Canadian government departments are balancing AI optimism with caution

By on 01/04/2024 | Updated on 01/04/2024
Image by

Artificial intelligence (AI) has the potential to improve services for the public and transform the way civil servants work. During a recent Global Government Forum webinar, experts from government organisations in Canada discussed the critical importance of trust when introducing AI.

Canada’s Benefits Delivery Modernization programme has been described as “the largest information technology project undertaken to date by the federal government”. It aims to redesign the way three federal statutory benefits are delivered – Employment Insurance, Old Age Security and Canada Pension Plan – and includes rethinking all user channels and touchpoints, whether by phone, online or by post.

Putting users at the centre is a key operating principle of the initiative, said Aaron Jaffery, chief product owner and architect for the Benefits Delivery Modernization programme, Service Canada, and this “client centricity” also applies to the use of emerging technologies such as AI.

Aaron Jaffery

“We’re seeing the emergence of potentially the most disruptive technology since the Industrial Age,” he said. “We’re looking at bureaucratic systems that were automated for large data entry from the 1950s through computerisation and now we’re applying the concepts of human-like intelligence onto our data in mass bureaucratic systems in a period of ever-decreasing trust in our public institutions.

“This is a big challenge for all public servants, and particularly in a benefit delivery modernisation programme.”

He is taking an “optimistic but cautious” approach.

“We’re certainly very open to and indeed embracing the way that artificial intelligence can improve the public service delivery of benefits,” Jaffery said.

This could include supporting someone who has become unemployed to understand the help available and guide them during the process, through to aiding a service officer who is making a decision on a claim.

He commented: “Our hypothesis is that if we deliver a better client experience, if we provide more transparency, if we enable clients to understand better how decisions are made and understand their journey and get the outcome that they are entitled to, that will have a positive impact on the trust that people have within services and within the systems.”

Sign up to receive Global Government Forum’s monthly AI newsletter

Challenge-winning chatbot

The webinar also featured three speakers from Agriculture and Agri-Food Canada (AAFC) who were part of the team that won last year’s Public Service Data Challenge. The challenge is organised by Global Government Forum in partnership with Natural Resources Canada and Statistics Canada.

Jay Conte, policy analyst, labour and market efficiency policy, Agriculture and Agri-Food Canada, came up with the winning idea to use generative AI to help users find information and resources more easily on the AgPal Program and Service Finder website, which provides information for agricultural businesses on government programmes and services.

Steve Rennie

The project focused on putting the customer “front of mind”, he said. “How can we make this a really pleasurable experience for somebody to get the information that they need, rather than a very challenging experience?”

The team has learned several things which are applicable to future AI-related projects.

One is the importance of collaboration. Nicole Johnson, senior data analyst, results and data analytics, at Agriculture and Agri-Food Canada, described how the Data Challenge team worked closely with colleagues in the AgPal business unit as well as running two hackathons which brought together multidisciplinary teams from across the department with industry representatives, experts from Algonquin College’s AI programme, and AAFC’s chief data officer.

Partnering with the college during this work also seeded the idea for using ‘talent pipelines’ more widely.

“We also realised early on that while our team had a lot of capability, we certainly didn’t have the ability to take on all the work that may come our way,” said Steve Rennie, director of data-driven technologies at AAFC, which is working to advance the responsible and ethical use of AI across the department in a “coordinated fashion rather than a siloed way”.

This has led to collaboration with colleges and universities across the country where students get the chance to work with real data and use cases from the AAFC department – typically problems that need to be solved but can get sidelined due to time constraints. The students receive mentorship and guidance from department staff.

“It also allows us to identify talented students that may be interested in joining the public service, which really helps us to diversify our workforce,” said Rennie.

Read more: Canadian public servants issued guidelines for using generative AI

‘AI in a box’

The team also plays a ‘connector’ role and has been engaging closely with around 80 departments across the government of Canada to understand the stage they’re at with AI and facilitate knowledge-sharing.

“We’re really of the view that to scale artificial intelligence across the government of Canada, we have to work together,” said Rennie. “It doesn’t make sense for every department to work in a silo and develop their own thing in isolation.”

Nicole Johnson

He said that departments sharing their code and their documentation contributes to an ‘AI in a box’ resource that can be made available across the public service.

“That really will help raise the level of ability in artificial intelligence, but it also helps departments that are very early in their AI journey to accelerate much more quickly,” he added.

Rennie said AAFC is committed to going beyond “the bare minimum” on ethics. For example, although chatbots don’t fall within Canada’s directive on automated decision-making systems, its principles were still followed for the development of the AgPal chatbot.

“We took the view that somebody getting information from a government chatbot is likely to make a decision based on that information,” he said. “And for that reason, we decided to make sure that we followed every step in that directive, so doing impact assessments and making sure that we were doing things the proper way. I am really excited to see where it’s going.”

The AgPal chatbot has also been developed to mitigate against the risk of ‘hallucinations’ by connecting it directly to the data source rather than allowing the tool to go out to the internet or the large language model to formulate answers.

Johnson also highlighted the importance of keeping “a human in the loop”.

“We did tons of user testing to see what responses came from the chatbot, but then even over time as it progresses [it’s] keeping that testing component within that process,” she said.

‘Transformative experience’

Asked where the main potential of AI lies for governments, Jaffery believes it will materially impact both the way services are delivered and the work of civil servants.

Jay Conte

“I think the skills pipeline, the training and the consideration about the way we will work in the future and the different skills that we all will need are going to be fundamentally important,” he commented.

Conte believes that governments have an opportunity to improve the user experience and make it more in line with what people have come to expect from private sector companies such as online retailers.

“I think Canadians and citizens of other countries will notice that change very rapidly,” he said. “If I go to a government site [and] I need information about a programme, I may not even know that the programme exists, I just know that I have a problem and that I need some help. If I can describe my own problem in two or three sentences and within 30 seconds, a chatbot gives me the exact resource that I need – that’s a transformative experience in terms of what you’re getting as a customer. And I think that there’s a real opportunity for us to do that.”

Watch the full webinar for further discussions on:

  • Infrastructure modernisation requirements to harness the full power of AI
  • The creation of AI guidelines and strategies
  • AI governance structures
  • Return on investment for AI
  • And much more

Join Global Government Forum’s LinkedIn group to keep up to date with all the insight public and civil servants need to know.

About Sarah Wray

Leave a Reply

Your email address will not be published. Required fields are marked *