Ottawa's new artificial intelligence registry promises to track how federal agencies use and test artificial intelligence systems.
Organizing labels to classify insects found on farms (or in food); scanning the parcel for the presence of a gun; calculating the best way to provide medical care to astronauts in space.
“Artificial intelligence is changing governments.”
These are just some of the more than 400 systems in 42 departments where the Canadian government is either exploring or currently using AI in its operations, according to a new public registry released Friday.
The register can be viewed on the Canadian website. Open Government website. It includes searchable categories such as the name of the government department using the system, what each AI system is used for, and which vendor developed the program. The registry only includes well-known artificial intelligence projects, but does not use programs that the government considers “low risk,” such as spell checkers or virtual assistants.
In the case of a system used to scan packages for weapons, the registry shows that the Canada Border Services Agency (CBSA) is currently testing a tool from Israeli software company SeeTrue.
The government says the registry was released as an early version, containing only basic search functionality, as part of a commitment made under Canada's federal AI strategy to provide a more transparent view of AI activities in the federal workforce. The government says it will be updated and refined to include more details and features in future versions based on public feedback next year.
“Artificial intelligence is transforming governments, and we are committed to providing Canadians with information about how it is used to support programs and services. The AI registry is an important step in strengthening public trust and ensuring the responsible use of AI in the federal public service,” Treasury Board President Shafqat Ali said in an emailed press release Friday.
Arvind Gupta, professor of computer science at the University of Toronto and member of the Ottawa AI Strategy Task Forcesaid in an email to BetaKit that at first glance, many of the registry's elements appear to be tools built in-house and designed to be used fairly simply, but without knowing more about the architecture, training sets, and testing, it's hard to tell whether AI best practices are being followed.
He also hopes to see more Canadian SMEs included on the list in the future. American companies Microsoft and OpenAI are mentioned in the document 75 and 15 times, respectively.
“When I first heard about this list, I hoped it was the start of a new government procurement strategy,” Gupta said. Gupta pointed to the Small Business Innovation Research (SBIR) model in the US, which requires each department to allocate 2 percent of its discretionary budget to procuring new solutions from small and medium-sized enterprises. (Implementation of the American model is currently on hold pending Congressional funding.)
“Given how many [Canada] spends on [Scientific Research and Experimental Development] …we could redirect some of these funds to a large and effective SBIR-inspired program and encourage SMEs to consider selling to the government. Maybe the AI list really is the beginning of this kind of thinking, maybe not,” Gupta said.
According to a government press release, artificial intelligence has been used by the Canadian government since 1994. The registry shows that the first use, which is still ongoing, was by CBSA for a “fuzzy search” software tool that flags high-risk travelers by searching names in CBSA systems.
The circular on the registry website itself mentions that it was created using machine translation, but the results were verified by experts.
Image provided Unsplash. Photo by Nick Linnen.
