Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)T
Posts
3
Comments
70
Joined
2 yr. ago

  • this is obviously talking about their web app, which most people will be using. In this special instance, it was clearly not the LLM itself censoring the Tiananmen Square, but a layer on top.

    i have not bothered downloading and asking deepseek about Tiananmen Square. so i cannot know what the model would have generated. however, it is possible that certain biasses are trained into any model.

    i am pretty sure, this blog is aimed at the average user. while i wouldn't trust any LLM company with my data, i certainly wouldn't want the chinese government to have them. anyone that knows how to use (ollama)[https://github.com/ollama/ollama] should know these telemetry data don't apply to running locally. but for sure, pointing it out in the blog would help.

  • fair enough. i guess the usa never did a great job ar limiting their presidents power. that way he can extend his reach way further down, than he should...

  • well google always displays the locally official names and borders. so just business as usual.

    but why does the president of the usa get to decide what places are called? isn't there a cartography department or something?

  • as @damnthefilibuster@lemmy.world already mentioned: GitLab CI

    Jenkins is a CI application from before CI was cool. GitLab CI is integrated and can trigger on certain events. Additionally you mentioned, that you want to publish on a public repo anyway.

    You are probably are comfortable with containers. So GitLab CI should be easy for you to learn - as it pretty much starts up a container to do certain tasks. I've seen suggestions for Kubernetes, which for sure is the more mature solution. But i would question, whether you need the added functionality and complexity of K8s for a home setup.

    To gain access to your local network, you can use the runner for a secure connection (as described by damnthefilibuster). or you could SSH into the machine, as long as you have it in a DMZ. Drawback is that you have to be more sure about your network infrastructure. Benefit is that it is a more general approach. Obviously you need to store all certs, keys and preferably even addresses in secrets, not the .gitlab-ci.yml.

    As you can see from this thread, there are many ways which lead to rome. My advice is to start with something simple and lightweight, which you understand. adding complexity down the road is easier, than removing it.

  • The main angle is not to 'poisen' the training set. it is to waste time, energy and resources. the site loads deliberately slow and produces garbage, which has to be filtered out.

    as i said: not a silver bullet. but at least some threads where tied up collecting garbage painfully slow. as the data is useless, whatever their cleanup process is, has more to do. or it might even be tricked into discarding the whole website, as the signal to noise ratio is bad.

    so i would still say the author achieved his goal.

  • sure, it is easy to detect and they will. however, at the moment they don't seem to be doing it. The author said this after deploying a POC:

    Aaron B told 404 Media “If that’s, true, I’ve several million lines of access log that says even Google Almighty didn’t graduate” to avoiding the trap.

    So no, it is not a silver bullet. but it is a defense strategy, which seems to work at the moment.

  • AI would have used the correct german ö

  • we are all numbers. lemmy.ca has a user number for you, your government has a number for you, your local library has a number for you.

    that is just how a digital world works.

  • I don't think these computers where designed in the Netherlands

  • dressing up in a costume -> hobby

    harassing people -/-> hobby

    the dissonance in these two posts

  • the ceo is just the effect, not the cause. the us laws allow such bullshit and do not protect the weak (at all). what this one ceo did was, like what many other ceo's do, immoral but legal. you cant jail someone for legal stuff.

    change the system and force them to adhere to modern moral standards. if they try to pull some bs now, it is quite easy to lock them away.

  • you are painting an oversimplified picture.

    "i am sure you'd have preferred Gandhi to pick up a gun because he was met with violence?" we can chase eachother with such oversimplifications forever.

    reality is much more complicated than such simple statements. so lets not use their inflammatory nature and focus on the actual problem. which, in that case seems, that people feel disbanded by sociaty to such a degree.

  • that's a nice thought. however, life is not a hollywood movie, so i wouldn't count on it.

  • right now we don't have any real customers that use it - as the plugin did not sell yet.

    but from testing at customer sites with real people that would use it - we got only positive feedback. which is not hard to imagine: the RAG + LLM enables less experienced users to navigate a huge and complex network of information.

    but it for sure is also a buzzword execs like to see: they talked to us because we have AI. saw that the main product is good. bought the main product and decided the AI is too expensive.

    in the end it doesn't matter to me. the 2w of AI was a fun sidequest and it left us with a passive boost for sales.

  • i know it is an unpopular opinion around here. but currently AI features open doors for sales. that is important.

    for the software i help develop, we introduced an optional AI integration. just its presence allowed us to sell the main SW multiple times. the AI plugin was never sold so far.

    investment AI: 2 weeks of gluecode. i am not concerned with finances, but that plugin is for sure net positive.

  • that's the great thing: you don't have to use it

  • as the screenshot shows, it is opt-in

  • then stay at twitter 🤷‍♂️ for now blue sky is many people's choice for a reason.