What does AI mean for government?

Yasmeen Ahmad, director of Think Big Analytics, gives lessons from commercial enterprise

The largest global commercial companies are under pressure: they face a new world where digital disruption has changed consumer expectations forever. ‘Uberisation’ is occurring in all industries, as new business models no longer require physical products and traditional services. Instead, organisations rise from the monetisation of data through analytics; as a result, consumers are increasingly demanding better and more transparent services that meet their needs in real-time.

The question arises: how do governments benefit from new tech trends, leverage the vast amounts of data they collect and apply artificial intelligence (AI), machine learning and advanced analytics to create value? These terms are not new: they describe algorithms designed to find patterns, trends, and predictions in data. Arguably, governments should be at the forefront with these techniques, given the volumes of data across departments available for interrogation.

AI has proven to support multiple areas of government, allowing governments to understand citizen needs, improving citizen services, enhancing citizen security and identifying tax and fraud risk. Here’s how:

Decoding digitalisation to improve customer journeys
With 80 per cent of the UK population online regularly in 2017, we have seen the digital channel rise as a major interaction media for users of today. There are over 70 digital government service portals that enable citizens to engage. However, provisioning of these online portals is not enough; citizens demand an optimal experience to make activities smoother, less complex and faster to complete.

Although the days of never-ending paperwork may be behind us, online processes can be complex, for example registering for VAT services and self-assessment tax returns. The opportunity online, as opposed to previous paper-based methods, is to provide additional context information, real-time assistance with online chat, immediate feedback on incorrect form entries, as well as re-routing to the right services and digital estate based on a prediction of citizen intention.

These actions require the analysis of citizens’ online journeys to identify intentions, areas of difficulty or complexity, leakage from the digital estate and abandonment. Data available today, for example, digital clickstream data, can provide the means for such analyses to occur.

Furthermore, the opportunity of online services is the ability to watch and analyse citizen behaviours for both positive and negative intentions. By analysing customer journeys and implementing real-time capability to digest citizen actions, we can observe those who are ‘gaming’ online application forms. For example, as individuals change the values on their annual tax return, algorithms begin to assess the risk of fraud and create alerting to help with prevention processes.

Beyond this, we can leverage data connected across digital platforms with offline journeys to assess the root causes of why citizens abandon online activities such as applications. By improving the uptake of digital services, the government can benefit from a channel by which it is easier to serve high volumes of users, alleviating pressure from offline systems.

Understanding the voice of the citizen
Increasing digitalisation results in higher levels of data capture of everyday actions that would have gone undocumented previously. Digitalisation opens the doors to capture every conversation, comment, and piece of feedback submitted in the call centre, online, via social media channels or online chat aides. This captured voice can now be digitised to words and sentences – then made available to algorithms to decode.

Year-on-year we observe call volumes increasing, and citizens asking for more assistance and in real-time. The rate of change in citizen communities and complexity of personal and financial situations leaves us to engage with more government services than ever before. By decoding citizen voice using natural language processing and text mining techniques, we can identify key themes and topics of conversation, as well as analyse sentiment of citizen words and, importantly, the volumes of individuals affected.

This insight provides an ideal backdrop to mitigate calls by understanding key challenges and issues to then take relevant action, for example improving a digital form, providing clarity of instruction or more information during the application process or introducing new services that meet needs.

Within a call centre, analysis can be used to automatically route calls based on a prediction of citizen call reason. In addition, analysing call centre staff responses can lead to increased efficiency of call centre operations by identifying training requirements for staff.

Identifying risk and fraud
With an increasingly global data economy emerging, the complexity of business operations has increased exponentially. The volume of transactions, mobile payments, b2b connections and movement of funds between parent and subsidiaries is masking complex tax evasion schemes and other types of fraud. To understand this complexity, graph analytic techniques are supporting government departments in improving compliance, increasing revenue and collection growth.

Government audits previously focused on investigating and cross-checking between audited company and counterparts. However, when it came to secondary, tertiary or further connections it quickly became difficult to determine who might have been involved in the network of fraud. Furthermore, this effort spanned many manual hours to identify VAT evasion schemes, money laundering and much more.

By implementing robust analytical systems, we can enable automatic detection of inconsistencies such as non-filers, under-payers, return fraudsters and double invoices. Millions of electronic declarations can be analysed to locate inconsistencies and provide automated notices to citizens to take action or forwarded for further investigation or action by authorities.

A connected data landscape between departments also allows for high-risk companies to be identified faster with failings in one department used to increase awareness of potential failures in other departments. For example, failed audits by the healthcare authority often raise the probability of issues with the consumer supervision authority.

Empowering government departments everywhere
The power of data and advanced analytics, when used to create value across domains and government departments, is evident. The examples explored demonstrate the use of a wide range of data sources combined with a variety of analytical techniques, for example, path and pattern analysis to decode digital and multi-channel journeys, natural language processing to understand the voice of the customer and applying graph analytics to understand networks of risk and fraud.

Furthermore, the examples not only demonstrate a benefit for the citizen but in most cases also help government departments with cost savings, revenue generation and mitigation of risk and fraud. In an increasingly digital world, these use cases are not optional but essential to supporting business economies and citizen communities.

Please register to comment on this article