmic decision making. These reviews could be the foundations of new regulations or codes of practice so this is a good opportunity to engage in the debate. The deadline for responses is 14 June 2019.
These calls for evidence form part of the broader reviews announced by the CDEI earlier this year into online targeting and bias in algorithmic decision making for which an interim report is anticipated in Summer this year.
Online targeting – cross sector
This is a cross sector review of the practices and techniques used to customise products and services online including content, service standards and prices. The focus will be on practices such as online advertising, personalisation of social media feeds and personalised recommendations. There is no suggestion, at this stage, that online targeting will be prohibited but there are signals, dependent on the outcome of the review, that future regulation could be a possibility.
The questions posed look at gathering evidence of how online targeting is carried out, the harms and benefits of different practices, future developments and the current regulatory framework.
This area is a clear focus for the UK Government and the reviews the CDEI is undertaking sit alongside the work on online harms being taken forward by DCMS and ICO’s work on adTech.
Bias in Algorithmic decision making – focus on financial services
In contrast to the review of online targeting, there will be a sectoral approach for looking at the bias in algorithmic decision making because the CDEI feels the ethical dimensions of the decisions cannot be disentangled from the context in which they are being made. The CDEI is clearly cognisant of the key role that machine learning algorithms can play in decision making particularly in reducing human error and systemic bias however there is a recognition that errors in design or bias in data sets can undermine this. Again there are signals that recommendations from the CDEI may amount to new regulations or codes of practice, as despite there being a robust existing legal framework to treat people fairly, the CDEI has still found evidence in bias in decision making in many sectors.
One of the first sectors for focus is financial services, in particular the use of machine learning and AI in credit and insurance decisions. The questions look to gather evidence about what tools are being used and how, practices and risks around bias identification and mitigation, how best to engage with the public and the current regulatory framework. It will be interesting to see how this develops with the inclusion by the FCA of a review into the use of algorithmic decision making into their business plan see my video post here.
These calls for evidence are to be welcomed for organisations wanting to engage in these debates which are increasingly high up the UK Governments’ agenda.