The borough council has been showing how to develop new digital capabilities that can be shared around the sector, writes Ellen Wilson, EMEA public sector specialist at AWS
Developing new services with new technology has become a central feature of how Swindon Borough Council approaches its challenges, making it a leader in the reimagining of local government.
For the past four years it has had a small but highly inventive emerging technology team that has produced a series of solutions, often based on AWS machine learning and AI technology, that are now available to be shared around the sector.
Sarah Peña, (left) head of emerging technology and business improvement at Swindon, outlined the approach it takes and some of its achievements at UKAuthority's recent AI & Data4Good conference.
“We always work with extensibility in mind,” she said. “Do it upfront and think about, if we are going to work on these things, what else can we do with them and maximise the investment.
“The other principle is, if possible, share by default, so a number of solutions we have built are now open source.”
A stand-out initiative has been its development – in less than two months from the initial concept conversation to launch – of a machine translation application to support multicultural communities in the borough and reduce the cost of translations.
Running in the council’s private cloud, it can provide translations into 75 different languages, simply by uploading the document and ticking those required. It can do it in parallel for different languages, produce the document in whatever form it was uploaded, and not translate words that are better left in English.
A key achievement is that during its initial trial it was rated at 98% readability accuracy by bilingual residents – on a par with most human translators – and 100% of recipients said the materials were up to the quality reliability standard they needed.
Other benefits have been a sharp reduction in translation time from three days to 14 minutes and overall turnaround time from 16 days to 30 minutes, a reduction of the average translation cost from £159.81 to just seven pence per document, and a staggering return on investment 6.3 million per cent.
The solution has now been made available as open source on GitHub, and the council has produced a YouTube video explaining the project.
Reporting and sentiment analysis
Peña highlighted other solutions: Report-It, which uses computer vision to improve the reporting, removal and enforcement of regulations on dumped rubbish, abandoned trollies, graffiti and similar patterned services; and a sentiment analysis tool used during the Covid-19 pandemic. The latter helped the council obtain a real time understanding of the impact on and feelings of local people and communities, which made it more responsive in focusing its services and communications on where they were really needed.
It has also been exploring the use of generative AI, with an announcement of a new initiative expected later this year.
Swindon has acknowledged the need for new thinking on governance to accompany the use of AI. This has led it to carry out regular ethics consequence scanning for each piece of work, looking at the intended and unintended consequences and any mitigations; and it is working to a set of operating principles based on the Asilomar Principles for the development of AI.
Peña emphasised that the team’s achievements have been underpinned by AWS cloud services.
“Those cloud services have been a huge enabler as they give us the power of spinning things up really quickly at low cost and looking at testing, trialling and evidencing what works and what can bring us some benefits,” she said.
“In all of these cases they went from the initial conversation to something on the ground within three months. Then we trialed them for a few months, evidenced weekly the results, then had decision making around whether we took them further forward, whether we had production and extensibility of the solution.”
AWS is helping to extend the availability of such solutions through the AWS Solutions Incubator. It supports a range of open source digital business solutions developed by customer organisations in government, decoupled from their original environments and packaged for easy adoption by others. It then hands them back for storing in the organisation’s GitHub repository and helps in promoting them.
The incubator also makes it possible for an organisation to take some components from a solution and use them for its own.
It provides a great asset for the public sector that will become even more valuable with the further development of machine learning and AI, encouraging innovation where it is needed, and the reuse of other’s solutions where it is more appropriate.