Can we leverage AI to help tackle housing crisis?

The answer to solving Canada's housing crisis may come from removing people from the process.

Red tape, inefficiencies, profit maximization on luxury builds - there is a long list of reasons why Canada's housing crisis will not be easily fixed.

But if the country has any possibility of hitting the Canada Mortgage and Housing Corporation's target of 3.5 million new homes built by 2030, it's going to need to get more efficient at every stage of the process.

That's where artificial intelligence comes in, says Jason Cassidy, president of Shinydocs, a Kitchener-based company that focuses on leveraging data to make better decisions.

"We're not trying to build the Star Trek bridge computer that can answer any question. We're just trying to make something that is manual, slow and human-driven just a little bit faster," said Cassidy.

Consider the process of issuing building permits.

Each municipality has its unique bylaws regulating the permitting process, with a team of employees that need to read, research and approve each submission. It can be a labour-intensive process that results in builders waiting months on the sideline before getting their approvals.

There are ways to use technology to improve this process, said Cassidy.

Kelowna, B.C., has partnered with Microsoft to become one of the first municipalities in the world to use AI to speed up its building permit process by up to 30 per cent.

The project includes an information bot - similar to Chat GPT - that can answer questions about the building process before filling out an application. When it comes time to fill it in, an AI assistant shadows the applicant and tells them in real time whether they are compliant or not, and what areas need addressing.

Once submitted, the city will specifically tag all submissions that have used the AI help. Since officials know those submissions are already compliant, they can quickly review and approve those first.

Not only will wait times for builders be reduced, the system will also free up staff who currently spend most of their day answering customer questions or explaining errors on applications.

Beyond municipal requirements, there are other steps in the building process where AI can dramatically cut down time, said Cassidy.

It can make initial building designs in minutes that meet building codes, and immediately offer a variety of alternatives depending on customers' wants. During construction, AI can be used in robotics to automate processes, reducing the number of workers needed while also reducing project time and costs.

And for urban planning, AI can quickly leverage data to create master building plans that address affordability and effectively use the different types of mixed housing needed in communities across Canada.

Picture an AI software that understands population dynamics, socio-economic factors, building codes and other possible data that tells the story of a community, and then instantly develops a plan that incorporates price constraints and geographical considerations.

The name of the game here is data, said Cassidy. The more data you have, the better the AI modelling.

And for those worried about "AI stealing our jobs," Cassidy said it is important to get some perspective on the enormity of the housing problem.

"The idea that AI is going to solve this, and then we are going to have nothing to do, is absolutely just silly to me," he said.

"We are behind on everything, from planning to permit development to deciding whether or not we're going to build on these protected lands or those lands. There's so many decisions we need to make that AI can help with."

One of the things Shinydocs focuses on is looking at the historical data of a company and using that to model what to do next.

Cassidy doesn't think that's what Canada's housing problem needs.

"It's actually the exact thing we shouldn't do. Our historical data got us to where we are now. We overbuilt expensive homes, we overlent, we got ourselves into debt and we overengineered our neighbourhoods to make them less nice and less safe."

Just getting somebody to look at the data and make predictions will just create a master class on what shouldn't be done, Cassidy said.

His suggestion to different levels of government that are interested in using new technologies is to first establish how they are operating now.

"Computers don't work like people; they don't get into Microsoft Teams meetings and ask where something is and have different members of the group go find them. They want all the information accessible, and they want to make decisions immediately," he said.

It comes down to three main questions: how does the organization now make decisions? Where is the data? Is it in a format that computers can consume right now?

"If it's not, well, maybe that's something you need to work on if you actually want to start feeding it into AI models and make some decisions," he said.

Robert Williams is a

Waterloo Region-based

reporter for The Record. Reach him via email:

robertwilliams@torstar.ca