Hello! Let us begin with a method for styling pride flags into your dividers, using some neat CSS technique in-line.
Also, a![[site community profile]](https://www.dreamwidth.org/img/comm_staff.png)
dw_news post welcoming people moving to Dreamwidth from Reddit, since Reddit decided that it wanted to charge ludicrous amounts of money for API access, and doing so essentially kills third-party tools and other ways of accessing Reddit (including several methods intended to make the site more accessible.)
A reminder that financial entities like Venmo and PayPal are not incorporated as banks or credit unions and do not have deposit insurance in case of financial difficulties. They're convenient, yes, but they don't have the full protections you would need if the companies collapsed. Some of their items do, because they're actually held and transacted with banks, but not all of them, and not necessarily by default. Treat them with the same amount of trust you would treat cryptocurrency.
A helpline for people with eating disorders fired all of its human staff and replaced them with a chatbot, which was stupid both in the labor parts of it (they had just unionized), but also because LLMs aren't actually people who understand context and know how to respond appropriately, and what they've been trained on is not a corpus that I would trust to be able to handle sensitive situations and information. Lo, and behold, the chatbot failed utterly at providing correct contextual information to people with disorders, instead affirming choices that would have made them worse and was taken offline to be fixed. Not that this would have been all that effective to start with, anyway, because eating disorders come in all shapes and sizes, and how you effectively treat them requires understanding the person themselves. (At least the American Medical Association has finally decided the Body Mass Index measurement is not actually any good and is distancing themselves from it.)
If you would like a little bit of understanding about how LLMs and chatbots can be so easily and terribly manipulated, Gandalf is a chatbot with a password that you need to manipulate into giving you the password. Even though Gandalf has been told not to give you the password, it can be convinced to do so with some lateral thinking or some indirect instructions. Lakera, the company promoting their own LLM protection models and software, put out a post with spoilers of the passwords, but also the protection methods used on each level of the Gandalf game, with the exception of level 8, Gandalf the White, which is the actual model with all of the protections they've put in place and are then trying to get people to still break it so they can make it better.
The Organization for Transformative Works gave an update on their progress toward making the Archive of Our Own a less racist space. The comments ask about why the process is so slow and also would like to know about what the organization has done and is doing to protect volunteers from further attacks, some of which included sending child sexual exploitation material to the volunteers in retaliation for moderation decisions. As further information has come to light,![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
synecdochic criticized the OTW's handling of Trust and Safety issues and ignnorance of best practices in relation to them and then a follow-up explaining that the criticism of the OTW for its inability to protect volunteers and fans of colors from harassment is still true, and also the criticism of the EndOTWRacism campaign stemmed from a disagreement about what tactics are appropriate and the campaign's over-reliance on someone who has done a fair amount of policing and harassing other fans of color in addition to receiving it. And also that the OTW was remarkably quick in putting out a piece criticizing synecdochic compared to the glacial pace they have been taking on the more serious allegations leveled against them.
( And more, below the fold! )
Last for tonight, someone watched the Dungeons and Dragons movie with the audio description turned on accidentally and thought it was a better experience than without.
And a poem about all the things that you don't see when you look at someone. Like how well they keep bees.
(Materials viaadrian_turtle, azurelunatic, boxofdelights, cmcmck, conuly, cosmolinguist, elf, finch, firecat, jadelennox, jenett, jjhunter, kaberett, lilysea, oursin, rydra_wong, snowynight, sonia, the_future_modernes, thewayne, umadoshi, vass, the ![[community profile]](https://www.dreamwidth.org/img/silk/identity/community.png)
meta_warehouse community, and anyone else I've neglected to mention or who I suspect would rather not be on the list. If you want to know where I get the neat stuff, my reading list has most of it.)
Also, a
dw_news post welcoming people moving to Dreamwidth from Reddit, since Reddit decided that it wanted to charge ludicrous amounts of money for API access, and doing so essentially kills third-party tools and other ways of accessing Reddit (including several methods intended to make the site more accessible.)A reminder that financial entities like Venmo and PayPal are not incorporated as banks or credit unions and do not have deposit insurance in case of financial difficulties. They're convenient, yes, but they don't have the full protections you would need if the companies collapsed. Some of their items do, because they're actually held and transacted with banks, but not all of them, and not necessarily by default. Treat them with the same amount of trust you would treat cryptocurrency.
A helpline for people with eating disorders fired all of its human staff and replaced them with a chatbot, which was stupid both in the labor parts of it (they had just unionized), but also because LLMs aren't actually people who understand context and know how to respond appropriately, and what they've been trained on is not a corpus that I would trust to be able to handle sensitive situations and information. Lo, and behold, the chatbot failed utterly at providing correct contextual information to people with disorders, instead affirming choices that would have made them worse and was taken offline to be fixed. Not that this would have been all that effective to start with, anyway, because eating disorders come in all shapes and sizes, and how you effectively treat them requires understanding the person themselves. (At least the American Medical Association has finally decided the Body Mass Index measurement is not actually any good and is distancing themselves from it.)
If you would like a little bit of understanding about how LLMs and chatbots can be so easily and terribly manipulated, Gandalf is a chatbot with a password that you need to manipulate into giving you the password. Even though Gandalf has been told not to give you the password, it can be convinced to do so with some lateral thinking or some indirect instructions. Lakera, the company promoting their own LLM protection models and software, put out a post with spoilers of the passwords, but also the protection methods used on each level of the Gandalf game, with the exception of level 8, Gandalf the White, which is the actual model with all of the protections they've put in place and are then trying to get people to still break it so they can make it better.
The Organization for Transformative Works gave an update on their progress toward making the Archive of Our Own a less racist space. The comments ask about why the process is so slow and also would like to know about what the organization has done and is doing to protect volunteers from further attacks, some of which included sending child sexual exploitation material to the volunteers in retaliation for moderation decisions. As further information has come to light,
synecdochic criticized the OTW's handling of Trust and Safety issues and ignnorance of best practices in relation to them and then a follow-up explaining that the criticism of the OTW for its inability to protect volunteers and fans of colors from harassment is still true, and also the criticism of the EndOTWRacism campaign stemmed from a disagreement about what tactics are appropriate and the campaign's over-reliance on someone who has done a fair amount of policing and harassing other fans of color in addition to receiving it. And also that the OTW was remarkably quick in putting out a piece criticizing synecdochic compared to the glacial pace they have been taking on the more serious allegations leveled against them.( And more, below the fold! )
Last for tonight, someone watched the Dungeons and Dragons movie with the audio description turned on accidentally and thought it was a better experience than without.
And a poem about all the things that you don't see when you look at someone. Like how well they keep bees.
(Materials via
adrian_turtle, azurelunatic, boxofdelights, cmcmck, conuly, cosmolinguist, elf, finch, firecat, jadelennox, jenett, jjhunter, kaberett, lilysea, oursin, rydra_wong, snowynight, sonia, the_future_modernes, thewayne, umadoshi, vass, the meta_warehouse community, and anyone else I've neglected to mention or who I suspect would rather not be on the list. If you want to know where I get the neat stuff, my reading list has most of it.)
