The federal government’s AI Strategy for the Federal Public Service exists to guide public sector adoption of AI tools. The idea is that artificial intelligence can increase the efficiency and effectiveness of government operations. I certainly won’t fight that assumption.
However, I’m not convinced that all that extra efficiency and effectiveness will necessarily translate into universally better outcomes for regular Canadians. Sure, we’d all like to get our new passports or resolve mix-ups over benefits eligibility faster. But there might be some government functions where more doesn’t necessarily equal better.
First though, let’s take a look at the registry of existing federal government AI projects that accompanies the strategy. The registry currently contains 412 entries.
356 (86 percent) of the projects primarily target internal Government of Canada users, 25 are meant for both employees and members of the public, 23 are just for the public, and 6 - intriguingly - are for “Neither employees nor public”.1
180 of the projects were developed internally by government employees, 156 were purchased from external vendors, and 25 came from open source resources.
Microsoft and Microsoft Azure were, between them, responsible for 37 of the tools that came from external vendors, with OpenAI contributing 4 projects, and Thomson Reuters, 3.
88 of the projects involve the use of personal information. Data for three of those came from deidentified datasets from health records.
Canada School of Public Service has the most entries in the registry at 31. They’re followed by Innovation, Science and Economic Development Canada (26), Statistics Canada (25), National Research Council (25), and Global Affairs Canada (21).
There’s no question that certain operations will noticeably improve as departments build out their integrations. Expect to see more open consultations where Canadians are invited to share their thoughts about various policy questions. The fact that AI can both accurately summarize and classify millions of unformatted text (or audio) responses means that your opinions can be reliably represented and communicated to policymakers.
Internally, governments will be able to produce limitless impact assessments, regulatory impact analysis statements, and guidance documents. Played right, this could counteract suffocating regulatory review requirements - although it might just as easily trigger waves of new regulations.
Even if, ultimately, final decisions would still require human review for legal compliance, AI tools could still speed up processing for immigration, tax, or access to information cases. In the overall scheme of things, those sound like potential wins for Canadians.
The flip side to that of course, is that governments will likely improve the most where they’re already engaged. Those will almost certainly include tax fraud enforcement. If someone’s worried about being caught for tax avoidance now, expect the stress level to dial up to at least eleven once an AI-powered CRA kicks into gear.
But for those of us who - however reluctantly - reliably dig deep and pay our taxes each year, there might still be something just as dark waiting for us.
Consider how AI tools make facial recognition (from official archives, public webcams, and online videos and images) and inter-agency data sharing a trivial task. Now consider how such data integrations can facilitate 24/7/365 digital tracking of entire populations.
With all that data already in place, why would a government not take its obsession with online offensive behavior and “disinformation” to the next level? And if that government has already demonstrated its passion for “social” collaboration with elements of Chinese Communist Party security apparatus, why not quietly create some local version of a social credit scoring system?
Perhaps those are unrealistic fears. But without any comprehensive privacy legislation restricting the government’s use of AI in Canada, you’d be hard pressed to convince me that they’re completely unfounded.
Who Really Steers Federal Policy? Spoiler alert: it's the Privy Council Office
I’ll admit that I’ve rarely given much thought to the Privy Council Office (PCO). In fact, I was never all that clear about what it was, what it did, and what relationship it had with the Clerk of the Privy Council or the King’s Privy Council for Canada
Online Harms Legislation, the Tilting at Windmills Edition
Conservative private member’s Bill C-412 “An Act to enact the Protection of Minors in the Digital Age Act and to amend the Criminal Code” is an attempt to thread the needle between online safety for children and everyone else’s fundamental privacy rights. The bill certainly avoids the worst abuses of the Liberal government’s own Bill C-63. But I don’t b…
Bill C-4: Join a Political Party in Canada and Mess Up Your Life?
That “other measure” from “BILL C-4 An Act respecting certain affordability measures for Canadians and another measure” is bizarre. Ok. Bizarre isn’t quite strong enough. I’ll let you come up with something better after reading this post.
This is obviously yet more evidence for secret government cooperation with space aliens. Either that, or the results are just transferred directly to the appropriate department within the Chinese Communist Party.
I just loged in. equitynation.ca your comments or questions welcome
equitynation.ca make Canada the first