Forgot your password?
Close
typodupeerror
Canada AI Government

Does Canada Need Nationalized, Public AI? (schneier.com) 108

Posted by EditorDavid from the what-if dept.
While AI CEOs worry governments might nationalize AI, others are advocating for something similar. Canadian security professional Bruce Schneier and Harvard data scientist Nathan Sanders published this call to action in Canada's most widely-read newspaper (with a readership over 6 million): "Canada Needs Nationalized, Public AI." While there are Canadian AI companies, they remain for-profit enterprises, their interests not necessarily aligned with our collective good. The only real alternative is to be bold and invest in a wholly Canadian public AI: an AI model built and funded by Canada for Canadians, as public infrastructure. This would give Canadians access to the myriad of benefits from AI without having to depend on the U.S. or other countries. It would mean Canadian universities and public agencies building and operating AI models optimized not for global scale and corporate profit, but for practical use by Canadians...

We are already on our way to having AI become an inextricable part of society. To ensure stability and prosperity for this country, Canadian users and developers must be able to turn to AI models built, controlled, and operated publicly in Canada instead of building on corporate platforms, American or otherwise... [Switzerland's funding of a public AI model, Apertus] represents precisely the paradigm shift Canada should embrace: AI as public infrastructure, like systems for transportation, water, or electricity, rather than private commodity... Public AI systems can incorporate mechanisms for genuine public input and democratic oversight on critical ethical questions: how to handle copyrighted works in training data, how to mitigate bias, how to distribute access when demand outstrips capacity, and how to license use for sensitive applications like policing or medicine...

Canada already has many of the building blocks for public AI. The country has world-class AI research institutions, including the Vector Institute, Mila, and CIFAR, which pioneered much of the deep learning revolution. Canada's $2-billion Sovereign AI Compute Strategy provides substantial funding. What's needed now is a reorientation away from viewing this as an opportunity to attract private capital, and toward a fully open public AI model.
Long-time Slashdot reader sinij has a different opinion. "To me, this sounds dystopian, because I can also imagine AI declining your permits, renewal of license, or medication due to misalignment or 'greater good' reasons."

But the Schneier/Sanders essays argues this creates "an alternative ownership structure for AI technology" that is allocating decision-making authority and value "to national public institutions rather than foreign corporations."
This discussion has been archived. No new comments can be posted.

Does Canada Need Nationalized, Public AI? More

Does Canada Need Nationalized, Public AI?

Comments Filter:

  • Build it and they will spend money (Score:4, Interesting)

    by NotEmmanuelGoldstein ( 6423622 ) on Sunday March 15, 2026 @06:56AM (#66042268)
    Everyone needs water, heat/energy, healthcare, communication, transportation, shelter and education. What do people do with AI: What problem is solved by building more AI? Recent analysis suggests the cost in time, both computing and human attention, is not cost-effective.

    Like education and energy (electricity), the government providing a uniform service, reduces the burden on businesses. But the random quality and massive cost of current sentence generators does not make this a uniform experience for any person or entity.

    This is a 'solution' looking for a problem and some people are demanding the taxpayer pay the bill.

    • I think the logic behind this kind of shit is "The AI hypers are right! Every person who uses AI is suddenly 10X as productive! Mass unemployment is around the corner and the only solution is nationalization so the government can hand out money and nobody needs a job any more!"

      The good news is that Sam Altman, Peter Theil, et al, may well have burnt their political will pushing this shit as nobody, even gullible politicians and journalists, are going to trust them again when the shit finally hits the fan an

    • As a Canadian I think this is a very stupid idea for many reasons, but the general dismissive attitude of LLM technology here on Slashdot is shocking though not surprising. Since the start of my career in IT 25 years ago I noticed this intense cynicism of new technology from those in the field. They’ve lost all excitement and imagination for things except perhaps some personal niche interests. Some of it is earned of course, we’ve all seen vapourware and snake oil, but the instinctive behaviour

      • It's not the technology I fear. It's the insane megalomaniacs and psychopaths that will own the AI and most certainly will use it for their benefit and not society. It's easy to be stary eyes in your early 20s because you are naive about just how rotten people are.

