This article is written by Dr. Keegan McBride, Lecturer in AI, Government, and Policy, Oxford Internet Institute.

It has been commissioned and published as part of the Government AI Campus — an initiative by Apolitical to prepare public servants and policymakers to lead in the age of AI.

Key takeaways:

  1. Governments have always sought to collect, manage and use data and information; AI is just the next step in this process.
  2. AI should be viewed as part of a broader digitalisation strategy focused on driving innovation and offering new services.
  3. The successful implementation of AI-based solutions requires strong regulatory, technological and organisational foundations.
  4. The adoption of AI and other digital technologies is transforming the public sector, making it increasingly efficient and effective but less human and, in turn, more distant.

For AI to be implemented successfully, there is a need for strong supportive regulatory, technological and organisational foundations.

Failing to adopt a more holistic or systemic view of AI will lead to suboptimal outcomes.

With the launch of OpenAI’s ChatGPT-3 in November 2022, Artificial Intelligence (AI) became tangible for the first time. AI was no longer something of the future, a tool only for scientists – anyone could use it. As the world began experimenting with using AI to write emails, engage in conversation, or generate art, governments took notice and asked how they could best leverage AI within their bureaucratic structures.

Though governments have already been using AI for years (Estonia, for example, boasts more than 100 AI use cases across the entirety of their public sector), the recent hype cycle has led to the launch of numerous conferences, task forces, working papers and news headlines proclaiming that AI will transform or revolutionise some aspects of the public sector. 

💬 Let the author know what you think about AI in governments by leaving a comment below

While it is certainly the case that AI will create value and transform some aspects of public sector bureaucratic organisations, most conversations in today’s discourse fail to explore two things:

  • First, the necessary foundations for making AI work in the public sector.
  • Second, the broader implications that accompany an increasingly datafied and AI-driven public sector.

Let’s start with the foundations: what exactly is AI? Truthfully, what “is” or “is not” AI is still debated today, and the benchmark is always moving. While there is no definitive definition for AI today, there is growing consensus surrounding the definition proposed by the OECD:

AI is a “machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments.”

Put another way, today’s AI-based systems are statistical models that are “taught” through their exposure to data. Successful implementation of AI-based systems is dependent on the availability of data. This implies not only that data are being collected but that they are being stored and managed in such a way that AI-based systems can access them.

It is not possible to discuss AI on its own as if it were possible for the public sector to “just use AI” to fix a specific problem. Instead, AI should be viewed as one component within a broader digitalisation strategy. For AI to be implemented successfully, there is a need for strong supportive regulatory, technological and organisational foundations.

Failing to adopt a more holistic or systemic view of AI will lead to suboptimal outcomes.

This is easier said than done, especially in the public sector context. For example, regulatory restrictions may inhibit the collection or exchange of certain types of data. Technologically, if the necessary data infrastructure or data exchange systems are missing, it is harder to develop and implement AI. Organisationally, if the public sector is missing capacity, there may be a need to update the required skillsets for public servants, such as increasing knowledge of statistics, math, or technology. 

Ultimately, a public sector that can overcome these barriers and digitalise successfully represents a specific future vision of the public sector. It is a future where the public sector is able to collect and use large amounts of data, run by public servants who have strong technological and data skills and a regulatory environment that supports and prioritises the digital. For some, such a future is an intimidating one, but it need not be.

In fact, this transformation is not new; it has been unfolding for hundreds of years. States and the bureaucratic structures that run them have always been interested in the collection and use of data. Viewed from this perspective, AI is the continuation of something old, namely, rationalisation and the use of technology to improve the efficiency of the public sector. “Improve” in this context is associated with a growth in control, power and surveillance. Though such terms often evoke strong negative reactions, the truth is that they form the foundations of our governments today.

The provision of welfare by governments demonstrates the necessity for control and surveillance. If you want to provide benefits, you need to have information about those who are going to receive them. For example, if you wish to provide unemployment benefits, it is essential to know if someone is employed or not. Without the availability of data, this is a process that grows in complexity and risk. In such a situation, some governments may turn to AI to improve the provision of these services.

This is exactly what the Netherlands’ government did recently, attempting to use AI to identify cases of benefit fraud. Unfortunately, the system went wrong and was found to be illegal and discriminatory, leading to a fine of several million euros. There are numerous examples of governments getting AI wrong. This should not distract from experimenting or implementing AI-based solutions but rather highlight the importance of getting AI right and investing in the foundations.

If governments invest in the foundations and begin to implement AI across the public sector, numerous opportunities emerge. AI can be used to automate processes, augment decision-making processes, improve public safety, transform public policymaking and deliberation, enable higher levels of accessibility and reimagine digital public services. If developed and supported properly, these innovations lead to increased effectiveness of the public sector, cost savings, increases in citizen satisfaction and decreases in red tape, amongst numerous other benefits. 

Driven by the desire to leverage digital technologies, the public sector will be required to collect increasingly large amounts of digital data. The collection of such data is essential for the successful functioning of a digitalised public sector, and it will enable numerous advantages. It also represents a movement towards a centralised future defined by data and statistics. This transformation represents not only a change in the public sector and how it operates but also in how citizens experience and interact with it.

A digitalised and AI-driven public sector will bring it closer to citizens than ever before, with services being delivered digitally and proactively. At the same time, decreased in-person contact with the public sector, separating further as digitalisation becomes increasingly pervasive, will lead to a system that is increasingly distant. These transformations will create value and positive impact, but they also represent a serious and fundamental shift in how we understand the state, government and society.

Done reading? Make sure to share your own tips by leaving a comment below ⬇️

(Image credit: Pexels)