How might artificial intelligence affect the trustworthiness of public service delivery?

The first Long-term Insights Briefing report.

Summary

This report, on How artificial intelligence might affect the trustworthiness of public service delivery, is the first in an enduring series of public service-led Long-term Insights Briefings.

The Long-term Insights Briefings are a new initiative under the Australian Public Service (APS) Reform Agenda. They provide an opportunity for the APS to consider significant, cross-cutting and complex policy issues and how they may affect Australia and the Australian community in the medium and long term. The briefings will bring together, and help the APS to deeply understand the evidence (including the views of the Australian community, academia and industry), context, trends and implications of complex policy issues. This will build the capability and institutional knowledge of the APS for long-term thinking, and position the APS to support the public interest now and into the future, by understanding the long-term impacts of what it does.

These briefings are expected to form part of the evidence base for policy making over time. They are not intended to take the place of policy making across the public service and do not make recommendations for future action.

Artificial intelligence offers significant opportunities to improve public service delivery

This Long-term Insights Briefing explores how using artificial intelligence (AI) to deliver public services might affect the trustworthiness of public service delivery, now and in the future.

We will need to do more with less to meet community expectations of public services in the future

Innovation, including adopting digital and AI technologies, can provide opportunities for efficiency gains that will allow governments to do more with less and meet the needs and expectations of the Australian people. The community’s expectations around the quality of public services are growing: for a higher standard of care; for tailored and personalised services; and for greater responsiveness, convenience and efficiency when accessing services. Australia’s population is ageing, increasing demand for care and support services. At the same time, an increase in the share of older Australians in the population means fewer working-age Australians to help fund public services.[1] External forces, such as climate change, are also expected to increase demand for services while decreasing the resources (people and funding) available to provide them.

Artificial intelligence could transform public service delivery, leading to a better experience and outcomes for the whole community

Opportunities exist to use AI across the spectrum of activities carried out by the APS. These include:

  • automating ‘backroom’ administrative processes
  • improving efficiency and minimising errors in data management
  • improving service processing and response times, freeing up time for more creative and complex work
  • AI-enabled public interfaces that offer customised services to users and facilitate communication with diverse populations.

Indeed, AI capabilities are already being used in solutions in some Commonwealth government agencies, including:

  • chatbots, virtual assistants and agents in service delivery
  • document and image detection and recognition for border control and for fraud detection
  • data mapping to geographical areas.[2]

In the future, AI will be a critical tool for maintaining and delivering even better quality services to a growing and ageing population.

Using artificial intelligence in public service delivery is not without risk

Using AI for public service delivery is not risk free. AI systems can inherit biases present in their training data, leading to unfair or discriminatory outcomes. The collection and analysis of personal data for AI can have privacy and security risks. Complex AI systems might behave unpredictably, causing unintended outcomes at a scale and speed that are hard to control.

Although not an example of AI, the Royal Commission into the Robodebt Scheme examined an automated debt raising and recovery scheme for social security payments. In the Robodebt example, the underlying basis for debt calculation was flawed and resulted in many incorrect calculations. The scheme was then implemented quickly, and incorrectly calculated debts were raised and imposed on recipients at scale, thereby compounding the impact on the community. In her report, Commissioner Holmes AC SC concluded that When done well, AI and automation can enable government to provide services in a way that is “faster, cheaper, quicker and more accessible.” Automated systems can provide improved consistency, accuracy and transparency of administrative decision-making. The concept of “when done well” is what government must grapple with as increasingly powerful technology becomes more ubiquitous.’ This briefing and its insights, on How artificial intelligence might affect the trustworthiness of public service delivery, are not a response to Robodebt. However, it offers an important framework for agencies considering how to adopt and implement AI – namely that realising the benefits of AI in public service delivery will require agencies to identify and mitigate these (and other) risks in order to ensure that AI is used in a safe and responsible manner.

Stewardship of artificial intelligence in public service delivery

In just five years, AI and its capabilities have developed rapidly – from experimental applications to AI solutions and applications that are widely adopted and used across society – fuelled by an increase in computing power, investment and consumer demand. This rapid rate of change is expected to continue in the future.

This will pose a challenge for an APS seeking to realise the benefits of AI for public service delivery. A survey undertaken for this Long-term Insights Briefing found that people who are more familiar with and knowledgeable about AI have higher trust in government’s ability to use AI for public service delivery. However, more than half of people (57%) have zero or slight knowledge of AI, while almost two thirds (63%) have zero or slight understanding of when AI is being used.[3] Most people will gain a greater knowledge of AI and when it is being used over time, thanks to the increasing pervasiveness of AI technology in their daily lives. However, the APS will always be engaging with people who lack knowledge of and familiarity with the latest tools. Realising the benefits of AI will require the APS to steward the community through the transformations that AI will bring to how public services are designed, implemented, delivered and explained. This stewardship is necessary to ensure that AI contributes to the delivery of high quality public services and that the risks outlined above are well managed.