        The technologies we are creating *could* liberate us all from work and give us all paradise. Of course, why would the top .01% keep humans around when AI robots can literally do everything they need? It's a serious question. If they don't release

        • Your fears, like this Canadian notion, are ill-timed. The industry is still in a high-risk, nebulous reward, stage. The people you're worried about may be about to lose everything. Or they may be about to win big. Assumptions about what things will look like after that is resolved are foolish. Just as it is foolish for Canada to start looking at jumping into the fray now, before they can even hope to know if the invested tax dollars will be wasted.

          • Sure, but are they "to big to fail" and I'll have to bail them out anyway? Seems to be par for the course in the USA. Citizens get to bail out big business and those same businesses continue to make record profits.

            I mean, it does make sense when you realize USA is really just a big profit-driven business and not actually a governing body that cares for its citizens.

            • I'll try and ignore, for now, the pointless and childish insult laid against my country, and just say that I didn't agree with those bailouts either and they are not likely to happen again.

      • It is not sentence generation or regurgitation.

        But that's the problem, it is regurgitation, even if a targeted one from a vast array of stomach contents. You've been fooled by a stochastic parrot. Is it a breakthrough in human/computer interface if the human's input has a high likelihood of coming out mangled for reasons we can't understand or automatically correct? Is it really conversing with humans by spewing out a statistically likely response it has no real understanding of, or is it just running the latest successor to ELIZA?

        With AI's ability to p

  • Does anyone need AI? (Score:1)

    by Anonymous Coward

    Forget the "Nationalized, public" part - does anyone really need AI?

    This kind of initiative smacks of creating demand for AI services without ever saying what those services are for.

    • Which is the question they should answer before asking "do we need a public/national one?"

      Or better, "Would we be dumping tons of tax dollars into something that's about to go bust?"

      Perhaps the whole thing should resolve down to, "is it better to continue letting private funding take the lead and the risk while government waits and watches, or should government jump in now and put public funds at risk, in the hopes that something indefinably better may emerge?"

  • Provide it as a baseline, as long as is't mandated (Score:4, Interesting)

    by williamyf ( 227051 ) on Sunday March 15, 2026 @07:31AM (#66042282)

    A Canadian AI (or a national/multicountry AI) would work as a nice baseline for everyone in the country.

    Instead of having hamstrung gratis AI that can be withdrawn at a moments notice, you get fully fledged Gratis AI that depends solely on your country. If then you (or your company) wan to pay for some other AI, so be it.

    From a soveringty point of view is cool, as you are not beholden to for profit companies, be them national or foreign.

    And please remmeber that AI ius much more than LLMs and other Gen AI. Slef-Driving Cars, Platoon Driving trucks, (semi)Autonomus androids/robots, and context aware industrial machine tools/robots need AI too.

    I hope canada is contemplating these cases as well, and not fixated on LLMs

  • Teacher here. I know a guy who is pleading for a national LLM for schools. Free to use, unbiased, privacy oriented ... Never considered it, but he has a good point. The training data probably will get very political at some point. At the pace these things go, I guess we will get this in 25 years or so. If he reads this, thanks for all your work!

    • Re: (Score:1)

      by Anonymous Coward

      But will the training data of a nationalized AI be unbiased?

      Funny enough, for current AI most bias are (system-level) prompts. Use Grok via API and it is fine, use it via Twitter and it hails Musk. Download DeepSeek weights and discuss Chinese war crimes, use DeepSeek via their website and it declines to answer. Model-level alignment isn't easy without lobotomizing the model, but alignment via prompt works well enough to prevent Chinese people from asking about what their government did.

      • But will the training data of a nationalized AI be unbiased?

        At the very minimum, it will not be biased toward the interests of Canada's rivals, which makes it usable in applications where there the absence of foreign interference is a criterion, such as for school, government and defence.

