• betanumerus@lemmy.ca
    link
    fedilink
    English
    arrow-up
    15
    arrow-down
    1
    ·
    6 days ago

    As long as the energy is NOT from fossil sources, I’m ok with this.

    AI must be built on renewables. It must not rely on adding greenhouse gases and altering the natural course of climates.

    • brianpeiris@lemmy.ca
      link
      fedilink
      English
      arrow-up
      7
      arrow-down
      1
      ·
      6 days ago

      Or, you know, how about we not spend any energy on an inherently wasteful, unethical, financially bankrupt, slop machine.

  • CanadianCorhen@lemmy.ca
    link
    fedilink
    arrow-up
    13
    arrow-down
    1
    ·
    6 days ago

    I’m fine with that. If a AI company wants to build major wind, solar, nuclear, or hydro-electric project to power their bubble, thats great!

    We can buy it for pennies on the dollar when the bubble bursts

    • captain_oni@lemmy.blahaj.zone
      link
      fedilink
      English
      arrow-up
      4
      ·
      6 days ago

      Would be the funniest thing if after the burst techbros were like “I wanted to build skynet and all I ended up with was this bullshit solar power plant”

  • mHtt@lemmy.ca
    link
    fedilink
    arrow-up
    4
    arrow-down
    1
    ·
    6 days ago

    whats fucked up, is they wont even say why we need them, just that we NEED them now.

    im guessing it has to do with ai controlled drone swarms, but my guess is a good as anyone elses. its not about shitty cat videos. we ave those.

  • yeehaw@lemmy.ca
    link
    fedilink
    arrow-up
    4
    arrow-down
    2
    ·
    6 days ago

    I mean sure, make them pay for some massive battery banks + solar beside their datacenters. I’d be more OK with it then.

  • [object Object]@lemmy.ca
    link
    fedilink
    arrow-up
    7
    arrow-down
    7
    ·
    6 days ago

    The reality is AI exists, it’s here, and every other country is building cyber offensive capabilities, while nerfing cyber defence capabilities.

    We can pretend that human written code is so much better, but have you seen the CVE numbers this year?

    AI agents are finding and publishing 5x more bugs than humans ever did, even using the nerfed models. OpenAI, Anthropic, Kimi, Alibaba have all spent the last year focusing on “cyber” models, which means offensive attack capabilities.

    Pure vibe coded stuff is slop, yes, but AI is outperforming humans in most security critical areas now.

    We have about 22 data centres in progress across the country (9 in progress, 13 projects announced). A large size data centre in the US is 600-1200 MW, but the US has 500+ of these in progress right now. There’s not really a clear definition of an “AI data centre”, so the numbers you’ll get vary drastically, form the 5k total DCs in the US to a few hundred.

    Over one third of all our data centres are US owned, many of the service layers even in our Canadian owned DCs are run by subsidiaries of American companies (like Buzz, which moved their HQ to Texas a year or two ago). Then our next biggest owner of compute is OVH and French companies, who we’re still trying to become associate buddies with.

    Now, I do think we should protest some of these: O’Leary’s Wonder Valley actually does a decent job on water use and not being near people, but it is a gas turbine powered DC in the middle of a retired gas field — it is going to emit massive amounts of CO2, and I think that’s an abomination.

    We also have projects like Bell’s AI Fabric project (Kamloops and Merritt) where the new DCs are to be powered by hydro power. We can actually do hydro power and other greener energy sources for half the cost of US infrastructure, we just to make the compromises needed to do so.

    I think blanket protesting of data centres is kind of embarrassing, and this goes along with “why do we need to spend more money on the military!”. It’s because we are fucked if we don’t. If we don’t have a basically entirely new military in 5-10 years we’re going to become the next Ukraine. I’m not even sure we have two years of safety. We all saw the US meddling in Alberta’s referendum, we all see the Maga shitstains in BC, it just came out how much US based deepfake content there still is pushing separatism and subreddits like CanadianVisaReform, and Facebook is entirely foreign misinformation slop.

    I’m not saying we should be pushing towards the singularity and giving up all control, but we aren’t even a player in the game. We only have two major AI companies, Cohere and Waabi, and neither register anywhere in the list of remotely competitive models. Meanwhile, China is literally having a robot olympics.

