I wanted to see if AI could provide insight into local government. AI does analysis really well, those datasets are available, and it’s not something a normal person can do, unless they have hours to spare. Can AI simplify government bureaucracies so we can better understand our government. The answer? Yes, it can.
1. I needed to start with a large dataset to test this. I visited the Chicago Data Portal, then narrowed down by “Administration & Finance” and then chose a dataset that would be easy to understand – “Current Employee Names, Salaries and Position Titles.” What I really wanted was to drill down into the city finances, but I figured this would be a good place to start because it’s easy to understand.
2. I narrowed down the data by filtering for full-time, salaried, then I exported it as a CSV (only 2.5 MB – that’s easy to work with.)
3. I booted up Claude. Then I loaded the data and asked Claude where to start, with the following prompt:
“What analyses can you make from the attached data?”
It turns out, a lot. Claude immediately displayed some key insights:
- 24,831 Chicago employees are salaried and full-time.
- 70% of this workforce is public safety (49% is the Chicago Police Department), 19% is the Chicago Fire Department)
- The highest paid departments are the Department of Buildings (#1), the Department of Technology & Innovation (#2), the Chicago Fire Department (#3), the Office of Budget & Management (#4) and the Office of the Mayor (#5).
- Policer Officer is the single largest job title with 32% of all city employees – and this doesn’t include related titles like Detective and Sergeants.
- Some of the highest paid departments are also the smallest: Office of Budget & Management (.21% of the full time salaried workforce); Department of Technology & Innovation (0.39% of the full time salaried workforce), Office of the Mayor (.42% of the full time salaried workforce).
- Highest department payroll of full time salaried workers is Chicago Police ($1.44B), Chicago Fire ($608.8M), Aviation ($94.3M).
- There are 385 unique full time, salaried roles in Chicago (36% of total job titles, a percentage Claude considers “fairly high”). This includes roles like “Superintendent of Police,” “Fire Commissioner,” and “Chief Information Officer.”
Where this is positive for the future: I don’t believe nitpicking or nickel-and-diming organizations is good long-term. You can reduce costs, but you also need to give organizations slack to experiment and make mistakes. I do however agree with Martin Gurri that governing in the future requires making government understandable. That’s what gets me so excited about this experiment. It humanizes a large city and makes it easily understandable. That means more people with good intentions can make volunteer recommendations to make our cities better.
A future experiment: It would certainly be interesting to compare datasets from different cities. Do some cities do this better than others? A large-scale review of different city budgets could be quickly compiled by an analyst and used to guide meetings between large city mayors – helping them understand what works best across different United States cities. This could be used for open-source consulting guides, where citizens compile low cost organizational strategies that can be deployed across cities to improve operations.

Leave a comment