• china🇨🇳@lemmy.ca
    link
    fedilink
    arrow-up
    71
    arrow-down
    5
    ·
    1 day ago

    Genuine question; what’s the difference between DeepSeek gathering data and for example ChatGPT or Gemini?

    because China bad and USA good :(

    • comrade_twisty@feddit.org
      link
      fedilink
      arrow-up
      14
      arrow-down
      41
      ·
      1 day ago

      I mean we can generally agree that China is worse. That doesn’t mean we should willingly donate our data to the US, but if you have to choose between amputating your foot or your little toe it’s an easy decision.

      • Knock_Knock_Lemmy_In@lemmy.world
        link
        fedilink
        arrow-up
        51
        arrow-down
        6
        ·
        1 day ago

        I mean we can generally agree that China is worse.

        Can we? Which country had open sourced it’s models and weightings so that the LLM can be run locally and spyfree?

        • HelloRoot@lemy.lol
          link
          fedilink
          English
          arrow-up
          6
          arrow-down
          20
          ·
          edit-2
          1 day ago

          France, Mistral. There are also many others in US and worldwide.

          China is worse when comparing many other factors, but if ALL you care about is whether the model weights are os, then it is usually not decided by countries, but rather companies.

          • Knock_Knock_Lemmy_In@lemmy.world
            link
            fedilink
            arrow-up
            18
            arrow-down
            4
            ·
            1 day ago

            There are also many others in US and worldwide.

            We are comparing chatGPT and Deepseek.

            China is worse when comparing many other factors

            Again, not an opinion everyone agrees on. And this very much depends on which factors are cherry picked.

            but if ALL you care about is whether the model weights are os,

            If you want privacy then you need to run locally, and to run locally all you do care about is model weights.

            then it is usually not decided by countries, but rather companies.

            Not sure why this dimension is worth adding to the discussion.

            • Ulrich@feddit.org
              link
              fedilink
              English
              arrow-up
              1
              arrow-down
              8
              ·
              edit-2
              13 hours ago

              We are comparing chatGPT and Deepseek.

              You asked “which countries”.

              not an opinion everyone agrees on.

              Only people who don’t have their heads buried in the sand. You can’t possibly believe China is not much much worse than the US. I literally don’t believe you. You’re a bot or a paid actor.

              • Knock_Knock_Lemmy_In@lemmy.world
                link
                fedilink
                arrow-up
                6
                ·
                9 hours ago

                You asked “which countries”.

                Countries and LLM model providers associated with those countries.

                • All openai data goes to the NSA

                • Online deepseek data goes to the CCP. Local models stay private.

                You can’t possibly believe China is not much much worse than the US.

                When it comes to AI models, China is streets ahead in openness.

                But if you are introducing other factors, ask an Iranian what they prefer; money from China or bombs from the US and vassal states?

                The only people believing that Americans are superior in everything are Americans.

                • Ulrich@feddit.org
                  link
                  fedilink
                  English
                  arrow-up
                  2
                  arrow-down
                  3
                  ·
                  7 hours ago

                  When it comes to AI models, China is streets ahead in openness.

                  No one was talking about openness, we were talking about data collection.

                  All openai data goes to the NSA

                  Online deepseek data goes to the CCP. Local models stay private.

                  You’re comparing online models with local models. Local OpenAI models are also private.

                  The only people believing that Americans are superior in everything are Americans.

                  Good thing no one said that. It’s not just America either. Most countries are significantly more free than China (and Russia and North Korea), and if you don’t believe that, you need to open your eyes.

                  • Knock_Knock_Lemmy_In@lemmy.world
                    link
                    fedilink
                    arrow-up
                    2
                    arrow-down
                    1
                    ·
                    edit-2
                    7 hours ago

                    No one was talking about openness, we were talking about data collection.

                    The only way to ensure no data collection is to run locally, which requires openness.

                    You’re comparing online models with local models.

                    Correct

                    Local OpenAI models are also private.

                    They don’t exist, and privacy cannot be guaranteed.

                    Good thing no one said that. It’s not just America either.

                    So you do think that.

                    Most countries are significantly more free than China (and Russia and North Korea),

                    The word used was “worse” which is highly subjective.

                    and if you don’t believe that, you need to open your eyes.

                    Residents of “the land of the free” don’t understand irony.

            • Ilandar@lemmy.today
              link
              fedilink
              arrow-up
              2
              arrow-down
              5
              ·
              15 hours ago

              We are comparing chatGPT and Deepseek.

              Then why are you asking binary questions about countries? Meta released open source models and weights way before DeepSeek, but I guess that wouldn’t fit into your “China is better” narrative so we’ll just pretend Llama never happened.

              • Knock_Knock_Lemmy_In@lemmy.world
                link
                fedilink
                arrow-up
                2
                arrow-down
                1
                ·
                9 hours ago

                Then why are you asking binary questions about countries

                Because a highly opinionated binary statement was made about China. I’m not saying China is better than the US. I’m saying it’s not worse.

        • gaylord_fartmaster@lemmy.world
          link
          fedilink
          arrow-up
          7
          arrow-down
          9
          ·
          1 day ago

          I run deepseek locally so obviously I appreciate that they made that decision, but let’s not pretend that something can’t be given away for free with bad intentions. People running LLMs locally are a drop in the bucket compared to people just downloading an app.

            • gaylord_fartmaster@lemmy.world
              link
              fedilink
              arrow-up
              3
              arrow-down
              1
              ·
              4 hours ago

              I have no idea what point you’re trying to make. That means they have good intentions with the data they’re collecting about the users of their mobile apps?

              • Diurnambule@jlai.lu
                link
                fedilink
                arrow-up
                2
                arrow-down
                1
                ·
                edit-2
                3 hours ago

                That mean they have at least less bad intentions than chatgpt at least the opened their model on contrary of chatgpt.

                • gaylord_fartmaster@lemmy.world
                  link
                  fedilink
                  arrow-up
                  3
                  ·
                  3 hours ago

                  It does not mean that at all. Nothing about choosing to release a model for free or not has any bearing on whether or not the app will respect the privacy of its users.

                  OpenAI could feel like they’re making enough money off of their proprietary model that they don’t need to collect data (I’m not saying this is likely), while DeepSeek could have released the model for free hoping mass adoption leads to more app downloads and more data to harvest. I don’t assume either has good intentions.

              • Knock_Knock_Lemmy_In@lemmy.world
                link
                fedilink
                arrow-up
                1
                ·
                4 hours ago

                The point I’m trying to make is that I don’t think the CCP cares about the users of deepseek mobile app when they already have an ocean of data from tiktok. That’s the real trojan horse.

                Mobile access to an LLM is a similarity. The big difference between chatgpt and deepseek is the open model weights.

                • gaylord_fartmaster@lemmy.world
                  link
                  fedilink
                  arrow-up
                  2
                  ·
                  3 hours ago

                  You think them already invasively farming user data from tiktok somehow makes them less likely to do the same thing with another app, and not more likely?

                  If someone stole $1000 from you and then asked to borrow $20, would you give it to them? Surely they won’t steal that too, they already have $1000.