> ## Content Index > Fetch the complete content index at: https://preview.jaynath.com/llms.txt > Use this file to discover other available public pages before exploring further. # Government Needs More Leverage - URL: https://preview.jaynath.com/what-do-we-want-to-make-better/ - Published: 2026-10-11T03:10:50.000Z - Updated: 2026-10-11T04:17:02.000Z - Description: Why I see AI as a way to expand public capacity without relying only on bigger budgets. - Author: Jay Nath When I was Chief Innovation Officer in San Francisco’s mayor’s office, much of my work was about increasing government’s capacity. The city had important problems to solve, and our staff had limited time and resources to address them. Through Civic Bridge, we brought people from other sectors to work alongside city employees, adding skills, perspectives, and hands to the work. That experience shapes how I think about AI. I see another opportunity to increase what government can accomplish: giving public servants tools that help them get more from their time and expertise. The problem, as I see it, is a mismatch between the scale of government’s responsibilities and its capacity to fulfill them. We ask public institutions to maintain infrastructure, prepare for emergencies, manage public funds, enforce laws fairly, and deliver essential services. Each responsibility requires people to understand problems, make decisions, and carry out the work. When capacity is limited, something has to wait, receive less attention, or go undone. Adding people and funding can be necessary. But I don’t think larger budgets and more staff can be our only answer. We need economic growth to support the things we want to do as a society, and we need government to produce more value from the resources it has. Asking public servants to work harder will only take us so far. Better tools offer another way forward. They increase what a person can accomplish with a given amount of time and effort. That is the leverage I’m interested in: helping people use their expertise across more work, or examine a problem more thoroughly, without increasing the effort required at the same rate. My hypothesis is that AI can provide some of that leverage. It could help public servants find and analyze information, compare options, prepare work for review, and carry out routine tasks. If it does those things reliably, government could take on work it currently struggles to complete and improve the quality of work it already does. Consider a team reviewing inspection records to decide which infrastructure needs closer attention. Staff might spend days locating reports and assembling the relevant history before they can assess the risks. AI could help organize those records, identify recurring issues, and point reviewers to the supporting evidence. Engineers would still need to check the analysis and decide what warrants action. The potential gain is that they could examine more records or spend more time investigating the problems they find. The same idea applies to someone answering a question about a public service. An employee may need to check a rule, locate a record, and consult another department before explaining what the person needs to do next. A tool that finds the relevant guidance and shows its source could help them respond sooner and spend more time on cases that require judgment. These are possibilities to test, not results to assume. A system that produces plausible but incorrect summaries could create more work. Even an accurate tool might save time in one step without improving the overall process. If staff can review information faster but a decision still sits in the same queue for weeks, we need to understand what else is holding up the work. I would judge the value by what government can actually accomplish. Can a team identify a maintenance problem earlier? Can an agency complete work that had been postponed? Can an employee give someone a clear answer without another unnecessary call? Productivity matters because it can make those outcomes possible. That was also the point of drawing on other sectors through Civic Bridge. City staff brought knowledge of the problem and responsibility for the outcome. Outside partners contributed capacity the team could use. I liked seeing what became possible when people had more help and a different set of skills available to them. AI offers a different kind of help, but I’m interested in it for a similar reason. Public servants already have expertise that matters. I want them to have better tools for applying it. More capacity could mean clearer answers, more timely action, and more room for sound judgment across the work government does. Realizing that potential requires investment and effort of its own. Agencies need to learn where the tools help, check their performance, and change how work gets done. That is the question I explore in [The Capacity to Put AI to Work](https://preview.jaynath.com/the-capacity-to-put-ai-to-work/): how do we build the capacity needed to make useful adoption possible? As Chief Innovation Officer, I tried to help government accomplish more by bringing in capacity from other sectors. Today, I want to understand how much more its people could accomplish with AI. The tools have changed, but the question that interests me is familiar: what could we get done that we couldn’t do before? --- *Written with drafting and editing assistance from Codex. The experiences and 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.*