Earlier this year my friend (a software engineer at Google) sent this article and jokingly commented that a bot had done better than them at Google’s coding interview.

In the following days, I heard a range of stories highlighting the capabilities of this bot from generating a meal and weight loss plan (see tweet below) to birthday ideas for a 10-year-old to helping students cheat.

https://twitter.com/anothercohen/status/1599531037570502656

I took none of this with much seriousness until I heard Satya Nadela, Microsoft CEO ( Add Footnote: Microsoft Invests $10 Billion in ChatGPT Maker OpenAI) talk about ChatGPT and how it was helping farmers in rural India get access to e-government services in a local regional language. This is when I started thinking about ChatGPT and its potential to help local governments in the United States (US) and around the world.

Satya Nadella on How ChatGPT is Helping Villagers in India

What is ChatGPT? What makes it different from the existing Chatbots?

https://images.unsplash.com/photo-1662974770404-468fd9660389?ixlib=rb-4.0.3&q=80&fm=jpg&crop=entropy&cs=tinysrgb

Chatbots are not a new concept, in fact, they are pretty widely used across the public and the private sector. If you shop online or book appointments or even use social media chances are you have used/interacted with a chatbot. So, before we get into ChatGPT, here is a quick overview of what chatbots are:

Chatbots are software programs that use automation or artificial intelligence (AI) techniques to provide support and simulate a human-like conversation with the end user. Chatbots can primarily be categorized into:

(i) Rule-based (also known as menu-based or button-based); and

(ii) AI-based.

Rule-based chatbots follow a certain set of input instructions and decision trees to guide the conversation. You may have used one of these on a DMV website, your city’s website, or while booking a Doctor’s appointment.

AI-based chatbots, on the other hand, are powered by machine learning models. They are more sophisticated as they can understand the context of queries to solve problems and continuously learn to produce more accurate responses over time.

Add Footnote: Quick reads to learn more about Chatbot

What Is A Chatbot? Everything You Need To Know

What is a chatbot? | IBM

 

So now what exactly is ChatGPT and what’s the hype all about? ChatGPT is a large language model AI chatbot developed by OpenAI. While Google (Bard – join the waitlist to try it out) and Meta (Galactica – taken down three days after launch) ********have developed their own large language models, what makes ChatGPT quite the craze is its public availability. OpenAI launched the chatbot to the public in November 2022 and attracted some 100 million users who experimented and “played” with the bot. Since then OpenAI has launched a newer version ChatGPT-4 and introduced a paid version (ChatGPT plus) as the company plans to eventually monetize the chatbot.

OpenAI ChatGPT Landing Page Description

Source: OpenAI ChatGPT

Here is a Twitter thread that explains how ChatGPT works in simple words:

Major cities such as Singapore and Dubai are already exploring the potential uses of ChatGPT in making their services more efficient. The Singaporean government is considering using the technology to help reduce the time spent by their officials on knowledge work such as writing and reviewing policies and doing research. Dubai Electricity and Water Authority (DEWA) already announced that it will be using ChatGPT, supported by Microsoft’s digital and cloud services, to help its customers and employees.

While doing research for this article, I did not come across any cities in the US that are considering or have adopted ChatGPT. However, we know that both state and local governments have heavily relied on chatbots to manage the tsunami of public inquiries in the wake of the pandemic. Several cities have launched city chatbots to support their residents and provide city services, for example, see California DMV’s Miles, Kansas City Open Data Chatbot, Los Angles City’s chatbot “Chip”, and San Francisco’s procurement chatbot “PAIGE” (Procurement Answers and Information Guided Experience) to name a few. Since ChatGPT is a much more powerful version of AI-based chatbots, it is only a matter of time before cities and government agencies upgrade their chatbots. The conversational nature of ChatGPT can transform an array of smart city issues such as smart mobility, public safety, and civic engagement.

Closing Thoughts

While much remains unknown about the impact of ChatGPT, it has enormous potential to revolutionize how residents interact with their urban settings and vice versa. ChatGPT and similar AI technologies can potentially transform anything from smart mobility to public safety to climate change in smart cities. The low-hanging fruit (low-risk and high-impact area) will be knowledge work, civic engagement, and e-government services. It can reshape resident engagement and improve awareness and access to e-government services and programs. Just as in the example in rural India at the start of the article, ChatGPT can reduce city officials’ workload by tackling resident queries more effectively and accurately. Furthermore, it can establish transparency and efficiency by reducing the search cost and time a resident is likely to spend searching for information about City’s functioning, services, or resident benefits.

Similar to google and other search engines, ChatGPT provides information available on the internet, however, it has no ability to fact-check or verify the information it generates. The makers of ChatGPT also acknowledge the limitations of the bot which include inaccurate or wrong information, sensitivity to phrasing, and responding to “harmful instructions or exhibit biased behavior.” Furthermore, just like any other AI technology, ChatGPT is not free of bias.

The fact is ChatGPT is just the beginning of a new era of tech. There are already many technologies like ChatGPT that can help us design, write, code, translate, and research. Like it or not but very soon AI-powered technologies such as ChatGPT are going to become mainstream and widely used. Check out this long-term timeline of technology by Max Roser that will help you understand how much tech has already changed the world and how drastically it will change the world in our lifetime.

So the right question cities should be asking is how to ensure the ethical adoption of AI technologies. A two-step cautiously optimistic approach toward AI technologies like ChatGPT is as follows:

  1. Assess the AI maturity and readiness of your department. This is critical to ensure your department is ready to adopt an AI system and successful deployment and scale it. Here is an article on “An AI Maturity and Readiness Model for Governments” I co-authored during my time at UC Berkeley’s Goldman School of Public Policy.
  2. Commit to AI trustworthiness. Cities can ensure AI trustworthiness by aspiring for transparency and accountability in how resident data is managed through its entire data lifecycle. The U.S. National Institute of Standards and Technology (NIST)’s AI Risk Management Framework is a great resource to minimize risks arising from the use of AI technologies. To understand what AI trustworthiness means, I highly recommend reviewing this recently released paper, “****A Taxonomy of Trustworthiness for Artificial Intelligence”** by Jessica Newman.

My Experiments with Chat GPT

How can ChatGPT help smart cities?

Yes, I did take some help from ChatGPT to write this blog!

Give me ideas for a smart city proposal

Best smart city in America