Extending the Reach of Government

When I was Chief Innovation Officer in San Francisco’s Mayor’s Office, I worked to increase government’s capacity by bringing in help from other sectors. Through Civic Bridge, people worked alongside city employees on public problems, contributing skills and time that the city teams did not have enough of on their own. I was trying to expand what those teams were able to take on.

That experience shapes my interest in AI. Government has substantial expertise, but only so many hours in which to apply it. Better tools give that expertise more reach. I see AI as an opportunity to increase what existing teams can accomplish, including work they have had to leave undone.

The mismatch between responsibilities and capacity runs across government. Maintaining infrastructure, preparing for emergencies, managing public funds, and delivering services all compete for attention. A team can be fully occupied with urgent work while knowing that other problems deserve investigation. Those problems do not disappear because nobody has time for them.

More staff and funding can be necessary. But asking for more resources cannot be our only response, and asking the same people to work harder does little to change the effort a task requires. Technology gives us another way to increase capacity: reduce the work required to apply someone’s expertise, so that expertise goes further.

Consider an engineering team reviewing infrastructure. Before engineers can decide which assets need closer attention, they need inspection reports, maintenance histories, and a way to connect records about the same asset. Gathering that material takes time that is then unavailable for investigating what it reveals.

There is a practical role for AI here: assembling records and surfacing recurring issues, with supporting evidence for engineers to check. The engineers retain responsibility for interpreting the findings and looking for omissions. If the assistance reduces the total effort involved, they gain room to examine more assets or investigate a concern more thoroughly.

That is what I mean by leverage. The same expertise reaches more work, and the team has more room to exercise judgment. I am interested in that expansion because an agency’s workload tells us only part of the story. There is also the work it never gets to start.

The real test is whether the team has more capacity after accounting for the effort of using the tool. Checking and correcting a tool’s output takes time too, and an earlier finding still needs people and resources to act on it. Buying access to AI does not, by itself, increase a team’s capacity.

Even with those limits, I think this is worth pursuing. In government, I looked beyond the resources inside a department to find people who could help. AI offers a different kind of help, but the ambition is familiar. I want public servants to have the tools to take on a problem they have kept setting aside, or give an important decision the attention it deserves. That is the reach I want to extend.

Written with AI assistance. Final editorial decisions are my own.

I work at OpenAI. The views expressed here are my own and do not necessarily reflect those of my employer.