The Bloomberg Philanthropies City Data Alliance (CDA) program helps cities improve operations and produce better outcomes for residents. After an initial Accelerator course, cities take part in one of several specific tracks. The “AI Track” is a new, specialized course for CDA cities in 2025. The program moves cities from interest to practical implementation of AI projects. Over 8 months, cities will: 

• Build foundational knowledge, 
• Explore prototyping AI solutions, and 
• Optimize AI by building a strategy. 

Through coaching, workshops, and hands-on labs with experts in AI, design, product development, engineering, and change management. The goal is to help cities tackle real challenges using AI tools and frameworks.

When it comes to improving city services using data I’ve often said, “We just need to count things better.”

When I was Chief Data Office in Baltimore, I stole that line from a friend and colleague who did similar work in another American city. Artificial intelligence? Predictive analytics? Sure, maybe in a few cases. In Baltimore, we worked with researchers to develop an award-winning AI prototype that detected vacant buildings with collapsed rooftops so we could prioritize them for demolition. But that was with high-resolution aerial imagery, not the workaday datasets from the city’s systems of record – data extracted from legacy systems that could help us get to performance metrics but not much more. In countless cases we had conflicting or unreliable counts of the stuff we owned and operated and their statuses. 

This may sound like a knock on Baltimore in particular or city government in general but I promise it’s not; on the contrary, one ought to be impressed with how much a city can accomplish in spite of its constraints. Baltimore was founded before the Revolutionary War, is the birthplace of redlining, and is perpetually strapped for resources. The services cities provide – dealing with trash, building roads, policing – predate business analytics, Toyota’s lean management principles of the 1990s, and even Frederick Winslow Taylor’s theories of scientific management and Henry Ford’s implementation of them in the assembly line. On top of that, a city must manage services across an extraordinarily wide range of domains – human services, drinking water supply and waste water, education, building codes and enforcement. Just within transportation, building bridges and providing public transit are two wildly different things. I struggle to think of a business that must have expertise in so many areas.

Contrast this with the tech companies providing one-click shopping and auto-playing the next episode of your favorite show in an epic binge session. Not only have they been able to count your every click since the day they launched, each company is really only providing one or a handful of services. 

All this is to say the business of running a city is messy and complex. It’s not one entity, it’s an organization of organizations with sometimes overlapping, sometimes competing goals, with top leadership that is typically required to be replaced every few years (i.e. elections). Anyone who has served in local government will tell you they are simultaneously baffled by the inanity of some of the details but are generally awe-struck by how well it manages to function most days.

Facing this context, city leaders are now expected to answer the question, “How will you implement artificial intelligence?” Both inside and outside of government, true believers will want to wholly replace city services with AI and skeptics will want to prevent their government from using it at all. As with all things, the path forward will lie somewhere in between. We hope that this new AI track in the Bloomberg Philanthropies City Data Alliance will help cities answer that question for their own community.

To get participating cities started, we’ve designed a City AI Readiness Assessment. Knowing where and how to use a general-purpose swiss-army knife tool like generative AI among the myriad different benefits and services the city provides can seem an ill-defined and intractable question – the kind of question that makes you stop before you even get started. But regardless of what you want to build in shop class, your teacher is going to ask if you’ve ever used power tools before. You need the training to use the table saw, you have to put on your safety glasses, and you’re probably going to mangle a few pieces of scrap wood before you build that heirloom chest of drawers. (Oh, and please put the tools away and clean up the sawdust when you’re done.)

The City AI Readiness Assessment is our version of asking cities if they have any experience using power tools before they reach for the circular saw. We ask a few questions about people (e.g. training, internal capacity), policy (governance and rules), and technology (software and infrastructure). Cities get immediate feedback on where they stand and catered recommendations on how they can improve specific gaps. We’ve created the following framework as a guide to cities’ readiness:

This early assessment will also get AI Track coaches a head start on identifying where to focus efforts as cities develop their AI strategy and consider scaling up a prototype. Cities will take the assessment at the end of the program as well to measure where and how they’ve improved their practices in anticipation of further AI usage.

Counting things remains difficult for cities, but that shouldn’t prevent us from conscientiously deploying AI where a community wants it, the right policies are in place, the data are good enough, and the benefits outweigh the risks. We hope the AI Readiness Assessment will put cities on solid footing before they take the next step with AI.