Can AI Achieve the Broken Promises of Smart Cities?
An interview with Neil Kleiman on transforming municipal services through AI
Last month, the New America RethinkAI coalition, which works with communities to build AI pilots and transforms findings into guidance and policy recommendations, published a new report, “Making AI Work for the Public: An ALT Perspective.” Written by Neil Kleiman, Senior Fellow and Professor, Burnes Center for Social Change, Northeastern University; Eric Gordon, Director, Center for Media Innovation & Social Impact, Boston University; and Mai-Ling Garcia, Director, Emerging Technologies and AI, Bloomberg Centers for Government Excellence and Public Innovation, Johns Hopkins University, the report proposes a new framework for local governments as they integrate AI into their work.
I recently spoke with Kleiman about the report, and what follows is an edited version of our conversation. I used AI to transcribe the interview — AI transcription is a massive times savings for journalists — and as Neil notes, that’s an example of how human intelligence might benefit from AI, not be replaced by it. But will AI truly transform our cities, or will the hype produce another disappointment like the smart cities and civic tech revolution promised to us 15 years ago? I think some of the new city government pilots (see below graphic from the report) point to a hopeful future where we’re using AI to help cities catch up on the very time-consuming work of bureaucracy — I see you, building permits! If AI can help the public sector work at the speed residents want and expect, everyone wins.
The opening of your report reads almost like a eulogy for old civic tech. Many people remember being promised “smart cities” that never quite materialized. What went wrong there, and why will AI be different?
Kleiman: I don’t know if anything necessarily went wrong with Civic Tech 1.0 or Smart Cities. It was more that we lost sight of the prize. The prize was really to transform government into something that hadn’t previously existed—both well-functioning and having the trust of citizens and residents.
We fell into what almost any organization, particularly the public sector, falls into: trying to build a better widget faster. It was essentially an efficiency trap. There was this pull to focus on efficiency rather than significantly rethinking how the public sector operates.
The other issue was the tendency to think about tech in this fetishized, specialized way. It was either in the innovation office or with the tech folks— so those not part of innovation or tech discussions were often left thinking “that’s cool, but I don’t really understand it.” Improving government with technology has to apply to everybody. It’s everybody’s job to understand these new tools and use them effectively.
How is AI different from previous technology waves?
AI has the potential to truly transform government for two fundamental reasons. First, AI does a remarkable job of translating almost anything into simple terms—whether that’s coding, dense zoning regulations, translating different languages, or transcribing interviews.
Second, it’s insanely fast. It happens in milliseconds. You don’t have to hire a data scientist or computer engineer. You just open your computer, voice a question, and get an answer in simple terms.
We’re most excited about using this technology to combine community insights with traditional government data. For example, in Dorchester, Mass., a community with very little trust in AI or local government, we created a public common data corpus. We seamlessly combined years of community meeting notes and interviews with traditional crime data, creating a more complete reflection of what crime actually looked like in the neighborhood.
Your report introduces the ALT framework—Adaptation, Listening, and Trust. Why do cities need this framework to lead with AI?
The framework reflects how we’ve long thought government could improve and transform. Looking back at smart cities, one of the biggest challenges was that cities would set up innovations like mobile 311 or SeeClickFix, then be overwhelmed by new demands from citizens—new demands layered on top of the same old administrative service system. We need a new principle for engaging with residents: being adaptive based on their needs.
When you add AI into the mix, this becomes undeniable. Even if a government wants to put its head in the sand about AI, it’s still going to happen. Demand will rise in different places, so it’s incumbent upon government to adapt.
The listening component goes back to a commitment we had in the civic tech movement but lost sight of—that fundamentally, tech tools can and should be used to better understand what residents want.
The trust component is really traditional good government reform—it’s accountability. If we’re going to adapt and listen, we must hold ourselves accountable to those results.
You note that leadership on AI is coming from Chief Information Officers rather than mayors. Are cities aware of how this role is changing?