Every organisation has different levels of experience with emergent technology like AI. Some barely use it while others have robust governance. Recently, a member of Apolitical's AI in Government Community leaned into the experience of their peers by asking:

“Could anyone share (a) link or experience with policies for responsible and ethical use of AI in civil service and government? Do you have any strategy, policy or guidelines across civil service?”

This member's question sparked not only a crowdsourced library of policies but a discussion around AI ethics. This article highlights the key themes and resources mentioned in the resulting conversation.

The ethics of inaction: is non-adoption a risk?

A central point of debate involves the 'ethics of non-adoption.' Most traditional frameworks focus exclusively on the risks associated with deploying AI, such as bias or privacy breaches. However, community members argued that government officials have a mandate to provide the best possible public services. If AI can significantly reduce administrative burdens for clinicians, thereby increasing time for patient care, choosing not to use AI carries its own ethical weight. This perspective shifts the conversation from "how do we stop AI from doing harm?" to "what harm is caused by maintaining inefficient systems?" For example, if automated triage can process routine applications faster, allowing humans to focus on complex cases, then a delay in adoption represents a missed opportunity for public benefit. The consensus in the group was that staying still is not a neutral act. Instead, the risks of inaction must be weighed against the risks of implementation.

Trust and the familiarity curve

The discussion drew parallels between current skepticism toward AI and the early days of online payments or personal computers. Participants recalled a time when digital banking was met with significant fear regarding identity theft and financial loss. While those risks still exist, the technology became ubiquitous as public knowledge increased and risk management matured. Members noted that AI is likely following a similar trajectory, though the potential for unintended consequences is broader. Public trust is the most fragile element in this curve. As seen in recent reports of local governments using generative tools for document drafting, using technology to gain efficiency can sometimes erode public confidence if not handled transparently. A degree of caution is a functional part of the process, allowing public servants to build "civic competence" before these tools become standard infrastructure.

From theory to operations: practical safeguards

The conversation moved from abstract ideas to the specific mechanisms used to safeguard public interest. Immigration, Refugees and Citizenship Canada (IRCC)'s approach, for instance, utilises machine learning to triage applications while maintaining strict human-in-the-loop protocols, ensuring that final decisions are never left to an algorithm alone.

Key operational tools mentioned include:

  • Algorithmic Impact Assessments (AIA): Mandatory reviews that evaluate the potential social impact of a tool before it is deployed.
  • Bias testing: Regular audits to ensure that automated systems do not produce discriminatory outcomes.
  • Public reporting: Transparency regarding error rates and system performance to maintain accountability.

Takeaways for government

The goal is not to adopt AI for its own sake, but to use it as a tool for public good. Progress relies on moving away from "non-approaches" and toward measured, proactive strategies. This involves implementing robust safeguards and conducting thorough impact assessments. By focusing on solving the right problems rather than just applying the latest technology, agencies can ensure that efficiency gains do not come at the cost of public trust.

Resources shared by community members

Global & Research Institutions

Australia

Canada & United States

New Zealand & United Kingdom

Join the conversation

To participate in conversations like this, join the AI in Government Community.

Join the community

Contribute to the Government AI Navigator

Instead of crafting a governance framework for AI, have you applied AI to government work? Contribute your case study with AI applications and outcomes to the Government AI Navigator. The Government AI Navigator is an AI-powered, interactive database of real-world government AI initiatives from around the globe.

Contribute your AI case study

This article was written with the help of AI using anonymised conversation text and edited by a human before publishing.