Towards Meaningful Transparency in Civic AI Systems

Transparency alone isn't enough if people can't act on it. The paper argues that current AI transparency efforts focus on technical explanations that are difficult for the public to understand and don't connect to any meaningful way citizens can challenge or change decisions—meaning transparency becomes a hollow exercise rather than a tool for accountability.

AI in government services is creating real problems for citizens and workers. The research identifies that AI systems used for benefits decisions and parking fines are producing biased or incorrect outputs while simultaneously reducing the agency of both the people affected by these decisions and the civic workers implementing them.

"Meaningful transparency" requires understanding the full context, not just the algorithm. Rather than focusing narrowly on how an AI system works technically, the paper advocates for transparency that reveals the broader socio-material context—the organizational practices, human decisions, and systemic factors that shape how AI systems actually operate in government.

Current transparency approaches miss the human element. Technical documentation about algorithms often ignores how civic workers and citizens actually interact with these systems in practice, making it impossible for people to understand or influence the real-world impact of AI decisions on their lives.