Learning how to make the most of new technologies in government

On May 2, over 70 state government employees joined InnovateUS and Santiago Garces, Chief Information Officer (CIO) for the City of Boston, in a 90-minute workshop on “Making the Most of New Technologies While Avoiding the Risks and Pitfalls.” Before becoming CIO of Boston, Santi was the Executive Director of the Department of Community Investment in South Bend, Indiana, and has served as CIO of South Bend and Pittsburgh, Pennsylvania. In this workshop, Garces introduced the uses and advantages of new technologies for public sector workers as well as mitigation strategies for risks such as privacy, security, and bias.

To begin the workshop, Garces polled participants to gauge how many of them had used machine learning or artificial intelligence (AI) in their roles within government. Only 9% of participants claimed to have used this technology regularly in their position, while 20% said they used this technology sometimes. In contrast, 71% of participants revealed they have never used AI or machine learning in their government job. Garces used this data to explain why this training is essential for government workers, as it is necessary to understand this new technology as it develops.

The Risks and Pitfalls of New Technology

“There are times when we implement a technology and we don’t fully understand how it’s going to operate based on the novelty of it,” he said.

In introducing the idea of what these pitfalls look like, Garces described the flaws in the Correctional Offender Management Profiling for Alternative Sanctions software, or COMPAS. This technology, which was intended to identify the risk of reoffense for previous offenders, was found to overidentify the risk of reoffense for Black offenders and under-identify the risk for white offenders. In order to determine the machine’s bias, researchers tested the algorithm against existing, known data. With this example, Garces explained that the data used to train and teach these technologies is essential.

“These are systems that help come to conclusions based on really large amounts of data, but the data that is used to generate the intelligence matters,” he said.

Garces then went on to emphasize that technology is neither good nor bad, but it is rather how we use technology that brings value and risk. For example, while cameras are great tools for detecting people or objects, there are certain risks that come with their use, such as privacy and security concerns. Another example of this idea is blockchain technology. While it can be used for registration and permitting, there is an environmental cost that comes with using it. 

New technology has to create value through efficiencies that display an existing cost. When measuring the value of new technology, governments must assess if utilizing these resources are worth an increase in other fees. Additionally, Garces warned attendees to watch out for solutionism when considering implementing new technological resources. Common risks and pitfalls when it comes to technology include privacy concerns with its usage, data access and security, and system bias.

“Sometimes buying technology doesn't solve the problem,” he said. “But technology can help complement what you’re doing to solve the problem.”

How to Buy New Technology

Garces then discussed how to procure new technology in government. When polling workshop attendees, 13% of participants had a lot of experience procuring technology, 23% had some experience, and 64% had no experience with the process.

There are three ways to purchase new technology:

The first way is analyzing technology in development, otherwise known as pilots. This is the stage where state workers should partner with researchers, universities, and start-ups to test use cases and validate maturity. In these stages, it is worth considering shifting the financial burden in purchasing these resources to external grants or venture capital funding. 

The second way to buy new technology is by looking at active business models in development or test procurements. It is important in this process to identify comparisons with the incumbent technology, identify evaluation metrics, and integrate the system with existing operations. 

Finally, you can look at procurement as a completed, full solution. In these situations, the value of the technology is understood, and there are set expectations for the desired outcomes of implementing this technology.

The remainder of the workshop was set around three crucial case studies in exploring the risks and pitfalls of new technologies.

Case Study 1: Smart Sewers South Bend

To demonstrate successful implementations of new technologies, Garces highlighted his previous experience with Smart Sewers in South Bend. Previously, due to inefficient infrastructure, when it rained heavily a mix of stormwater and sewage would go into residents’ basements or into local rivers and oceans. To fix the problem, government officials went to Notre Dame to look for solutions. 

The university collaborated with a smart water technology startup and the City of South Bend to design a system that relies on a network of sensors to monitor the flow of stormwater and wastewater. When necessary this system will divert water to areas of excess capacity within a combined system to reduce or prevent overflows. This technology is estimated to save South Bend $437 million by 2038.

Case Study 2: Rebuilding Our Streets

Garces also introduced participants to another case study involving traffic and the quality of roads in South Bend. Residents of South Bend were disproportionately upset with the streets in the city compared to residents of the Great Lakes and the U.S. nationwide.

To fix the problem, the South Bend government consulted the Robotics Lab at Carnegie Mellon University in Pittsburgh, Pennsylvania. Researchers used machine learning to identify defects on the pavement, generate maps, and score the area using the rating system of the Streets Department. With this technology, they created a public dashboard and a $25 million investment in roads.

“Over three years the goal was to bring all of the streets under poor conditions back to being good,” Garces said. “[There was] a significant improvement in the satisfaction of residents in the condition of streets.”

Case Study 3: ChatGPT

With the rise of generative AI like ChatGPT, Garces emphasized the importance of creating specific protocols for using these technologies in government.

The City of Boston created the City of Boston Interim Guidelines for Using Generative AI  to serve as an interim resource for city employees. This is intended to identify different values, privacy expectations, and privacy security guidelines. The guidelines outline three important requirements.

1. Everything generated needs to be reviewed by a person
2. No private information can be shared in the prompt
3. Employees must be aware of the bias and inaccuracies that come with AI

Main Takeaways 

• To conclude the workshop, Garces highlighted 6 main takeaways from his presentation.
• Technology is neither good nor bad
• Technology can serve a public purpose to be used in government
• We can quantify the value of technology
• Some pitfalls to be aware of: privacy, security, and bias
• It is ok to experiment responsibly
• Manage and share risks actively: if it is not clear that something will provide value, maybe someone else should try it

Participant Feedback

Here’s what some participants had to say about the workshop!

“I found the content informative and well-organized.” -California Attendee, Health and Human Services, Advanced Career

“Good review for someone less informed.” -New Jersey Attendee, Sustainability, Advanced Career

To watch the workshop, check out the recording here! Make sure to sign up for our upcoming workshops here!