The state of AI in government: A mixed bag
The study revealed that some Asia–Pacific government agencies are using AI technologies to streamline operational processes, improve service delivery and increase overall compliance with public programs. These agencies are able to increase efficiency in areas like service provision and public security (think police, customs and immigration). These agencies are finding ways to optimize their scarce human and financial resources while better fulfilling their mission.
However, these desirable outcomes are not universal and usually depend on the level of AI maturity within each agency. For instance, only a quarter of government organizations in the Asia–Pacific region report that they have integrated AI or are using it in some transformative way. The biggest group, 29%, is using AI in a narrow way. A separate question indicated that the top five applications are aimed at addressing issues in social welfare, emergency response, tax and revenue issues and investigations.
These responses are not surprising to me. The benefits of AI in such applications, such as reducing inefficiencies and helping with sense-making, are easily argued and demonstrated, making it relatively easier for agencies to secure the necessary support and funds.
What’s working with AI
Respondents were also asked what aspects of AI they considered the most beneficial. While automatically managing and deploying models was the top answer, I suspect that would only be relevant for agencies that are quite mature in using AI – there’s no need to worry about managing models if you only have one or two on hand.
The three other benefits were more relevant to many agencies just starting out as they are much closer to everyday work. One benefit in particular to note is number two: streamlined agile decision-making process. I think it’s important to remember that this benefit will not be realized unless the agency sits down and does the hard work of clarifying its decision-making process. For example, it would be difficult for an agency to become a data-driven organization if decision-makers continue to be swayed by emotional arguments.
Challenges and benefits
However, there were also clear challenges to implementation. A key and perennial one was siloed data. Nearly a third of government organizations said that there were bottlenecks around data storage and a lack of standardized data management practices, making it difficult to draw on data from multiple sources.
It seems there is no easy answer to this problem. The technology to fuse data from different sources has existed for some time, but as respondents suggest, hurdles like legacy siloed systems, unclear data ownership, a desire for more security and lack of funding will need to be overcome.
Organizations also cited a lack of skilled staff as a key challenge. This shortage hampered not only the implementation of AI or data acquisitions but also an inability (or lack of confidence) to evaluate AI-based solutions. However, this might also be specific to the government because it is often hard to assess the return on investment in the public sector.
The future of AI in government
How can government organizations overcome these hurdles and realize AI’s promise for citizens and governments alike? The first step is to improve data management and governance.
Much government data is sensitive and needs to be securely managed. In the past, keeping data in siloes was desirable from a security and privacy standpoint. But that traditional approach is now hamstringing the success of AI projects in government. Governments must, therefore, start thinking of ways to cut this Gordian knot and make these disparate data sources readily and easily integrated while being secure enough.
Once that foundation has been laid, there are several ways to support agencies’ missions. At the root however, is that agencies need to be able to envision a pathway starting from using AI in narrow applications to eventually seeing the value of allowing all business units to access AI for their needs. To do that, agencies need to see AI systems as relevant and reliable – such as being able to explain how the AI models came to their results. This, in turn, means that agencies need to increase the general and specific AI skill levels and experience using AI of their officers and leaders.
Ultimately, the overall message for Asia-Pacific government agencies is that for AI to make a difference, they need to address data issues, have skilled people and have a vision for adopting AI both in the short and long term.
Discover comprehensive insights into data and AI success in government – download a copy of Data and AI Pulse: Asia Pacific, 2024.
Â