Edmonton, AB builds AI framework to deliver smarter projects





AI is one of the buzziest concepts in local government innovation.

The ability to use data and artificial intelligence to make decisions and streamline operations is being applied to everything from $ citizen engagement$  to $ traffic management$  to $ preventing shark attacks$ .

But while many civic innovators have tackled individual problems this way, $ Edmonton, AB$ , is taking a different approach. Rather than implementing AI for a particular initiative, the city has built a framework and culture that can apply artificial intelligence to solve challenges across the government.

This strategy has opened up the use of AI and data-powered initiatives on a much wider scale for Edmonton, and it’s allowing the city to make smarter, more efficient decisions and workflows. The best part? The city’s data science foundation is one that many other local governments can implement as well.

As Govlaunch works to build the global wiki for local government innovation, we’re highlighting a series of Innovators — cities, towns, and counties who are implementing transformative ideas and fostering a culture of innovation. We chatted with data scientists and codes & inspections workers in Edmonton to talk about how local governments of all sizes can take advantage of artificial intelligence.



What’s AI?

If the most you know about AI is that it stands for “artificial intelligence,” you’re not alone. Many people know of the concept but have never worked with AI to truly understand what it entails.

Essentially, AI uses algorithms or models to take action, make decisions or predict an outcome based on data. It’s done by a computer instead of a human (hence the “artificial”), and is becoming increasingly popular for its ability to automate tasks that don’t require a lot of critical thinking.

How local governments can use AI

Artificial intelligence’s uses are so broadly applicable, it provides a huge opportunity for local governments, says $ Ben Gready$ , Data Scientist at the City of Edmonton:

“There’s a ton of potential for leveraging AI and data and analytics in municipal government. I think it has so much potential to improve services and increase efficiencies and ultimately get citizens greater value for their hard-earned tax dollars.”

When Gready speaks of AI, and how Edmonton is applying the tactic, data and analytics are always front and center. This is because AI models are essentially complex data analysis, based on algorithms and inputs and outputs.

Because governments typically sit on a large amount of data, many of them are already in a good place to start building an AI framework. Says Gready:

“A lot of organizations already have analysts and other data professionals that can begin the process of putting foundational elements to lead up to a more advanced kind of AI and analytics.”

So how do you build a foundation for AI? Just follow Edmonton’s lead:



1. Start with quality data

To run AI algorithms, data must be robust and recent. It’s also important to understand the context around it — what biases may affect it and what gaps there may be.

2. Use strong data management processes

It doesn’t matter how good your data is if you can’t access it. Says Gready:

“At the city of Edmonton, we have an internal data portal that we can use to organize and give access to analysts and data scientists at the city.”

3. Promote data literacy within the organization

It’s critical to educate others in the organization about what data is available, and how it can be used. In addition to joining regular meetings with business partners when rolling out an AI project, Gready and his team host a large data & analytics meeting every two months or so.

“It’s just a place for people that are interested in data and analytics. There are lots of analysts who are embedded in different business areas who come together.”

4. Identify invested business partners

The more people in other departments across the organization understand the power of data and AI, the more likely they are to embrace it when tackling their own projects. In Edmonton, Gready says:

“Through that data literacy and education process, we’ve tried to get different business areas interested and they are able to see the potential.”

5. It’s best if leadership is on board

Using AI often requires a lot of change management, and a degree of risk. The more understanding and support coming from the top of the organization, the more likely projects are to succeed. Gready notes:

“We’ve been in the lucky position of having leadership from the very top, including the mayor and city manager, who have seen the potential for this sort of technology to help transform the city.”

By making sure its data was high quality and accessible, and by educating and advocating for its use, Edmonton created a framework that supports advanced analytics — including AI. Once this was done, AI could be applied across the agency. And that’s exactly what’s being done.

An AI take on safety codes

Gready’s team has applied this framework to a range of city departments, from fire rescue to pest management to building safety codes and inspections.

The safety codes and inspection efficiency project, or SCIE, uses AI to predict whether home builders will pass low-risk inspections.

To start, Gready partnered with $ Juan Monterrosa$ , Director of Codes and Inspections at the City of Edmonton, to dig through ten years’ worth of safety inspection data. Says Gready:

“With all of the inspections that come in for these inspectors, some of them are very high risk, and some of them are really routine, low risk inspections. So the idea is to leverage AI to help inspectors move their focus to the higher risk inspections. It takes the focus away from the lower risk inspections that they’re just turning up, having a look around, and then passing.”

In other words, some of the inspections sucked up a lot of the inspectors’ time and efforts, even though they almost always passed and didn’t carry a huge risk if they didn’t pass.