Framework for trustworthy use of AI in public service delivery

To build trustworthiness, government agencies need to deliver public services well, by meeting users’ needs and delivering efficient, timely, good quality and reliable services. There are significant opportunities for agencies to use AI to do this. Drawing on research, survey evidence, and consultations with community representatives and AI and service delivery experts, this briefing offers insights into how agencies can design, develop and implement AI in public service delivery in ways that build trustworthiness (Figure 1).

Figure 1. Framework for trustworthy use of AI in public service delivery

Trustworthiness of public service delivery is built when: 1. Integrity is established through: Regulation and processes to protect personal data; Clearly communicating to the community about their rights and protections; Frameworks to ensure ethical, fair, accountable and transparent use of AI. 2. Empathy is demonstrated by: Offering face-to-face service delivery, especially for people experiencing vulnerabilities and those with complex needs; Taking into account individuals' needs, contexts and experiences when making decisions. 3. Performance is improved by: Using Al to better meet users' needs and deliver efficient, timely, quality and reliable services; Upskilling frontline agency staff so that they can clearly explain Al outcomes to end-users. 4. Competence is built by: Scalable and reliable technology infrastructure to support Al solutions; Investing in the skills to steward the community through the transformations to public service delivery that Al will bring about. Trustworthiness of public service delivery is eroded if: 1. Integrity is undermined by: Security and privacy breaches; Failing to communicate how individual's personal data and information is being used; Failing to establish lines of accountability and avenues to appeal Al outcomes. 2. Empathy is lost when: Agencies fail to offer enough of a relationship to service users; End users' experience fake empathy in an Al-facilitated interaction. 3. Performance is reduced when: Agencies fail to address unintentional biases and stereotypes perpetuated by Al; Agencies fail to accommodate the digital experience, connectivity to Al knowledge of the community that the agency serves; Artificial intelligence makes it harder for people to access and engage with public services. 4. Competence is undermined by: Lack of workforce skills and system capability to develop. use and implement Al; Failing to train Al models on high quality and representative data; Outcomes that are biased or perceived to be unfair.

Source: Summary of insights from workshops and focus groups held for this briefing, involving people from 15 organisations representing the community, 9 organisations representing academia, industry and youth, and 16 APS Agencies.

Insight 1: Artificial Intelligence must be designed and implemented with integrity

If the community does not trust AI, and the APS still uses it within a service offering, the APS may itself be seen as untrustworthy. Implementing AI in public service delivery well, in ways that demonstrate and build trustworthiness, critically depends on establishing and acting with integrity. This means that the people and organisations employing AI are accountable for the outcomes of the AI that they use, and transparent about how AI is being used; practising ethical values and principles when designing, developing and implementing AI; and ensuring personal privacy and data security.

At the same time, AI regulation and frameworks will only build trustworthiness if they are clearly communicated and explained to the community, including how they work in practice and the protections they provide.

Insight 2: Using artificial intelligence shouldn’t come at the expense of empathy

AI will increase the trustworthiness of public services if it is designed and implemented in a way that demonstrates empathy. Trustworthiness is built when the APS demonstrates empathy for the people it serves. In practice, this means providing enough of a relationship with public services, where what that looks like depends on an agency’s trust history, the community it serves, and the type of service it offers. A relationship with frontline staff is likely to be particularly important for people experiencing greater vulnerabilities and those with more complex needs.

Insight 3: Artificial intelligence should improve performance

AI will significantly erode trustworthiness if its use reduces the performance of public services.

Using AI in public service delivery will erode trustworthiness if end users have a poor experience with the service they are seeking. This might be the case if AI perpetuates unintentional biases and stereotypes (as a result of being trained on biased datasets or failing to include diverse perspectives in the design process); or makes it harder for people to access and engage with the services. New skills and capabilities will be needed across the APS if it is to adopt and use AI in ways that improve public services, and steward the community through the transformations to public service delivery that AI will bring about.

Insight 4: Successful service delivery depends on supporting people to engage with AI-enabled services in the long term

Maintaining trustworthiness requires the APS to deliver services to the whole community. Public services are for everyone, including those who don’t want to engage with digital and AI-enabled systems or provide additional personal data, to ensure that using AI in public service delivery doesn’t entrench disadvantage. Nevertheless, in the long term, opting-out will not be an option in a more connected world, where AI will be critical to address future challenges. It will be important to invest in building the AI literacy and digital connectivity of the community, particularly cohorts experiencing vulnerability and those that support them, in order to bring everyone along on the AI adoption process.