The role of AI in helping democratic governments maintain public trust
Democracy is enabled through Government transparency. A core element of such transparency is the systematic release of historical Government records in accordance with the Freedom of Information Act (FOIA) and Public Records Act (PRA). The release of Government records carries sensitivity and security concerns, alongside the resource-heavy nature of reviewing material prior to release. Such review demand has increased geometrically since the introduction of IT systems which enabled digital record generation – with the risk of being so expensive that many departments may consider themselves challenged in compliance with their statutory obligations, which threatens the quality and availability of information in The National Archives (TNA) and risks diluting public trust.
Every UK Government department is obliged to transfer relevant public records to TNA, delivering on their obligations for transparency and public accountability. With the immense volume of digital files across government departments, the task of a Digital Sensitivity Review, where sensitive information is identified and redacted before transfer to TNA, presents serious challenges and a risk of human error.
The sheer number of digital files selected for long term preservation and taken through sensitivity review is vast and requires innovative solutions, both to ensure the records are processed securely and to achieve the greatest possible release to the public in a timely and cost-effective manner.
Alongside digital record systems, Government departments require new technology that helps Sensitivity Reviewers work more quickly, reduces risk, and which is robust enough to meet rigorous Government security protocols, as well as being easy to use. The problem was, this technology didn’t exist – but it does now.
Working with the University of Glasgow through a Knowledge Transfer Partnership (KTP) and a range of expert suppliers, the team from SVGC developed a digital system incorporating a combination of rule-based and AI powered methods for Government departments that accelerates review processes, and reduces risk by surfacing areas of potential sensitivity in accordance with the FOIA and PRA.
Our team of technology engineers are demonstrating the use of AI to accelerate and assist human decision-making on sensitivities through the novel application of cutting-edge Natural Language Processing techniques, such as Transformer based Deep Learning models. This work also uses other machine learning processes to aid in structuring and de-duplicating large volumes of public records and improve the efficiency and effectiveness of the review.
The KTP with the University of Glasgow enables us to collaborate on the development of new technologies and procedures using state of the art hardware and software, based on methods which have been extensively peer reviewed, tested and piloted before deployment.
This use of cutting-edge software and solution technologies helps us to alleviate pressures on highly skilled Sensitivity Reviewers by accelerating the review processes through accurate identification and management of potential sensitivities. The reviewers have been able to use our intuitive solutions with ease, acknowledging that the introduction of semi automation has reduced human effort in those areas where it makes most sense to do so.
We continue to make significant progress on other applications to enhance core capability, providing relevant insights, reducing the time required to analyse material. Our modular approach builds in the ability to take advantage of emerging technologies to continue leading the way as the challenge evolves with new file-types and document complexities.
Intrigued to know how AI technologies could help your business processes? Get in touch with one of our experts today.