Generative AI (often known through services such as ChatGPT or Google’s Bard) is a rapidly evolving field with the potential to revolutionise many aspects of our lives, including the way our cities are run.

Many cities are still at the beginning of their journey in using generative AI to improve local government and address some of the challenges they face when implementing this technology. However, a few cities have begun to explore the potential of Generative AI and early results show significant potential benefit. To help UTA members understand better this important area, this article presents some background and lessons based on initial investigation from Bloomberg’s City initiative, London’s Office of Technology and Innovation (LOTI) and Japan’s Yokosuke city.

What is Generative AI?

Generative AI is a type of artificial intelligence that can create new content, such as images, text, and music. It does this by learning from a large dataset of existing data and then using that data to generate new outputs. Text based Generative AI, because it has been trained on large amounts of English language text, can understand natural language questions and generate suitable responses. It has the potential to be used for a wide variety of applications and is currently the subject of intense experimentation worldwide.

How are Cities Using Generative AI?

Cities are using generative AI in a variety of ways to improve their operations. Both Bloomberg and LOTI surveyed city staff to understand current usage as has Yokosuke city in Japan, who work closely with one of UTA’s academic members, Keio University.

The Bloomberg, LOTI and Yokosuka surveys showed that some of the most common applications include:

Basic use cases

  • Internal documentation: Generative AI excels at writing text, or reading and summarising text and documents. Many city staff are already using services such as ChatGPT to summarise complex documents, or help them write text and documents for internal use – see the graphic example from LOTI.
  • Brainstorming and ideation. City staff are using Generative AI to help them generate new ideas and to explore possible approaches to problems. A typical workflow is to generate some initial ideas using GenAI, tailor these based on staff expertise and then challenge the GenAI to critique and provide feedback
  • External communications: City staff are using GenAI to help them create external communications. Popular examples are summarizing complex policy/legal texts in language that residents can understand, tailoring existing comms material to target demographics, or developing surveys to better understand residents needs.
  • Improving data analysis by city staff: Some city staff have already started using GenAI to help analyse data for internal use. In simple cases this is asking for help with Excel spreadsheets (See the Yokosuka presentation at the end), in more complex situations, GenAI is helping write python code to analyse complex data sets.
  • Research and Education: Generative AI can be used better understand issues and to personalise learning experiences for city staff aiming to improve their skill sets. It can provide staff with real-time feedback, help them self test, and suggest ways to improve their knowledge base.

This example is from the LOTI guide for City Staff which contains more suggestions and examples to guide usage by city staff more.

More complex usage

These uses-cases will probably go beyond just using Generative AI ‘off the shelf’ and may require fine tuning or augmenting existing models could be used to handle tasks such as:

  • Social care support: triaging cases, providing summaries etc, but also offering targeted support suggestions for care staff. For example, one London local council is already using such an approach to help local residents who are supporting a family member (dementia, chronic health issues) with tailored health care support suggestions. 
  • Improving access to city services and information: One of the most popular uses of Generative AI is through ChatBots. Leading cities are starting to use GenerativeAI to develop smarter chatbots that handle basic citizen queries through the cities’ website – these can be used to support city staff, or in some cases alone.
  • Building, Planning, Procurement and Licencing: Planning applications triaging and summarisation: Generative AI can be used to improve the efficiency and resilience of planning – for example identifying missing sections in applications or to identify potential future risks (fire, safety etc)
  • Public safety: Generative AI can be used to improve public safety by predicting crime patterns, identifying potential threats, and optimising emergency response efforts.
  • Climate: Generative AI can be used to mitigate the effects of climate change by identifying opportunities for the use of renewable energy sources, optimising energy consumption, and improving disaster preparedness.

See the LOTI guide for CIO’s for more complex examples (Section 08)

Challenges of Implementing Generative AI

Despite the many potential benefits of generative AI, there are also a number of challenges that cities face when implementing this technology. Some of the most common challenges include:

  • Reputational Risk: Generative AI that is used in citizen facing tasks risk affecting trust and reputation. Apart from internal biases, tools can also generate plausible but incorrect results. Careful consideration of where and when to use Generative AI and processes to check output is critical
  • Data: Generative AI requires large amounts of high-quality data to train. Cities may not have the resources to collect and store this data, and they may also face privacy concerns when collecting data from residents.
  • Ethics: There are a number of ethical concerns related to the use of generative AI, such as the potential for bias, discrimination, and misuse. Cities need to have a plan for how they will address these concerns before deploying generative AI solutions.
  • Cost: Generative AI can be expensive to implement and maintain. Cities need to carefully consider the costs and benefits of using this technology before investing in it.

Bloomberg offer the following guidelines for cities considering how to get started with Generative AI:

  • Designate a leader who is free to explore uses and ask vital questions.
  • Learn through testing the technology—and dreaming big.
  • Share early guidance and guardrails without immediately imposing policies that discourage exploration.
  • Understand how generative AI is already being used in city hall.
  • Create a (safe) space for experimentation.

Conclusion

Generative AI has the potential to transform cities for the better. However, cities need to carefully consider the challenges of implementing this technology before deploying it. By developing a plan for how they will use generative AI and addressing the ethical concerns, cities can harness the power of this technology to improve the lives of their residents.

Further readings and resources

The London Office of Technology and Innovation has produced a number of helpful guides including:

General guide for city leaders

A guide targeted at city staff

A guide target at city CIOs

A longer report analysing feedback and discussions from a series of surveys and workshops they carried out – “state of play report”

InnovateUS has produced an excellent video that explains GenerativeAI, and then provides hands-on training on how City staff can use it in their daily tasks. It goes through several common tasks, e.g. generating a memo, simplifying text and summarising a document showing pros and cons of using GenAI for each use case. It’s 13 mins long, but worth watching all the way through if you are just getting started with GenAI.

Yokosuka city have produced a document detailing their surveys and training – it’s in Japanese but you can copy and past the parts that look relevant to you into your favourite translation tool, or even a Generative AI tool such as ChatGPT or Bard 🙂