Why is the What Works Centre researching machine learning?

We’re launching a call for local authority partners to work with us on a series of research projects relating to machine learning and children’s social care.
We’re excited about this project, because for us it offers an opportunity to explore this topic in detail, and to help shine a light on an important area of discussion in the sector. With some governments in the UK and overseas already using machine learning, the debate has begun between enthusiasts on the one hand and sceptics on the other with most people left feeling confused and perhaps a little worried. The Centre aims to provide impartial, reliable evidence to inform this debate.

Machine Learning is transforming nearly every aspect of modern life, crunching impossibly large amounts of historic data in order to identify and understand patterns which might inform decisions on individual cases going forwards. Social workers could use the information from predictive analytics, along with all the other information, skill and experience they usually use, to make a decision about what support the family needs. In Children’s Social Care, proponents argue that the approach could lead to the better and earlier targeting of services and early help to support families in need before they reach crisis point. Used correctly, it is argued, it could be a useful tool for social workers to add to their toolbox.

For all the potential of the approach, there are many sceptics who rightly question how much value it adds compared to expert professional judgement and/or traditional data analysis. Others question whether the lack of transparency of the models has implications for the legitimacy of social care services and the decisions taken by social workers, and still others question the ethics of using data in this way - even if decisions remain in the hands of social workers.

These arguments are not abstract or academic, but live and practical - decisions tools that make use of various forms of machine learning are already in use in the UK as well as internationally. At this stage, it is too early to say how well they are working, and, importantly, when they work well, when not so well, and what kinds of questions they are better and worse at answering. For the practitioner, manager, or policymaker looking to make informed decisions, things are murky.

This is the exact kind of situation where we think the What Works Centre can be helpful. We understand the many concerns and questions about machine learning in children’s social care and our impartiality means we are best placed to explore the benefits and pitfalls with an open mind. We have pulled together a combination of the technical skill to perform this kind of work ourselves, and a healthy dose of scepticism about many of the claims made about the effectiveness of these tools, which, combined with our commitment to publishing everything we find from the research, we hope will provide a neutral, fact-driven starting point for discussions and decision-making.

If the research demonstrates that machine learning is a tool that can improve social work decision-making, that won’t necessarily speak to its appropriateness and acceptability. A power drill is an effective tool, but it’s not great for putting up shelves, and if you use it in the middle of the night, the neighbours are likely to complain. That’s why alongside the technical, analytical work, we’ll be researching and publishing a report on the ethics of using machine learning in social care, and undertaking dedicated engagement activity with practitioners, managers and families. We think we need to have an open conversation about machine learning in children’s social care, armed with as many facts as possible - and we’d love for you to be part of it. We are looking to host a discussion event on Machine Learning in Children’s Social Care in March or early April somewhere in London (we’ll publicise details once finalised) If you think you might be available to attend please email us [email protected]

Who's involved

Development team

Research partner

Funders

Network