        If you're a simple citizen it's a different scenario. Your experiments tend to demonstrate the biases weren't introduced at the level of the training dataset, but there is no guarantee this remains the same in the future.

        • At the very minimum, it will not be biased toward the interests of Canada's rivals.

          Are you sure of that? How in the world would you measure and audit that?

          IMHO, bias is in the eye of the beholder. What one person might call biased another might call obviously correct. Forget AI, every human is biased to every other human. When I talk to my 20-something daughter, she (lovingly) thinks I'm a biased anarcho-capitalist-libertarian, I (lovingly) think she's an indoctrinated progressive feminist wackadoodle.

          • I mean the model owner controls the bias related to the selection of the dataset. I have to assume that OpenAI/DeepSeek/... will select whatever dataset will make their government happy, for example do not include books that present a version of History that isn't popular around them. In the context is "nationalized, public AI", a Canadian agency would perform the training, to be used by other Canadian agencies (and citizen). It could be that the resulting model still is biased, but at least the Canadian go

            • I mean the model owner controls the bias related to the selection of the dataset... It could be that the resulting model still is biased, but at least the Canadian government users would be free from the voluntary effort of foreign governments in biasing their models.

              No doubt. And hopefully the Canadian government is less heavy-handed than other governments. I wouldn't assume that's the case but maybe it will be.

              What I'm most comfortable with is there being lots of models, created by lots of organizations, funded by a plethora of sources, produced in different societies, and freely (as in speech) available to all humans. That way, if one is too biased, people will ignore it in droves.

      • IMO for anything nationalized in terms of software should be open source, and that is all of it including the training data and methods, prompts, etc. Maybe this becomes political football but we should all be able to observe the game taking place, that should minimize the amount of fuckery.

        As for the training data use what the government has access to legally, I would assume that is the Library of Congress and similar materials, there isn't the pursuit of profit to motivate the system to train off of ever

  • Bruce is from NY. He isnâ(TM)t Canadian.

    Though he is very polite.

    • Re: (Score:2)

      by Entrope ( 68843 )

      He's also a fool when it comes to politics. People should want government writing rules and picking winners for AI just as much as for social and news media: not picking winners at all, and setting as few and as narrow rules as possible. Do you want the default (or only) AI service to run like the DMV?

      Betteridge's Law continues to hold true.

      • Do you want the default (or only) AI service to run like the DMV?

        Although governments all over the world are criticized by their citizens - and often rightly so - it's probably a good idea to recognize that the experience of bureaucracies from one country / jurisdiction to another can vary greatly.

        At least here in Ontario, our Ministry of Transportation is pretty efficient, fair, and convenient. I think an AI service run like our MoT would probably be at least OK, and maybe even much better than that. And that's in a province whose premier is firmly in the pockets of dev

    • Are you telling me that this whole thing is actually two Americans telling Canada how to do AI so Americans don't tell Canada how to do AI?

  • Fucking no (Score:2, Informative)

    by drinkypoo ( 153816 )

    I'm not saying this for Canada, I'm saying this in general.

    People follow AIs like sheep following a shepherd to food. A state media AI is not a good idea, unless you really really want fascism. Once people get used to listening to it they will never, ever stop.

    • MAID (medical euthanasia) is already one of the leading causes of death [calgarysbusiness.ca] in Canada. AI will be used to accelerate this, because MAID is cheaper than treating illness.

      • If they wanted to use MAID to save treatment costs, they wouldn't refuse half of the applicants.

        • If the #5 cause of death in a country is government funded and sanctioned suicide, ether something is horribly, monstrously wrong, or they ran out of other things to die of. Which of those sounds more likely?

          • Or there were just that many people suffering before but now you know about them.

            • No. That doesn't track. That many people desperately wanted to die but were waiting for government permission? Not believable.

              • There are a lot of people out there suffering. It's probably way higher in the US.

