People-first AI delivers twice the returns for leading cities, study finds

Cities taking a people-first approach to artificial intelligence are achieving more than twice the financial returns of their peers, according to new global research that also reveals widespread concerns about costs, public trust and the ability to scale AI projects.

A study by research and economic analysis firm, ThoughtLab, found that cities classified as AI leaders recorded a 96 percent return on investment, compared with 42 percent among other cities. They also recovered their investments in an average of 39 months, against 49 months for their peers.

The findings come from Building a People-First, AI-Powered City, which draws on a survey of 525 senior officials across 215 cities, a review of 363 municipal AI plans and interviews with government, business and academic leaders.

Lou Celi, CEO of ThoughtLab, told Cities Today that the strongest-performing cities are distinguished by their focus on public value rather than technology alone.

“Leaders achieve these better results by starting with a human-centred mindset that defines success as driving public value,” he said. “They build a long-term AI vision and plan around people-first outcomes, then put in place the governance, data, IT, and partnerships to deliver it.”

The research identifies 37 AI-leading cities, including Amsterdam, Barcelona, Dublin, Austin, Montreal and Seattle, using a benchmarking model covering governance, cybersecurity, data readiness, workforce development and partnerships.

The challenge of public trust

Despite the reported financial benefits, 69 percent of cities identified public trust and acceptance as major obstacles to AI adoption, while 78 percent cited concerns about surveillance and privacy.

Celi highlighted a particularly stark observation from Pittsburgh’s Director of Economic Development, Steve Wray, who noted: “The two least trusted bodies in the US are government and AI platforms.”

The comment underlines a challenge for municipalities introducing AI into public services: convincing residents that systems handling sensitive information and influencing government decisions can be trusted.

Some cities are responding by introducing explicit safeguards and greater public scrutiny. Seattle requires human review of generative AI outputs before they are used officially and prohibits social scoring, helping ensure that automated systems do not replace human accountability in government decisions.

In Europe, Amsterdam and Helsinki publish public registers of algorithms used by their administrations, providing greater visibility into how automated technologies are deployed across municipal services.

Turning investment into results

The research highlights tangible operational benefits, including New York’s Metropolitan Transportation Authority reporting a 24 percent reduction in bus maintenance material costs through AI-powered predictive maintenance.

However, achieving similar results remains difficult for many municipalities. Some 84 percent cited concerns about unexpected or long-term costs, 82 percent reported shortages of AI skills and talent, and 80 percent struggled to justify investment returns to stakeholders.

Celi identified funding, workforce readiness and data as the three biggest barriers to scaling AI, warning that some existing roles could become obsolete without adequate retraining.

He also cited San Antonio’s Geoffrey Urbach (Product and Partnerships Lead, Emerging Technology), who observed: “You cannot build reliable AI-assisted decision-making on inconsistent, siloed, poorly governed data.”

The research will be presented at Smart City Expo World Congress in Barcelona, alongside a free AI playbook and benchmarking tool designed to help municipalities assess their AI maturity.

Main image: Lesia Sementsova | Dreamstime.com