Driving forward: navigating the AI landscape in the Government of Canada

By Stephen Burt, Chief Data Officer, Government of Canada?

While the ideas for this post originated in my office, please note that ChatGPT was used to help edit and generate part of the text.? This blog post is intended for federal institutions and public servants interested in exploring how artificial intelligence (AI) can be used as a tool to support and enhance their operations or daily work.?

Introduction

Automated tools and solutions like generative artificial intelligence, robotic process automation and predictive analytics are increasingly capturing our collective imagination with their transformative potential.? Generative AI in particular has created excitement across the Government of Canada (GC) about AI’s potential to enhance services and increase government efficiency.? This is accompanied by prudent concerns about AI’s implications.??

Canadians are coming to expect AI-enhanced services that are frictionless, immediate and personalized, but they also express concerns about the risks of AI, and the privacy and security of their personal information.? To meet Canadians’ expectations of government, it is crucial that we explore and showcase the benefits of AI technologies, while also mitigating the risks that could hinder our success.??

The AI foundation: strong policies?

AI integration into government operations will significantly shift how we work and deliver services.? However, without a solid policy framework, the rapid adoption of AI technologies can lead to unintended consequences.? Strong policies serve as a safety mechanism for AI in government.? They empower us to drive forward with confidence, knowing that we have the necessary controls in place to navigate the challenges that can arise.?

Recognizing this need for a policy framework, the Treasury Board of Canada Secretariat (TBS) issued the Directive on Automated Decision-Making in 2019.? The first of its kind globally, this directive sets mandatory requirements for the use of automation—including AI—within federal service delivery, ensuring transparency, accountability and fairness.? This directive is a testament to our commitment to safe and responsible AI use.??

Following this directive, the publication of the Guide on the use of generative AI in September?2023 marked a significant step in refining our policy approach to newer AI technologies.? The FASTER principles (Fair, Accountable, Secure, Transparent, Educated, Relevant) outlined in the guide ensure that generative AI use and deployment are thoughtfully considered, balancing innovation with risk management.??

Complementing these efforts, Canada actively participated throughout 2023 in the development and adoption of twelve guiding principles for the use of AI at the Digital Nations forum.? These principles guide our overall approach to AI and are crucial for maintaining public trust and ensuring an ethical approach. I encourage you to review these principles and consider how they can be applied in your organization.??

AI: building momentum?

With our foundational policies firmly established, it’s time to accelerate the federal government’s AI adoption.? The call to action in the Guide on the use of generative AI is clear: federal institutions should explore AI tools to support and enhance their operations.? As you embark on this journey, let the FASTER principles and the Guiding principles for the use of AI steer your explorations, helping you to identify and mitigate risks effectively.?

The AI landscape has matured significantly over the past few years, offering us pre-trained models, powerful secure application programming interfaces (APIs), and expansive AI capabilities through cloud computing.? These tools are at our disposal to enhance service delivery and improve operational efficiencies.??

As we adopt these tools, however, security and privacy need to be the cornerstone of our approach.? While some APIs may offer robust security and privacy features, public tools available on the internet do not meet the same standards.? It is imperative that institutions prioritize access to secure, verified AI tools.? Successful examples include SSC’s CANChat and NRC’s AI Zone, which provide tailored, secure environments for AI exploration.? Moving forward, chief information officers (CIOs) must continue to prioritize secure solutions that meet employee and program needs.?

The success of AI initiatives hinges not just on the technologies themselves but also on the quality of the data that fuels them.? Not only does the right data need to be available, but it also needs to be curated to remove inappropriate biases and ensure that it is representative of the population being served.??

While perfection in data is often beyond our reach, as a government we have made a dedicated effort to understand and improve the quality of the data we steward to achieve significantly improved results.? Chief data officers (CDOs) have an important leadership role regarding institutional data holdings and the AI tools necessary to extract its value.? They should be at the forefront of the discussions about AI adoption to ensure that your data is robust, well governed and ready to fuel your AI applications.?

We also recognize that AI maturity levels vary across the government.? Many of our organizations are well staffed with data experts, and our scientists and statisticians regularly use AI to process large volumes of data, develop useful technologies and reveal new data-driven insights.? These experienced colleagues are a great resource and can provide leadership as we build capacity across the GC.???

Planning the destination?

As we embark on this journey, we need to chart a clear path forward.? For this reason, TBS’s Office of the Chief Information Officer is leading work to develop a comprehensive AI strategy for the federal public service that will align with our public service values and ethics and lay out a vision for AI adoption.??

In May, we brought together leading AI experts from Canada’s national AI research institutes and academia for a ministerial round table to help shape the strategy.? This was only the first in a series of engagements to be held over the coming months with a wide range of important stakeholder groups.?

As we refine the AI strategy, we will also engage GC stakeholders from technology and data leadership forums and specific domains, such as science, human resources, service delivery and national security.? The unique challenges and opportunities in these sectors demand tailored approaches to AI integration.? It is critical that each sector not only prepares for, but also actively shapes, how AI will transform their operations.?

A human-centred approach?

As we continue down the path to greater AI adoption, our clients and our workforce must be at the heart of our efforts to use these technologies.? We need to prioritize AI deployments that build trust and enhance services to Canadians and other clients, while also ensuring that they address the every day challenges faced by our employees.?

We need to equip our workforce and leadership with the skills and tools to deliver the best possible outcomes for Canadians.? Increasing AI literacy must be a key priority for all public servants.? Understanding and leveraging AI should become as fundamental as using a computer or navigating the internet.? If you haven’t yet engaged with AI tools, I strongly recommend the DDN321 course on Using generative AI in the GC from the Canada School of Public Service.?

We need to collaborate across policy divides to ensure that the adoption of AI is effective and safe.? Program and corporate services experts need to identify the problems that can benefit from innovative AI solutions to address user and organizational needs.??

CIOs need to provide access to modern software and systems, while CDOs work with policy and program teams to complement their operational expertise with the data science that can help solve their problems and the data governance that will ensure high quality, unbiased data to fuel AI models.? Cybersecurity teams must make sure that the use of these tools is secure, and privacy officials must ensure the protection of the personal information of those we serve.?

The road ahead is both exciting and daunting.? AI’s potential to enhance how we serve Canadians is immense, yet it requires thoughtful handling to truly benefit the public good.? Let’s be sure to embrace the transformative power of AI along with the responsible approach it demands.? Let’s drive towards a future where AI supports and enhances the government’s ability to serve the interests of Canada and its citizens.?