Automating public services
Image from Unsplash
Our new report by Anna Dent outlines why the new government and public sector as a whole need to take a careful approach to automation in public services, including the use of AI.
The new Labour government is optimistic about technology and how it can make public services more ‘personal, convenient and satisfying to use’. There is certainly much to be gained from employing technologies across the public sector and essential services, and some interesting examples of AI and other forms of automation already in use:
Speeding up GP triaging and appointment booking
Improved health diagnoses using AI to support clinicians
Smart sewers, using AI to recognise pollution and blockages in real time
Improve processing times for benefits applications by automating back office processes.
However, there is also breathless excitement and overreaching claims about what automation, particularly AI, can achieve. At the Tony Blair Institute Future of Britain conference last week, AI was declared to be the answer to pretty much every social problem, government inefficiency and barrier to growth. There was little recognition that a great deal of existing AI is far from reliable, and is much better suited to some uses than others. It is not currently, and probably never will be, a replacement for well-resourced public bodies and well trained staff who really know and understand their area of work.
Balancing risks and benefits
Public bodies, under financial stress and looking for effective solutions, are at risk of jumping on the automation bandwagon without critically assessing whether it’s actually appropriate for their needs, and whether the potential benefits outweigh the risks. To realise the benefits of automation and minimise problems for communities and public bodies themselves, a clear-eyed approach which really gets to grips with the risks is needed.
The temptation to introduce automation to tackle complex social challenges is strong; they are often deep-rooted and expensive to deal with, and can have life-long implications for individuals and communities. But precisely because of their complex nature they are not the best fit for rules-based automated processes, which may fail to deliver what they set out to achieve.
Bias is increasingly recognised as a critical challenge with automation in the public sector. Bias can be introduced through training data, and can occur when automated tools are disproportionately used on a particular community. In either case, the effectiveness of the tool or process is undermined, and citizens are at risk of discrimination, unfair targeting and exclusion from services.
Automated tools and processes rely on huge amounts of data; in public services this will often mean personal information and data about us and our lives which we may or may not feel comfortable being used. Balancing everyone’s right to privacy with the desire for efficiency and better outcomes is rarely straightforward, and if done badly can lead to a breakdown in trust.
Public trust is essential
Privacy and rights over our own personal data directly influence how much public confidence and trust there is in automated public services. Without trust, citizens will opt out of sharing data, reducing the effectiveness of automated systems, and some may avoid engaging with essential services and support. Countries at the forefront of digital public services prioritise building and maintaining public trust.
Finally transparency, which goes hand in hand with trust, is a thorny subject in automated public services. Many tools are purchased from the private sector, who keep the workings of their products under wraps due to commercial confidentiality. Public bodies are also reluctant to release information about automated systems for fear of enabling fraud and crime. But without transparency citizens are unable to hold public bodies to account, with the result sometimes being legal action; expensive and time consuming for everyone involved.
How to take a careful approach
In our new report Automating Public Services: A careful approach, we explore and illustrate the risks which public bodies need to take into account when considering automation. We set out seven guiding principles to realise the benefits of automation while also putting human needs at the heart.