• 7 Posts
  • 191 Comments
Joined 1 year ago
cake
Cake day: June 4th, 2025

help-circle







  • So when it comes to using genAI, you are going to have a lot of people (including leftists) just using a corporate model, then you have a smaller subset trying out “open” models like deepseek, then a smaller subset locally hosting a distilled model, and a smaller subset fine tuning a model or using RAG, and smaller and smaller subsets of people maybe try to build something on their own.

    The db0 instance’s official stance is against using corpo AI, so you’re talking to the smaller subset already. So any arguments based on “most people” aren’t really relevant to the complaint you’ve made about this instance.

    So in reality, most of these users will be making predictable and easily influenced output using these tools.

    If you’re talking about the AI output, predictability is kinda the point. If you’re talking about the “influence” effect it has on certain people, watching youtubers or consuming any media can also influence people significantly. Does this mean listening to or creating podcasts is unethical and shouldn’t be tolerated? No, it means you should be careful what kind of media you consume regardless of its source. I am reasonably certain everyone on this instance knows genAI is quite fallible and biased.

    Also, I think the basic function of these tools, which will be true for open or corporate models, makes us less dependent on eachother, which I think will lead to even more self centered societies then we currently have and not the community centric goal that I think a lot of us share.

    First, there is a difference between dependence and dependability.

    • You do not want to be dependent on specific members of your community; that would imply you and others couldn’t get by without them, meaning your community is not resilient.
    • You do want to be able to depend on others in your community (and vice versa); this makes your community stronger.

    Secondly, one could make this argument about any information sharing system ever. Youtube tutorials have made people less dependent on repair-men. Does that mean tutorials on how to fix an air conditioner are unethical?

    Furthermore, genAI is a hobby for some people (especially those on this instance who use it). People like sharing cool things they made with/about their hobbies, and finding people with shared interests is a great way to build a sense of community. However, if an instance were to assert that no use of genAI is ethical ever, that would alienate people who use it in an ethical way, preventing/discouraging them from sharing/creating and/or participating in communities that they would have otherwise. I’d argue that is literally, directly against the “community centric goal” which “a lot of us share.”


  • I don’t use genAI basically at all, but I have experience with machine learning in general. Here’s my take:

    Generative AI itself is literally just statistical modeling. Very complicated statistical modeling/prediction but still the same idea as any other method for prediction/classification.

    The only problem is how certain models are trained and/or used.

    You can (and imo should because it’s fun/interesting) create your own generative AI using responsibly sourced training data and open source code. Is that a crime? Does that hurt anyone? Does that harm society as a whole? No.

    The concept of “GenAI” is not the problem. In fact, utilizing it for protien/drug discovery is pretty fucking cool, as is the fact it makes language translation significantly more accessible and accurate.

    The problems you listed mostly apply to strictly corporate GenAI models. They are problematic not because they are generative statistical models but because they are built, maintained, controlled, and used by capitalist institutions in a capitalist society which allows them to do any damage they want as long as they can continue bribing politicianslining the pockets of investors


  • Learn mathematics. If something/someone tries to assert the following fallacious syllogism, then it is propaganda:

    1. P implies Q
    2. Q
    3. Therefore P

    If someone/some-organization asserts that a simple correlation means causation—especially in the case of surveys or polls—thats propaganda.

    Hell, a lot of the time the correlation doesn’t even verifiably exist.


    If someone asserts that because the publication/school publishing a paper has prestige, that paper must be well founded and correct, they are using it as propaganda regardless of whether it draws solid conclusions or not.

    The same is, unfortunately, true in reverse. Just because something is a biased source or even a superspreader of propaganda, doesn’t necessarily mean they are always spouting misinformation. Though if the conclusions are sound, you should be able to find a better source than any clearly biased news/journalism organization

    If they cite a paper/study:

    • Ask what the purpose of the paper was
    • Look to see if the article actually draws the same conclusion as the paper or if they omit crucial details/flaws that the researchers mention
    • Ask what department/background the researchers were part of and if that background fits with the research being done
    • Ask what the procedure for data collection and analysis was
      • Compare with standard procedure for similar but non-political experiment(s) to see if the data collection method followed good procedure