Open budgets and open data may not be at their fashionable peak, but AI changes what we can see in the data, and what we cannot.
Building a simple budget query tool for a middle-income country’s open data API, I found an open data portal reported at roughly one-fifth of the real figure. The dataset was clean, machine-readable, and systematically incomplete.
For over a decade, the infomediator bottleneck hid this. When only a handful of skilled developers could query a government API, partial data caused limited damage. AI removes that containment: a technology that could potentially automate transparency ends up automating the detection of how little there is to work with.
A few implications for the next ‘AI for accountability’ wave in the full piece (link in the comments). Grateful to Jonathan Fox and the team at the Accountability Research Center for the encouragement and editorial support.