    We need to be building sovereign AI systems in Canada. We need to do that sustainably, and we need to do that quickly.

    • Avid Amoeba@lemmy.ca
      link
      fedilink
      arrow-up
      2
      ·
      edit-2
      2 days ago

      It’s nice to say we need sovereign tech. Things tend to only happen when there’s material interest for them happening within each class. The American investor class has material interest in DCs to so that’s happening. People don’t want DCs because there’s nothing for them in it - they make their lives worse and they reduce their real estate wealth (often the only wealth they have) when they’re built close by, all the while their tech subs are climbing in price. Wanting to build sovereign tech beyond the want therefore means changing the material conditions. For example by first and foremost giving something concrete to people that is conditioned on or provided by the DC buildout. For example a decent public pension that’s not a patchwork poverty pension. If the gov’t truly believes these imvestments are going to improve productivity significantly, that’s one dividend they can front load to get buy-in from workers. Then for the domestic investor class - create demand for domestic tech using state procurement - the tested way to get markets to do something they don’t want to. If that doesn’t work, create a crown corp, hire people and start working. And at every step when there’s deliverable, give it to people. Got new hosted Nextcloud with Office for gov’t? Provide it to Canadians at cost so they switch away from MS. Of course the private domestic providers would never do that by themselves because they will profit maximize so those deliverables have to be gov’t provided or managed.

      BTW while the Europeans are doing something, I’ve seen past attempts fail so don’t count their digital sovereignty hatched yet. Big Tech will be working overtime to ensure they and us fail at these initiatives.

      One more wrinkle, any sovereign infra / R&D investments that are in private hands are subject to foreign purchase under the current free flow of capital regime. So there’s also that to think about.

      • [object Object]@lemmy.ca
        link
        fedilink
        arrow-up
        2
        ·
        2 days ago

        Good points, I agree with you

        On the returning value to the people, that is a huge issue and a large part of why we need sovereign AI. Any job losses here are never going to be compensated by the Americans, that’s the whole objective. But on that note, this is why I’m excited for the new sovereign wealth fund, though we are mostly using our oil windfall to fund this budget to invest instead of distributing it.

        You could argue either way on this one that what should be a public benefit is going to corporations, or you can argue that these investments unlock future value later. We should be deeply sceptical of that, but at the same time I’m kind of done with governments who keep pushing for less tax, less services, and never investing in ourselves and then complaining that very lack of investment means the system is failing.

        I do think it’s difficult to walk the line. Increasing our efficiency in-theory increases tax revenue. If you increase efficiency without workers, tax revenue goes down, which is a large part of what AI pitches to do. So then you need to crank up corporate taxes at some point to make up the shortfall, and I don’t see the current government or a future Libservative government doing that under their current political direction. At some point we’ll have to reconcile that issue.

        create demand for domestic tech using state procurement

        This one I worry about. State procurement is famously not very good. Technically, we can do procurement better already, but nobody wants the risk of anything but the lowest bidder or setting high technical bars, so we keep overpaying for under planned contracts.

    • [object Object]@lemmy.ca
      link
      fedilink
      arrow-up
      6
      arrow-down
      1
      ·
      6 days ago

      Just to add, we still don’t have a sovereign operating system, we spent ~$720M on US tech this year. We have no plan to migrate off Microsoft 365. We haven’t joined the EU’s sovereignty projects.

      We’re so so so far behind. I’m glad the government is finally starting to move, and honestly we have 15 years of letting Canadian tech and ownership languish to make up for.

    • Grimy@lemmy.world
      link
      fedilink
      arrow-up
      4
      arrow-down
      1
      ·
      6 days ago

      Awesome write up. I don’t get why they don’t build more in Quebec since they run on hydro.

      • [object Object]@lemmy.ca
        link
        fedilink
        arrow-up
        2
        arrow-down
        1
        ·
        6 days ago

        Partially it’s because it’s very hard to actually get a list of what is an “AI data centre” vs not.

        Quebec is weird though, it’s almost entirely foreign companies building there. Apparently the advertised the electricity low rates to get investment and then they rug-pulled and hiked the costs for data centres above 4MW (which is all of them).

        There are also a lot of crypto mining data centres that are migrating to AI workloads and that’s not counted in any of the lists of DC buildouts.