The federal government will work to prevent artificial intelligence from discriminating against people applying for jobs in federal government departments, says Treasury Board President Anita Anand.
In a wide-ranging year-end interview with CBC News, Anand acknowledged concerns about the use of AI tools in hiring.
“There is no question that at all times, a person’s privacy needs to be respected in accordance with privacy laws, and that our hiring practices must be non-discriminatory and must be embedded with a sense of equality,” Anand said when asked about the government’s use of AI in its hiring process.
“Certainly, as a racialized woman, I feel this very deeply … We need to ensure that any use of AI in the workplace … has to be compliant with existing law and has to be able to stand the moral test of being non-discriminatory.”
Some government departments have started incorporating AI tools into their hiring processes. In its response to a question posed by NDP MP Matthew Green, the Department of National Defence said it used Knockri, an AI-driven skills assessment tool, in hundreds of pre-recorded job interviews to “reduce bias and promote equity in the recruitment process.”
“Knockri uses machine learning to analyze the speech-to-text content of candidate responses and compares them against determined performance indicators,” the department wrote.
Anand spoke to CBC News less than a month after an investigation by Radio Canada found that several government departments had purchased technology that can be used to extract data from cellphones or computers issued to employees.
Anand rejected any suggestion the government was using spyware on its own employees.
“This is only used in cases where there is an investigation or a belief that something’s going wrong,” Anand said. “It’s not used routinely.”
Government guidelines for AI
In September, Anand launched new guidelines for public servants who want to use generative artificial intelligence tools like ChatGPT on the job. At the time, she said the government will be monitoring the way AI is used to guard against potential problems like bias or discrimination.
That didn’t save Global Affairs from having to apologize last week after it posted an AI-generated picture on a social media account to represent an Inuit woman.
Anand said Global Affairs could have used the tools Treasury Board has developed to guide employees and departments, such as the AI guidelines or the Algorithmic Impact Assessment (AIA) process.
“This is a perfect situation where one might use that algorithmic impact assessment tool that I mentioned,” she said. “The AIA tool is a series of questions, and those series of questions should be asked when AI is being used in the workplace. They reflect legal and policy and ethical considerations.”
Anand said she also wants to fill gaps in Canada’s regulations on the long-term use of AI.
“There’s no question AI is a transformative technology,” she said. “Another thing that I am looking at is really the long term, because it requires us to ask how we think about regulation more broadly.
“For example, with the advent of autonomous machinery, such as autonomous vehicles, we need to ask ourselves whether the regulation that we have in place is relevant and is applicable in the long term.”
Anand said that, even as the federal government explores the use of AI-driven tools, Canadians should always have the option of speaking with an actual public servant.
“At all times there should be a telephone number available so that you can speak with a live person,” she said. “That is fundamentally important. I do not expect that individual workers should be in any way impacted from employment. We need strong talent.”
The federal government is moving to adopt a framework for the use of AI in federally regulated companies through Bill C-27, currently before committee. Anand was noncommittal when asked whether the legislation should apply also to federal government departments and Crown corporations.
“The intention is to ensure, from a Treasury Board perspective, that we are also in tune with the need for guidelines relating to artificial intelligence. And of course, Bill C-27 does that for the broader economy,” she said. “But as we know, many businesses look to the government of Canada for best practices and the area of artificial intelligence is no exception.”
One of the biggest projects on Anand’s plate in the coming year is a major effort to overhaul the aging technology the government uses to deliver such benefits as Old Age Security (OAS), the Canada Pension Plan and employment insurance (EI).
In June, the government made the first transfer of 600,000 foreign OAS recipients to the new platform, said Anand. In the coming year, she said, the government plans to launch phase two of the program with improvements to the delivery of OAS and EI payments.
“Research shows that there’s a direct correlation between the confidence that citizens have in our government (and) the services that they receive,” said Anand. “So we expect to make tangible progress on the replacement of 45-plus year old systems.”