Democracy around the world faces fundamental challenges. Their citizens don't think they can achieve results. Recent research from countries, which are members of the Organization for Economic Cooperation and Development, shows a decline in trust in governments that have driven partially by perceptions that public institutions are neither responding nor transparent. Additionally, a series of Pew Research Center polls have clearly reduced satisfaction with democracy across 12 developed countries, including the United States.
To improve the perception of the US government within a subset of voters, the next Trump administration has pledged to curb waste and promote greater efficiency. It is unclear what that actually means, but the goal of providing better service to the government to people is not new, but it is important across democracy. With a new wave of government advisors, which many come from the space of “technological optimists,” it seems inevitable that technological advancements, including the deployment of artificial intelligence (AI), will become part of the proposed solution.
The ever-growing number of research highlights the benefits of using AI in the workplace. Examples of recent federal deployments of AI-enabled tools and other technology solutions show clear promise. For so-called “high-impact service providers” (open divisions of federal agencies such as the Internal Revenue Service and customs and border security), AI-supported performance improvements could improve Americans' perception of the overall capabilities of the U.S. government.
However, the “fast move, break” approach that uses technology to improve government efficiency can also have great results. With the presence of widespread mistrust in AI systems across the political spectrum and the presence of several well-documented failures around the world, whether or not the harmful deployment of these tools (by government) alienates citizens can lead to deeper skepticism of potential benefits or even delay the development of transformative solutions. It could also erode the public's trust in government warrants. Human interaction and judgment, as well as continuing attention to risk mitigation, remains important, despite the next administration exploring technology pathways to improve how governments can help citizens.
The Possibility of AI in Government
Recent experimental evidence highlights the potential of AI to transform the workforce. From university-educated professionals, programmers, aspiring lawyers to customer support agents, creative writers and consultants, AI support can improve production, save workers time, reduce performance disparities, and increase job satisfaction. These benefits have undoubtedly been transplanted into the public sector workforce around the world. For example, one study found that UK National Health Services experts estimated that they could save a day of work per day by leveraging generated AI to support more common and bureaucratic tasks.
This recognition of possibility has led to rapid experiments within the scope of US federal agencies despite ongoing challenges related to regulations and standards that conflict with data quality. In 2023, the federal government disclosed 710 AI use cases across the agency. By 2024, that number had more than 1,757 times higher.
For example, the US Patent and Trademark Office has deployed AI tools to improve patent classification and search processes and reduce application processing time. The State Department has helped employees leverage AI to help them use their time more efficiently, and deploy tools that can use open source and US government data to search for email drafting, document translation, brainstorming ideas, departmental policies, summarise articles and free up time for other tasks. Using a “crawl, walk and running” approach, Transportation Security Administration (TSA) has begun to integrate AI-enabled technologies into operations to speed up the airport screening process and improve customer service, but data security questions remain.
In impactful areas such as airport security, streamlining access can make tedious processes more efficient, strengthen the quality of everyday interactions among citizens, and ensure that benefits from AI are distributed more evenly. Over time, these technological investments can help improve government perceptions in theory.
The risk of excessive reliance on technology solutions
Despite these benefits, there is a grumpy history of technology that uses AI to improve decision-making and modernize operations. For example, in 2015, the Australian Robodebt scandal involving an automated debt collection system miscalculated social welfare beneficiaries' debt, causing serious economic and psychological distress. In 2019, childcare provided a scandal involving self-learning algorithms to identify early benefits frauds involving the Netherlands. The system, known as Syri, used indicators such as dual citizenship, low income, or “non-western appearance” as signals of potential fraud. The outcome was disastrous: children's separation from their families, poverty, and even suicide. A similar problem occurred in the UK visa application system. It claimed that algorithms designed to streamline visa processing contain “established racism,” leading to a halt in 2020.
These cases highlight the key role of data quality and system monitoring in deploying AI-enabled tools to modernize government functions. Without proper supervision, and if the underlying data is defective or biased, these issues will be replicated in subsequent output. This can lead to unintended negative consequences, especially in areas such as law enforcement and social services. Faulty AI applications can pose a question about other viable, technology-based solutions and completely undermine trust in government.
It is also possible to overestimate the importance of “efficiency” as a factor of satisfaction. For example, while the US has seen increased awareness of airport security experiences, a survey of over 13,000 TSA customers found that this shift was not related to perceptions of waiting time. This has found that facial recognition technology can help improve “interpersonal communication” such as “understanding professionalism, respect, and security procedures.” In short, friendly TSA agents can make a greater difference to the traveler's experience than a quicker wait in line. In this case, technology can certainly play a role in shaping positive experiences, but it may not be the main factor that changes perception.
Balancing efficiency and trust
The Trump administration's previous AI executive order mandated the incomplete but useful enforcement inventories for AI use cases, but it recognized that unimplemented systems could outlaw technological solutions deployed to the federal government. But as the second Trump administration takes shape, push to dismantle guidelines designed to mitigate risks or streamline the procurement process could do more harm than good. Without a doubt, many of these areas could benefit from improvements (such as agency-wide data standardization practices and FedRamp modernization efforts). However, an approach to inability to properly identify, describe, or communicate the potential risks of these systems, or to move quickly, identify and communicate properly can backfire.
Recent polling data demonstrates the need for a progressive and intentional process. With increasing awareness of AI, there are concerns about its misuse. In the US, the majority of respondents support higher testing and safety standards for a variety of AI-enabled systems, but there are variations between parties as to whether this should be self-regulation or enforceable government oversight.
To address the growing skepticism of Americans about AI, it is important to focus on reducing risk before deploying technology solutions. Clear guidelines for AI use need to be developed, and methods for humans to clarify the decisions and reasons for AI systems, engage the public, and encourage independent surveillance. They also need to make it easier for people to opt out of technology if necessary. In tandem, policymakers need to streamline and dismantle bureaucratic processes regarding access and procurement of data that overly complex or burden the deployment of promising AI solutions.
It is becoming increasingly clear that collaboration with humans can improve productivity and quality. However, without public buy-in, it is difficult to build trust in these systems. Despite the major push by the federal government, AI continues to be shaped primarily by private sector investment and market forces. Technological innovation can undoubtedly improve government efficiency. However, when governments leverage the potential for AI transformation, transparency must prioritize, mitigate risk and maintain a critical role in human decision-making.
