Taking AI experimentation to the next level in cities

Whether it’s speeding up the procurement process, guiding residents in accessing essential services or involving citizens in long-term planning, generative artificial intelligence is starting to change the way cities work. But it’s not technology alone that’s driving these changes: it’s also local leaders working to identify the most appropriate uses of technology. And at Bloomberg Philanthropies, we believe that by constantly experimenting with how generative AI can be integrated into their daily practices, innovative cities will exploit its full potential while being aware of its limitations.

To deepen their understanding of generative AI and how it can be ethically applied to their work, 120 data and innovation leaders from more than 90 cities recently came together for a City Innovation Studio at Bloomberg City Lab In Mexico. There, they explored exactly how their innovation skills can take AI experimentation to the next level and what that work can, in turn, produce for the people they serve.

Introducing the “irregular frontier” of AI in government.

Mitchell Weiss, professor at Harvard Business School and senior advisor to Bloomberg Harvard City Leadership Initiativepresented those gathered in Mexico City with a comprehensive framework for them to be more ambitious about artificial intelligence in their work. The approach is based on a working document co-authored by some of his colleagues, which states that genAI helps people increase their speed and produce higher quality results when used to tackle specific tasks within the “frontier irregular”. This boundary is best understood as a curve demarcating what lies “inside” and “outside” the technology’s current set of capabilities.

While the paper explored the use of genAI in the private sector, Weiss believes that embracing the idea of ​​an irregular boundary can help urban innovators do more with the technology on behalf of the public. What he means in practice is that they won’t know which tasks are and are not within the capabilities of generative AI until they discover for themselves where this irregular boundary lies. That means pursuing constant, bold experimentation, and that’s exactly what they did as a group in Mexico City.

“The problem with AI, at least generative AI, is you don’t actually know what it’s going to do,” Weiss tells Bloomberg Cities. “There is no real user manual. There is no definitive list of tasks he can do, either with you or at your direction, until you try.

Experiment to reveal utility and face limitations.

The specific tasks that fall on either side of this irregular boundary – which may or may not fall within the current capabilities of generative AI – may surprise city leaders who have not yet devoted much time to the technology. As Weiss demonstrated at the City Innovation Studio, a request as simple as producing a group of sentences of a dozen words each was too difficult for one of the most popular AI tools.

“The tools are not designed to be quantitative tools. These are linguistic models,” explains Weiss, before noting that what works one week or one month may be different the next: “They acquire more and more quantitative skills as they go along. »

Weiss believes city leaders may not fully appreciate how useful this technology can already be. At the City Innovation Studio, local innovators gained perspective on this topic when they took on a hypothetical challenge of prototyping a new heat mitigation solution in a city and discovered that generative AI tools produced a wide range of concrete results in no time. . Specifically, generative AI has helped innovators very quickly identify useful data sources, generate new ideas, review case studies from other cities, and create Minimum Viable Product (MVP) models. and project documentation.

“They used the tools of problem definition, ideation and prototyping, and in a very short time (it was a real shooting exercise) they came up with some truly amazing results,” explains Francisca Rojas. , academic director of Bloomberg Center for Public Innovation at Johns Hopkins University.

Improve data and other key inputs to produce better results.

Just as urban innovators know that it is always important to test, evaluate, and adjust solutions as necessary, it can be helpful for them to keep an open mind that a task that is not current capabilities of AI is not necessarily a reason to stop trying. In other words, past behavior is not an indicator of future capability in generative AI tools.

An example that appeared during a task session potentially Being outside the boundaries – understanding the gaps in services for a specific demographic in a specific city – represented an opportunity to improve the tool rather than a failure, Weiss noted.

“A question you should have in mind with these things that we think are outside the boundary is whether you could, in any way, by virtue of your skills or knowledge, to actually bring these things to the border,” he told city leaders. . After all, in a case like this, a city may simply not have published clear documentation of its limited ability to provide services. To incorporate this into its work, a generative AI tool must be trained on the relevant dataset.

Likewise, a task such as capturing the perspective of a minority community living in a city may, at first glance, seem beyond the capabilities of technology. Certainly, these tools cannot replace engagement with actual residents. But one of the key takeaways here is that if leaders direct generative AI tools toward new data and insights, they can begin to address these gaps.

“What we’re seeing is that cities that are actively engaged in finding ways to adopt and integrate this type of general-purpose technology into their municipal operations (are) training the models with their own data,” Rojas explains.

Meanwhile, Beth Blauer, associate vice provost for public sector innovation at Johns Hopkins University, where she oversees the Bloomberg Center for Government Excellence and the Bloomberg Center for Public Innovation – advises cities to keep in mind that generative AI alone does not include all the precautions that should be part of every government activity.

“It was not designed to protect personally identifiable information. It was not designed to take into account the regulatory frameworks that the government faces when it has to make decisions,” she notes. Still, she encourages cities to explore the irregular frontier “to understand what they are good at, what they are not good at, and then demand” more from their private sector partners who develop these tools.

Cities will therefore always need to ensure they use engagement to involve residents in the journey and be aware of ethical considerations in their genAI work. If they succeed, innovative cities are now able to advance the entire public sector.

“Use it in what you do every day,” says Weiss. “I know that doesn’t seem profound, but that’s what I would advise.”