Why Open Source Is Even More Important in the Age of AI

by Mike Gifford, CivicActions

Artificial intelligence has changed how we think about digital services. For the first time, we have tools that can generate large volumes of text, code, and content faster than any human team could ever manage. That speed is extraordinary, but it is also destabilizing. Between the hype and what people are observing, much in our society is about to change. In the public sector, where technology underpins essential services and policy delivery, we need to ask a hard question: What keeps our systems safe, resilient, and aligned with the public interest when AI is producing so much, so quickly?

The answer is not “more AI.” The answer is stronger governance, shared oversight, and community-driven stewardship. In short, the principles and practices that open source communities have developed over decades.

Code Volume Does Not Equal Progress

AI can now generate thousands of lines of code in seconds. But sheer volume is not a measure of progress. In fact, treating it as such is dangerous. What matters is not how much code is written, but how new ideas are added to an existing project in ways that don’t compromise either the current implementation or future opportunities.

This distinction is crucial for governments, where digital systems often underpin healthcare, benefits delivery, taxation, immigration, and justice. A single misaligned feature can cascade into service failures with real consequences for citizens. We generally use technology to amplify someone’s power, but both the speed and the “confidence” of modern LLMs can be quite convincing. Producing more code faster is meaningless if it undermines the reliability and trustworthiness of the service.

Open source communities have always understood this. These communities have always been run literally by humans with real human experiences and with a real investment of human effort. These communities measure progress not by output of code alone, but by the quality of integration, the clarity of direction, and the ability to maintain functionality over time. In the AI era, this discipline is no longer optional. It is the only way to prevent systems from collapsing under the weight of unchecked change.

Governance and Process as Safeguards

Governance is often dismissed as bureaucracy. In reality, it is a safeguard, especially in a world where AI accelerates complexity. Automated tests are useful, but they can only confirm that code behaves as expected in narrowly defined ways. They cannot confirm that a new feature aligns with public needs, nor can they anticipate whether a design decision today will block a necessary improvement tomorrow.

That is the role of human oversight. Open source projects have built governance models that include proposal processes, structured review, community voting, and long-term roadmaps. These processes are not about slowing things down; they are about making sure that acceleration serves a coherent, sustainable vision.

For governments, this kind of structured oversight is essential for building public trust.

Why we need to monitor AI

To use AI responsibly, we need to trust, but verify. A common metaphor is to think of AI as a brilliant toddler. There are problems with this, as AI isn’t human and isn’t learning to be. It is capable of astonishing things: assembling knowledge, writing code, producing text, and suggesting solutions in ways that are surprising and impressive. Many can already pass the Turing test, but they aren’t human. AI lacks judgment and cannot effectively weigh trade-offs, anticipate long-term consequences, or understand the values at stake in public service.

The metaphors for AI are still evolving. Perhaps you want to see AI as a vine which or hedge which needs to be constantly trimmed. Perhaps, like any iterative process, it just needs to be supervised to ensure that it is going in the right direction.

The mistakes made by AI are sometimes amusing, sometimes dangerous. Generally we can assume that AI is not malicious, but we cannot assume the intents will be good. To harness its brilliance safely, AI needs boundaries, supervision, and guidance from people who understand the broader context.

This is where open source communities come in. They provide the “supervision” that keeps the brilliance of the AI contributions aligned with shared objectives. Community members don’t just check whether code compiles; they ask harder questions: Does this serve users? Does it continue to respect accessibility? Does it strengthen security rather than weaken it? Will it still make sense next year, or in five years?

The Real Scarcity: Human Oversight

The risk AI introduces is speed. Even if the models and training weights are made fully open, the fundamental problem remains: AI can regenerate or rewrite functionality at a pace that far outstrips human review. Entire sections of a system can shift in minutes. The bottleneck is no longer writing code, but ensuring that the code aligns with objectives and values.

In practice, this means that governance and review are now the scarcest and most valuable resources in digital projects. Code is cheap; thoughtful evaluation is not. The role of open source communities is to make that evaluation collective and structured, rather than leaving it to a few overstretched individuals. People are ultimately invested in the stakes of technology and its impact on society. Contributors are drawn to open source and civic technology because they want to be able to improve their communities. This is not true of AI.

Community Oversight Beyond AI

It is important to stress that this challenge is not limited to AI. Even human contributors can introduce instability if changes are not carefully reviewed. But AI raises the stakes because it lowers the cost of generating code to nearly zero. With more changes flooding in, the responsibility to filter, assess, and align them grows.

Open source communities provide the right forum for this oversight. Contributions are evaluated in the open, by diverse reviewers, using documented processes. This creates accountability and resilience. No single actor can unilaterally shift the direction of a project without debate, review, and consensus.

For governments, who are being held increasingly to account, this kind of shared oversight is especially valuable. It reduces risk, spreads responsibility, and ensures that critical systems are not beholden to a single company’s priorities.

Why Open Source Matters More Than Ever

For the public sector, the importance of open source in the age of AI comes down to a few key principles:

1. Direction: Communities ensure that rapid code generation serves long-term goals, rather than introducing instability.
2. Responsibility: A broad base of human contributors spreads the work of reviewing and validating changes.
3. Resilience: Community stewardship protects projects from being derailed by sudden shifts in technology or single vendor decisions.
4. Judgment: AI can propose changes, but people decide whether those changes align with public needs and values.
5. Transparency: Open source communities document discussions, decisions, and rationales, making governance itself visible and accountable.

How AI Should Be Used in Open Source Communities

The most constructive way to think about AI is as an assistant, not an authority. It is a tool to streamline human work, not a replacement for human oversight. In open source communities, AI can play several valuable roles:

• Code review assistance: AI can scan proposed changes for security risks, performance issues, or accessibility concerns, flagging them for human attention.
• Documentation support: Generative AI can draft first-pass documentation or tutorials, giving contributors a starting point that humans refine.
• Data auditing: AI can surface anomalies in training sets, such as duplicates or mislabeled data, which human reviewers then evaluate. It can write tests for functions, and provide much better test coverage.
• Community support: AI can handle routine triage of bug reports, and supporting basic questions. It can also provide support for compliance reporting. This will help free human maintainers for higher-value tasks.

In each case, the AI provides acceleration, but the final responsibility rests with people. The AI can experiment, and often surprise us — but the supervisors need to set the boundaries.

Implications for the Public Sector

For governments, the implications are clear. If AI is to be used to deliver public services — whether in code generation, document drafting, or data analysis — it must be embedded in systems that are transparent, accountable, and governed by communities with the authority to say no.

Relying on closed development, or assuming that AI-generated outputs can be trusted at face value, is reckless. Public services cannot afford the fragility that comes with unreviewed, unaligned changes. Citizens depend on these systems to work consistently, fairly, and securely.

Open source offers the structures we need to make AI safe to use: governance processes, transparent debate, documented decision-making, and distributed responsibility. Without these, the brilliance of AI is wasted — or worse, becomes harmful.

Conclusion: Supervising Brilliance

AI can provide compelling and often brilliant responses. It can astonish us with what it creates, but it cannot do so without responsible supervision. It needs communities — especially open source communities — to provide the governance, oversight, and direction that keep it aligned with human needs.

For the public sector, this is not an abstract debate. It is about ensuring that digital government remains stable, trustworthy, and resilient in the face of a technology that can produce too much, too fast. Open source is not just helpful here; it is indispensable.

AI gives us speed, but open source gives us safety. The task ahead is to combine both — guiding the use of AI, constantly finding better applications, but never leaving it unattended.