                • That is a non sequitur, not a point. There are billions suffering right now. There are people suffering under actual dictatorships. There are people starving to death for no fault of their own. There are people permanently crippled by conditions long since eradicated in the developed world. Suffering is part of the human experience. Bones and hearts break.

                  And the overwhelming majority still get up in the morning and go on living.

                  • But it's not for you to say they should keep on suffering, because you have no idea what they are going through. Life is truly shitty. If I didn't have people who would miss me I would leave myself.

                    • And that's why governments should never endorse suicide. You are telling me that you suffer from the symptoms of a treatable psychiatric/medical condition. I would not let you kill yourself. I would tie your ass up and take you to a therapist first. Canada appears (from some of the reporting I've seen) to be letting people top themselves just because they're depressed. If that's incorrect, good. But will it stay that way?

                      Oh, and just to be clear - your life is not truly shitty. You have people who

                    • You act like there is a drug for all ills. Most psychiatric meds only work by turning you into a zombie. So you advocate that people should zombify themselves with drugs and live that way?

                    • The waiting time "to see a shrink" where I am is two years for even the most serious cases. The fact that a person is su8cidal is sadly not enough to move you ahead in the queue.

                    • Where the hell are you that you have to wait 2 years?!?! If it's Massachusetts or Costa Rica, I might be able to help. And you can always talk to me, though this is a pretty public place.
                    • I take Wellbutrin for my anxiety&depression. It helps quite a lot without zombifying me, but there were years of talk therapy before that. Exercise and sunshine may be almost as effective for some people. There have been promising results with psilocybin as well.

                    • Because I know someone who became totally non-functional due to her psychosis and she had to wait two years. Sure, I suppose people with serious mental issues should just use ChatGPT right?

                    • Had to wait 2 years where? You can walk into a US hospital, say you're in a mental health crisis, and get admitted to the psych ward on the spot.

                      Back in 2001 I was in a particularly bad spot. My therapist happened to also work at the local hospital's psych ward. I told him I thought I needed to be admitted, and he offered to take me over and check me in. I said, "how about tomorrow, I have to go perform on the radio tonight" (used to be a musician). I was in the next day. I was broke and had no ins

                    • But that psych ward will only keep you two days for observation and then let you go. They may give you a prescription got zombie meds and send you on your way. No hospital in the US has a way to keep people until they are better unless those price are wealthy and for it.

                    • I think observation is 3 days. And no, that's only when someone else puts you there. They'll let you stay, but one of the finance people will come bother you about signing up for assistance. At least that was my experience. Usually, just a couple of days plus a scrip means the patient didn't want to be there. Or didn't need to be there. A lot of people who end up in a ward involuntarily just weren't taking their meds.
                      Zombie drugs are only given when needed, and they don't just put people on the hea

                    • I think things are very different in the US as opposed to Canada. American thinking is that a person is only worth as much as they earn. So this is why healthcare is attached to having a job and earnings. Canada has a different philosophy that you keep as many people healthy as you can and in turn they will be productive over all. There is really no way that these meet in the middle. I know of two cases personally where people were seriously or fatally ill (one person ended up killing his own mother in

                    • So, you're in Canada! Well, I guess that explains the bits that didn't make sense to me - like a 2 year waiting period. That sounded so bonkers I had to ask where you were that it was true.

                      I don't know why you're trying to turn it into a criticism of the American healthcare system. Yes, there are problems with it, but at least a person in a mental health crisis can be seen right away instead of having to wait two years. And, as I said, I went through that process while unemployed and entirely broke.

                    • Mental health care for lowincome people in the U.S. is technically available but extremely difficult to access in practice. Medicaid is the main source of coverage, and it pays for most mental health services nationwide, but access varies widely by state because each state sets its own rules. Even with Medicaid, people run into major barriers: huge provider shortages, long waitlists, and the fact that many psychiatrists don’t accept Medicaid due to low reimbursement rates. More than half of U.S. count

                    • I don't want to argue about the advantages and disadvantages for the American and Canadian (I almost typed comedian, dunno why) systems. They're both flawed, arguing whose is worse seems silly today.

