The promise of artificial intelligence is no longer a nebulous concept on the horizon. The private sector is already exploiting its possibilities, and governments are following suit. We use the broad term “AI” to refer to computers simulating human abilities and performing tasks that people typically do, using technologies such as cognitive computing, predictive analytics, robotic process automation and machine learning.
AI represents a significant economic opportunity for the United Kingdom. Recent research by IBM and the Confederation of British Industry found that around 20 per cent of British firms have already deployed AI in practical applications. In March, Emmanuel Macron laid out his plan for French leadership in AI, which included a pledge to invest €1.5bn of public funds. Since then, 25 European countries have signed a Declaration of Co-operation on the most important issues raised by AI. Most recently, the House of Lords has published its view on how the UK can be best placed to take advantage of AI.
Used in government, AI can also stand for “augmented intelligence” – giving employees the ability to make more informed decisions.
The earliest successes of AI adoption in government have happened in the US. Citizenship and Immigration Services now uses an AI-based online virtual assistant to answer questions from citizens and immigrants, while the Intelligence Advanced Research Projects Activity has tapped the technology to improve facial recognition. Other agencies are testing whether AI can improve purchasing processes or relieve employees of tedious work. A recent survey found that 77 per cent of US Federal managers said their agencies will need to deploy AI over the next five years “to keep up with the increasing pace of work”.
As government agencies around the world start to adopt AI, a number of repeatable use cases have started to emerge. The overwhelming majority of the world’s data – an estimated 80 per cent – is held in formats not easily used prior to the emergence of AI. This data may be held as unstructured documents, both electronic and hard copy, as video or audio. AI can analyse these files, recognising the content of images, videos and text, and then help people to understand them and use them to make informed decisions.
Machine learning systems improve over time, building better simulations with the more data they are given. Alexander Measure, economist at the Bureau of Labor Statistics in the US, called the possible applications of this technology in government “too many to list,” but the wide range of activities, he says, can include everything “from analysing satellite images to processing Social Security disability claims.”
How governments are using AI
Across public sector organisations, a number of very different agencies – both defence and civilian, and in local government and academia – are using AI technology to transform their work.
The cases presented here describe both works in progress and end results. Other agencies can learn and benefit from these organisations’ early experiences, particularly if these first stages end up being a springboard to significant shifts in agency practices.
…for a safer world
Law enforcement agencies use, or plan to use, AI to outsmart criminals. Several US law enforcement agencies, as well as wildlife rangers in two other countries, have used AI to plan their patrol routes to guard against terrorism and poaching. In addition, recent advances in video analytics that can interpret facial sentiment and bodily disposition are adding enhanced security capabilities to border agencies.
…for improved service
Agencies such as the US Bureau of Labor Statistics at the Labor Department are looking at AI to assist staff through taking away tedious, repetitive tasks from employees and save hundreds, even thousands of work hours. Employee time can then be redirected to more important tasks. Similar organisations are looking at AI in speeding the assessment of benefits eligibility, resulting in faster citizen satisfaction and ensuring that incidences of missed or improper payments are reduced.
Tax and customs agencies are looking to AI to identify instances of tax evasion, by the analysis and interpretation of many different unstructured sources of data.
…for social benefit
AI can also help break down government silos, analysing disparate data sources from different agencies to serve citizens and protect vulnerable populations. Child welfare agencies are using AI to help identify children at risk, and to recommend interventions to prevent abuse. One Kansas county is using the technology to identify substance abusers at risk of arrest, and to get people in these at-risk groups into county services before they break the law.
…to reduce facilities cost
Government agencies with large land estates, along with many private-sector counterparts, are starting to use AI in asset and facilities management to predict failures of critical equipment and maintain them before they fail. For defence and intelligence agencies, too, predictive maintenance can help ensure mission readiness, safety and agility of response.
The education community is looking to AI to help students find out about services. A leading university is using AI technology to better provide students with access to information on
benefits and where to get help from social services.
…for better value for money
The US Air Force plans to use AI technology to make sense of complex acquisition regulations, so that it can speed the process of buying goods and services. Doing so could open government procurement to more small businesses and companies that have previously avoided working with government agencies because the acquisitions process has been too difficult to navigate.
…for policy decisions and legislative guidance
The ability of AI to navigate mountains of existing legislation, rules, regulations, ordinances, and to determine the impact of change in legislation on other parts of government, has increasing relevance in a world where the public demands clear action in complex cases. This ability to analyse unstructured data, in conjunction with prevailing legal frameworks, can have equal applicability in determining if a legal case will stand effective prosecution.
…for happy citizens
AI already plays an increasing role in everyday life – citizens now interact with chatbots as a matter of course. In the public sector, this can be effective in both citizen helpdesks and IT call centres, providing the user with access to insights and expertise derived from the most experienced and knowledgeable employees, and delivering that advice in a way that is easy to understand. In Australia, AI is being used in conjunction with an avatar of a real person to enable citizens with physical and cognitive disabilities to access key benefit services.
Trust remains a key factor
AI is progressing well, with more solutions on the way to help people find smarter ways to live and work in a world where big data is a new natural resource.
But to fully reap the benefits of AI, society must trust it. This trust must be earned through transparency, and through repeated experience. In the same way that people learned to trust that an ATM will register a deposit or that a car will stop when the brake is applied, AI must prove itself a reliable and useful technology.
Government has a role to play here, to ensure that use of AI goes hand-in-hand with the principles that ensure trust in both the technology itself and in the handling of sensitive citizen data. AI systems must be accountable, and developed with the capability to explain their decisions. With responsible handling, AI can be a powerful tool.
Ian Baker is partner and vice-president at IBM Global Government Industry.