Having access to AI is unlikely to be much of a competitive advantage for very long.

The technology is moving too quickly for that. Models are getting better, cheaper and more accessible. Competition between providers is pushing prices down. What once required significant investment is becoming something that more organizations can access with relatively little effort.

We can already see this happening in China. Alibaba’s Qwen family of open models recently surpassed 3 billion downloads globally, with more than 460 models released and more than 300,000 derivatives built around them. Capable AI models are becoming cheaper and easier to access at remarkable speed.

AI will still be incredibly important. But as access becomes easier and cheaper, the focus will shift from having the technology to figuring out how to use it to actually change the way we work.

We have seen this before with other technologies. At some point, access to cloud computing stopped being particularly interesting. The advantage came from what organizations built with it and how they changed the way they operated.

In other words, the organizations that get the most out of AI will be the ones that figure out how to incorporate it to transform the way work gets done.

The operating model matters

AI touches nearly everything about how an organization works, and other jurisdictions are already adapting:

  • People: how roles change, what skills are needed and where humans add the most value. Australia is requiring foundational AI training for public servants and establishing Chief AI Officers across agencies.
  • Process: what can be automated, simplified or eliminated. The UK’s Office for National Statistics is using AI to automate occupational classification, saving hundreds of hours of work and allowing staff to focus more on quality checks.
  • Service delivery: how AI can change the way services are delivered. The UK’s AI Exemplars program is testing AI across areas including healthcare, education, planning and justice, using a test-and-learn approach.
  • Technology: how AI fits with existing systems, architecture and infrastructure. Australia is developing GovAI as a shared government service, providing secure access to generative AI alongside the infrastructure and technical capabilities needed to support it.
  • Data and performance: how we use information and measure whether AI is actually improving outcomes. Singapore has made AI a strategic priority and is driving adoption across government to improve operational efficiency and service delivery.
  • Security: how we protect systems and information as AI becomes embedded in government. Australia’s GovAI is being developed as a secure, APS-only environment for public servants to use AI with government information.
  • Governance: how we manage risk, accountability, privacy and responsible use. Australia is establishing an AI Review Committee to provide whole-of-government oversight and review high-risk AI use cases.

The examples are different, but the broader point is the same: AI adoption is not just a technology decision. It touches the way the organization operates.

This is true of any major technology shift, but it is even more relevant with AI because of how directly it can change the way work gets done.

The risk is that we use AI to make existing processes more efficient without asking whether those processes still make sense.

Canada has already recognized the opportunity

The Government of Canada has its 2025–2027 AI Strategy for the Federal Public Service and is starting to put some of the pieces around it in place. The strategy covers AI capability, better services, responsible use and workforce readiness.

But a strategy can only take you so far.

The GC has a lot of history embedded in the way it operates. Processes, policies, systems and organizational structures have accumulated over decades. Some are necessary. Others are simply difficult to change.

AI gives us a reason to question some of them. For example:

  • If we can automate a process, why are we still doing it the way we did before AI?
  • If AI can reduce a day’s worth of work to an hour, what do we want that employee doing with the time we’ve freed up?
  • If a service can be delivered differently because AI can handle some of the work behind the scenes, should we continue designing the service around the old process?

These are much harder questions than “where can we use AI?”

Canada also doesn’t have to figure it all out alone. Australia, Estonia, Singapore and the UK are already experimenting with different approaches to AI-enabled government. We don’t need to copy them, but we should learn from what they’re doing.

The people issue

There is also a people issue that I think we need to be more honest about.

AI is going to change jobs. Some tasks will disappear. Some roles will change. Some work will become much more productive. I don’t think we do anyone a favour by pretending otherwise.

That also changes the skills we need. If AI takes on more of the routine work, being able to assess an answer, challenge it and know when it is wrong becomes more important.

The GC also has an opportunity to automate some of the work that consumes a lot of people’s time without necessarily creating much value. That could give employees more time for the things that still benefit from human judgment: solving problems, working with citizens, making decisions and dealing with situations that don’t fit neatly into a process.

Experimenting at the edge

Getting there will require some willingness to experiment and make it easier for people to try things.

A team shouldn’t have to solve every enterprise-level question before it can test a relatively small idea. The UK’s AI Exemplars program is one example of a government deliberately using a test-and-learn approach, with teams testing real use cases before scaling what works.

Let people experiment, see what works and build from there.

That also means accepting that some experiments won’t work. I suspect this may be one of the harder cultural changes for the GC. We are generally very good at managing risk once we understand it. AI requires us to make decisions while some of the risks and opportunities are still evolving. Waiting for everything to become clear before acting may turn out to be its own risk.

The GC has already recognized that AI is going to be an important part of the future of the public service. The next step is making sure we don’t simply add AI to the way government works today, but question whether government should work differently because AI now exists.

AI is here to stay. The organizations that benefit most will be the ones willing to change how they work because of it.

Written by Srivatsan Vijayakumar, Founder and Managing Director, SV Strategy Services.