August 10, 2026
(7 Min Read.) This is a condensed Q&A taken from the recent City & State Tech in the Cities Podcast, Bonus Episode.
To access the full 12 min. podcast interview with more questions, click here!
Ron Greenberg spent three decades and five administrations working in New York State and City government, including two decades in the New York State Division of the Budget. These days he's focused on a different kind of public service: helping government leaders think clearly about artificial intelligence.
Ron Greenberg: So I was in five administrations. I think I'm of that age now where I can say five administrations. But really I'd say over my three decades, and really three governors, I never walked into a room and said, let's sell a technology project.
A good example of that is, under one of the former governors, I led his shared services initiative. When I went in there, I never once said, this is the reason we need to do this. I had what I called a top 50 list, broken up into 7 or 8 categories, and we simply gave people the problems.
We said: did you know that your servers are below the waterline in the city of Albany, at risk of flooding? Did you know security was bad enough that you could climb from the bathroom into the data center? Or how about this: you have 15 separate email systems. We even walked a million square feet at one point and found hundreds of thousands of square feet of rented space sitting vacant. I joke with people that I knew this resonated because it ended up on the front page of The New York Times. I don't know how it got there. But it showed me: figure out what problems you're trying to solve before you get into how you're going to implement anything.
The pivotal moment for me, though, was going down an AI rabbit hole and taking courses online. I came across a company called Ubenwa, which means "the cry of a baby" in Nigerian, that was exploring how AI could listen to a baby's breathing and crying to predict the risk of asphyxiation, one of the top three causes of newborn mortality worldwide, particularly in developing countries. I just froze. We're not talking about AI and hallucinations here, we're talking about solving a real problem. Are there lots of problems like that we could solve?
For years, technology helped government work better. What I see with AI is the possibility of helping government decide better. That's what started me on this journey, the framework, the podcast, all of it. How do we help people make better decisions, before we even get into why it doesn't work?
Ron Greenberg: It starts with a level of enthusiasm. Someone has a demonstration, or heard something, or wants to provide a service. But the problem is everybody owns a part of the process, and no one owns the outcome.
I call it "the big three." Someone says, do we have budget for this? Can we even procure it? What's legal going to say? Dig deeper and the tech folks start talking tech that no one else understands, and the cyber people talk about cyber. Every one of those questions may be appropriate, but the problem is sequencing. The room spends an hour talking and ends up nowhere. It's like a rowboat going in circles.
Government organizes itself around functions, but citizens experience outcomes. And then we wonder why nothing gets done. I shouldn't say nothing gets done — I spent 30 years there, and a lot gets done. But I'm passionate about the opportunity AI presents to fix that dynamic.
Ron Greenberg: The first question I'd always ask is: what problem are you trying to solve? Former colleagues still joke with me about how hard I pushed on that. I'm closer to 60 than 40 now, so I say I'm more Zen about it than I used to be, I was a little angrier back then. But we really beat that into people. Step back before you talk about what you can and can't do. What problem are we trying to solve?
The reason is that everyone in the room thinks they're solving a different problem. One person thinks they're saving money. One thinks they're standardizing systems. Someone's worried about security, someone else about service levels. So the first question is: what policy problem are we trying to solve? The second is: what outcome do we want to see? Those are leadership questions, and we don't ask them often enough.
Ron Greenberg: Government is right to focus on the things it focuses on like inaccuracies, biases, all of that. But when I thought about the Ubenwa example, I realized the risk of not doing something is sometimes greater than the risk of doing something. If we don't act, baby asphyxiation remains a top-three concern worldwide.
So I asked myself: if leaders in New York City and New York State have identified problems they want to solve, could this be a tool that helps, before we simply decide it isn't right? That's the risk on the other side of the ledger.
Ron Greenberg: It really grew out of all the work I've done over the years, trying to frame this properly. It comes down to five questions. The first two are leadership questions, about purpose. Then it moves into the technical questions.
To be specific: "let's modernize government" is not a policy problem. "Let's use AI" is not a policy problem. But "children at risk" is a problem. Climate might be a problem. Natural disaster response might be a problem.
So the five questions are: What's the policy problem? What's the outcome we want? Do we have the data to achieve that outcome? Do we have governance and oversight in place? And finally, the key question, is AI even the right tool? It may not be. That analysis might tell us we need a policy change, or a redesigned process, or a different kind of leadership decision. Or AI might genuinely be the solution. But if we start by assuming AI is the answer, we never actually get there. In some ways it stopped being just an AI framework and became a framework for solving problems in government generally.
Ron Greenberg: My hope isn't that AI makes government less human. My hope is that public servants get more time to do the work that requires humanity. What if a caseworker had more time to spend with a family? What if a nurse had more time with a patient? What if people in New York City had more time to think about the most effective way to plow the streets, or pick up the garbage, or handle public safety?
That's the misconception: AI is a tool. You can control it, or you can't. If you control it and define it the way we've talked about, then success isn't measured by how much AI government deploys. It's measured by whether citizens receive a better government.
Ron Greenberg: What public policy problem are we trying to solve? If leaders can't answer that, they haven't actually defined the technology or the solution. It forces clarity. It stops people from chasing different objectives under the same project name.
That matters even more with AI, because the technology can influence decisions that affect people's lives — and that's where people get nervous. They worry about a button being pushed and a decision being made for them. But that's not what has to happen. People have to keep asking what public policy problem they're solving. I'll close with this: better government is the mission. AI is just one of the tools.
To contact Ron Greenberg and read his bio, click here.
This interview has been edited for length and clarity. To access the full 12 min. interview with even more questions, listen here!
Updates, news, and information for your industries.