                      Though I wonder how many of the problems can be traced to both the US and Canada being mostly empty. Of course there are a huge number of US counties that don't have psychologists, I'm surprised that it's only half! There are so many rural counties without enough residents to keep a shrink in business. It'

                    • That's why there are governments that collect taxes. They are there to solve such problems that won't be solved by capitalism.

                    • Capitalism is the economic system. During a frightfully rare moment of insight, Marx pointed out that you can't really separate economic and political systems. They pair up with their compatible partner. Capitalism, where individuals make their own economic decisions and use what they have to profit, really only works/exists when paired with a pluralistic (let's call it Democratic, because we're lazy instead of accurate) Republic, where individuals make their own political decisions and participate in pub

                    • You must be in the US, the only country where people don't seem to want the government to help them. Meanwhile, your neighbor to the north has managed to be very successful with capitalist socialism.

                    • There's no such thing my friend. Unless you want to count "State Capitalism", which was supposed to be a transitional phase between Capitalism and Socialism. And which China has ended up in.

                      But you probably mean something far simpler. In a Capitalist Democracy, the line between public and private services can be drawn in a number of places. In the US, we prefer more private to public, but in Northern Europe, the line is drawn much further on the public side. But still Capitalist Democracies. Canada,

                    • Ok will Canada is, in fact, capitalist socialism. I'll believe every school of economics over you. You just want to discount it because it isn't perfect. That's what the right wing does. They discount everyone because they see a flaw but they don't correct their own.

    • That ship has pretty much sailed when it comes to de facto state owned media like in Canada.

    • There's also the more obvious and immediate risk that they'll spend billions on something that nobody wants, or doesn't work, or turns out not to have been necessary for some unpredictable reason. AI is an immature industry beset by wild speculation and may turn out to be a bubble. No nation should the thinking about jumping even further ahead of the facts than private investors have. That's irresponsible.
  • What could possibly go wrong if the government knows every question we ask, everything we search for, ...
  • All Universities, colleges and schools need to kick AI permanently out of the institutions and focus on high quality human research as their edge. Chatgpt.com needs to be blocked at the firewall of all human resecting nations. In an era where slop is the default, the humans that fight back will be kings.

    • Re: (Score:3)

      by TWX ( 665546 )

      The insidious part about using free AI is that you're training something owned by someone else to do your work. That means whomever owns that AI now has the option of replacing some aspect of your work, rendering you redundant.

      For the short term, free AI empowers the unskilled masses because they can use what skilled people misguidedly trained it to do. For the long term, AI only empowers the wealthy who own it, to the harm of the unskilled and skilled alike.

      • This sounds more like lashing out about your own capabilities and training rather than reality. Congrats on recognizing your own limitations I guess, even if it’s subconscious and manifests as such.

        Change, and advancement in automation and technology, have been constantly altering the specific work and roles of humans for all of history. Shouldn’t we be used to that by now? I have humility in the sense I don’t believe myself to be intelligent- which funny enough is perhaps my greatest i

        • Re: (Score:2)

          by TWX ( 665546 )

          It's interesting that you took this to intelligence and software coding when I was talking about skill and had not limited my criticism to software coding. Honestly at the time I made my post, software wasn't even on my radar, and skill only weakly correlates with intelligence.

  • I think there's way too many people who imagine AI to be some sort of Stuxnet, and they're letting their imaginations run wild. It's all pareidolia [wikipedia.org] at work. AI is just an amalgamation of training data. Think of it like hamburger...when you look at what comes out of the meat grinder, you can't say to yourself, "That morsel came from the shank, and that little bit must be the filet, and that tidbit there came from the rib." It all clearly came from somewhere, but when blended together, you can no longer di

    • Can we perhaps stop trying to anthropomorphize an algorithm?

      Yeah! AI hates that!

      People who don't know how to think can't recognize the difference between token diarrhea and thinking.