Instead, the AI model uses data to analyze factors like the type of inspection, location, and level of risk, to automatically pass certain inspections without the need for an in-person visit. But the technology is also only part of the equation. Monterrosa notes that even with the data powering the program, the city built in physical audits to ensure the AI was working correctly:

“We started to feel comfortable along the lines of, how can we provide a program where builders don’t get inspected every time, but we provide an audit function to make sure we go out and check every 20th one.”

This audit validates that the AI is working. It also facilitates the change management and stakeholder education process, which Monterrosa and Gready say is a big part of any AI project.

Now, not every inspection is run through the model — Monterrosa estimates it’s about 25 to 30 percent of the City’s overall inspections. But of those that go through, about 40 percent of them pass using AI. This equates to a significant reduction in manpower, and allows Edmonton inspectors to focus their efforts on the cases that carry a higher risk to citizens.

What’s more, when Edmonton’s public servants had to shift to a remote working model recently, the inspection team was already primed to rethink the way they work. Says Monterrosa:

“Once we were all sent home, we couldn't do inspections in person anymore. So the team that originally led the AI portion that went through and just did it, using different types of apps, whether it was FaceTime, Google Duo, Google Hangouts, and doing inspections through phones, coordinating with homeowners, and with builders.”

Having AI as a tool in their kit helped Monterrosa and his team stay openminded to new technology, and adapt to changing needs quickly.

Common AI misconceptions

While the concept of using artificial intelligence is fairly well-known, there are a lot of misconceptions about how it actually works. Gready says the following myths are common among those new to AI:

AI will result in job loss

If an algorithm is making decisions or eliminating the need for services that a person previously did, won’t it put that person’s job at risk? Gready explains that’s usually not what happens:

“We worked really hard with the frontline staff to educate them about what this was and the fact that it’s not designed to take everybody’s jobs. It’s designed to help them deal with the incredible volume of work they’re trying to do.”

Public servants are too often faced with an incredible amount of work — AI is meant to ease the load and let workers focus on the tasks and situations that require a person to think or act. In Edmonton’s case, there are still many inspections that can’t be passed via AI; now workers can give these the time and attention they require.

AI is fair by design

Computers don’t have human bias, so decisions made by computers are inherently unbiased, right? Wrong. Data often comes with historical biases, and if the AI model isn’t designed with these in mind, the biases will work their way into decision making.

For Edmonton’s safety codes and inspection project, the team was concerned certain inspection data, such as a neighborhood that happened to have a few bad contractors, could skew the outputs and disadvantage that neighborhood. Says Gready:

“We were very conscious about actually taking out some of the attributes because we didn’t want to accidentally bias the model. It can be fair, as long as it’s designed with a lot of thought.”

AI is the silver bullet solution

As is the case with most technology, AI should never be viewed as the magic cure-all for a problem. It’s a tool that is dependent on a lot of other factors. Gready explains:

“The truth of the matter is that AI is very helpful, but it’s only as good as the data that it is fed, and it’s only as good as the design of the project and how it’s rolled out.”

In fact, just like every other innovation, AI will fail. It’s part of the process. Gready explains:

“Many projects will fail when you’re rolling out AI algorithms. The key is to identify early that it’s going to fail and then move on and not go down a road that’s not going to work out.”

Gready says Edmonton has had a number of projects where the team tried to apply AI over the course of a few weeks, and it just didn’t work:

“We looked at the data, worked with the business area, and eventually we just came to the conclusion that there's not an opportunity here in this case.”

Edmonton’s broader use of data

One of the reasons Edmonton is so successful in its application of AI, is the city’s overall use of data when driving innovation.

In 2019, the city received a Willis Award for Innovation from the Canadian Association of Municipal Administrators for its $ urban Primary Land and Vegetation Inventory (uPLVI)$ . The inventory created a data portal that details vegetation and tree canopy around the city. This is then used to drive urban planning and development projects. The project was so successful that Edmonton now provides versions of the data to other local governments and researchers.

Data and analytics have also played a role in $ monitoring the performance of a new electric bus program$  and $ piloting an adaptive street light program$ . Data even helped the city $ house 4,000 people experiencing homelessness $ by tracking resources and support services for the area’s homeless population.

Gready expects that this reliance on data is only going to increase:

“I suspect that going forward, data and analytics and AI are going to be more important than they were before when we think about reimagining the city of Edmonton. And I think it’s going to be a common theme for many municipalities.”

As local governments grapple with challenges like tighter budgets, reduced workforce, and the need to digitize citizen services and internal operations, data is one of the most powerful tools at their disposal. By building a foundation of solid data, analytics resources and a framework for AI application, like Edmonton did, cities can make technology carry some of their load, and make smarter decisions for their citizens.