AI for Public Data

Public data do not always arrive as a ready-made dataset. In my Payment Reform research, useful information was scattered across cost reports, historical registries, state records, and other public documents. AI helped bring these pieces together.

I created Public Data Reconstruction, a reusable skill for finding and recovering the data a research project needs from verifiable public sources.

From public records to Payment Reform research data

Explore how 35 public source families connect to 8 data components. Select a node to see what it contributes; drag to rotate and scroll to zoom.

What AI makes possible

Find dispersed evidence

Follow identifiers, past names, and document references across archives to find the missing fact.

Recover usable observations

Extract tables, write repeatable processing code, and reconcile entities, years, and definitions across files.

Keep the work traceable

Record the source behind each value, flag contradictions, and resume unresolved cases without repeating earlier searches.

AI lowers the cost of this detailed work. Values are accepted only when the evidence supports them; unresolved gaps stay explicit.

Try the skill with your own data question

The skill works across topics, industries, and countries. Start with a desired dataset or an existing table with gaps, for example:

I need a yearly dataset of school openings, enrollment, and staffing. Find public records, recover the fields they support, and keep the source and remaining gaps for every school and year.