    • Yeah, keep believing it's not going to get baked into the National security defense network... they'd never do something like that!
      Right now, it's not intelligent... trust us computer geeks when we say 'they're already working on pushing it into AGI and beyond (why would they be working on human brain tissue grown on a processor?).
      Isn't this (the current iteration) how stuff like The Matrix or SkyNet or I, Robot began? Someone tried it, applied it to robots, automated everything, the AI took over and decid

  • The best option is to trust AI to a multinational corporation thaat has the primary motivation of ever greater profits.

    Because corporate greed will "trickle down" to the plebs in the form of minor life improvement that they should be thankful for.

    It's true the government often has a poor reputation running things...but ultimately it has been running national services for decades. So it can be done. It's not perfect but can we really trust a handful of companies with all our AI queries, automation, reapo
    • Can you trust a government to jump into a fury of speculation and make all the right choices? Is it reasonable to make assumptions about the end state of something that is still emerging? Should a nation put billions of the taxpayers' funds at risk in a turbulent market that may collapse at any minute?

      Or should governments wait and see how the private investment turns out, see what the outcomes are, and then start asking if they have any need to establish their own public models? Which, at that point,

  • monopolies (Score:3, Interesting)

    by 7311587 ( 755664 ) on Sunday March 15, 2026 @10:35AM (#66042482)
    corporations are not allowed to have monopolies so they use the government to create de facto monopolies via regulations.

  • ...governments are "aligned with our collective good"
    In the real world, governments are aligned with the powerful

  • I would certainly welcome a national fund (could even be multinational and accept donations) for an open weight AI with full ecosystem even including hosting, that is fully open source and available for local LLM use as well. Not a Canadian but just by building it, could benefit from network effects.Including info about the training set, training methods and system prompts and allowing it to be configured freely could usher in a eave if experimentation and support the growth of centers of excellence, and a

  • When I hear "nationalized", I hear "government takes ownership of privately owned organizations." I don't think this is what they're proposing. Without just compensation, that would be wrong.

    I don't see the Canadian government shelling out, say, $400 billion to buy Anthropic or whoever but maybe they would. It sounds like what they want instead is for the Canadian government to create an entirely new LLM and AI apps using nothing but government funding. That's not a cheap undertaking.

    However, if that's what

    • Don't read too much into it. Under all the flowery language, what they want is more govt money for AI to compete with the US.

  • Why not just nationalize everything? The argument would be the same.

  • The country already doesn't make any sense, and is pretty much communist. So expect it.
  • Does Canada Need Nationalized, Public Magic?

  • It's eh-I you hoser

  • We all met at the Tim Hortons, and the answer is Yes! Now that the stinky orange meanie in the whitehouse has kicked us out of the proverbial basement, a big 'ol maple-flavored AI brain sounds like a good idea.

  • ok (Score:1)

    by Crenor ( 6176856 )
    sure everyone "should" but we cannot afford the current high cost. So maybe when the costs come down. Or kiss the butts of a few big AI companies. Just don't place your bets yet on who is going to win. Waaaaaayyyyy too early. But yea, expect to be some big companies bitch until we get our own.
    • I don't know about the "should", but I can't argue that it isn't too early.

      Canada is getting out over its skis if it's even thinking about this now. "Hey, let's make a lot of ill-founded assumptions and throw a ton of taxpayer funds after them. What could go wrong?"

      No. Wait to see how the industry shakes out. Wait to see if you actually need to take any action. Don't put taxpayer funds at risk unless you really have to.

  • Compute is the issue: to build one's own frontier models, we need a lot of it. It's far from free. We know how to build the models, we just lack the compute. There's no intrinsic reason public funding of AI compute can't work, if there's the will.

  • An AI bot that will report you to the Mounties if you ask it why your hip replacement will take 15 months or your neighborhood is suddenly full of Congoloids.

    An idea so great, it must be made illegal not to finance it.

    Looks like America's Hat has decided to be even more Marxist. Color me unsurprised.

Slashdot Top Deals

Fundamentally, there may be no basis for anything.

Close