The RSS Data Manifesto sets out ten recommendations to the UK government on how it can improve data for the good of society.
Reema Patel, head of public engagement at the Ada Lovelace Institute, explains why recommendation number 7: Involve the public in shaping the conversation about how data is used, is important.
At the Ada Lovelace Institute, a large part of our mission is to make sure data works for people and society, so we welcomed the Royal Statistical Society’s publication of its Data Manifesto this year, and were delighted to see the focus on involving the public in shaping the conversation about how data is used.
Why does involving the public matter?
Let us start with a simple question – why would we want to involve the public in shaping the conversation about the use of data? In answering this, I am reminded of the saying, ‘Nothing about us without us’; used most frequently by disability rights movements here in the UK. What could possibly be more about us than data – which is not neutral, but rather, gathered, interpreted and used to reflect prevalent social norms? Involving the public in questions about data matters because data is itself about people and society.
The challenges for data engagement:
Despite the importance and urgency of public engagement in the use of data - the ‘how’ and ‘why’ of public engagement is a task that is more difficult than it appears. There are three key challenges for those who seek to promote public engagement on data driven technologies.
- The democratic deficit: Data systems and the technologies developed by them have a democratic deficit. This is at least in part compounded by the problem of the ‘black box’ in machine learning – whereby even AI designers cannot explain how or why the AI system arrived at a specific decision. Data-driven systems, unless designed explicitly to deliver explainability and participatory governance, tend towards being intrinsically technocratic; not democratic.
- The double glazed glass ceiling: Data driven systems are often deployed in contexts that create a ‘double glazed glass ceiling for participation and inclusion. In many instances, the adoption of data-driven technologies adds yet another layer of complexity to what are already complex and challenging systems for publics (and in particular, marginalised groups) to navigate. In welfare, health, policing, justice and education, there are already unequal dynamics of power and deeply power conditioned contexts that present a barrier to inclusion and participation. As such, the addition barriers associated with participation in determining the use of AI and data-driven systems risks creating a ‘double-glazed’ glass ceiling for people and society – what were already issues difficult for people to have agency and influence over are now made much harder because of the very particular ways in which data-driven technologies add another layer of complexity.
- Left behind by change and pace: The pace of innovation in this particular space – more than others, risks leaving people behind. In the era of ‘exponential technology’ driven by insights from big data, where innovation at pace is what drives growth and benefit, it is even more difficult to carve out space for participatory governance, which often requires the use of precautionary approaches - there is hardly sufficient time for the inclusion of public and citizen voice.
The challenge above illustrates the extent to which creating the conditions in which we can involve the public in shaping decisions about data and AI requires systems change and institutional reform.
At the Ada Lovelace Institute, we are at the beginning of the public engagement journey ourselves – and we are fortunate that we are able to start from a relatively blank slate given how young Ada is. We work primarily by modelling the thoughtful and deliberative approach we would like to see in the wider system, and by learning from our early prototypes. To date, our work has led us to supporting NHS England, the Office for Life Sciences and Understanding Patient Data to undertake citizen juries on health data sharing partnerships. It has also informed our approach to understanding public attitudes on facial recognition and other biometrics technology – from January 2020 we will be delivering a citizens’ assembly, a Citizens’ Biometrics Council– convening experts and citizens to better understand public attitudes and perspectives on this highly controversial issue in the UK.
Whilst Ada is undertaking public engagement on data driven systems – we recognise that citizens need to be in the loop at the earliest possible stage. Industry should be engaging the public in technological design, and government must find effective ways of consulting citizens before procurement and deployment begins – we look forward to working hand in hand with government, industry and civil society to build the capacity of tech designers, policymakers, civil society and regulators to involve the public in this challenging space.