yeah, the future of online discussions has arrived in the #fediverse as well: found the replies of a guy regarding #nuclear energy to be a bit too suspiciously samey sounding, so I checked with an #ai detector. lo and behold!
(is there maybe a browser plugin to detect this stuff automatically?)
@kyonshi remember that these ai detectors are as trustworthy as the things they are checking, meaning not at all. And they can cause harm, like flagging actual people (potentially neurodivergent people more often).
@marta yeah, and the arguments against it also always are the same, and because both sides value different things neither can come to am agreement
(Anyway, in this case I noticed that of his ten comments 8 were shown to be AI generated and 2 were human created. The AI ones were the same length, had proper formatting, and the usual AI abundance of filler words, while the other two were shorter, did have non-standard formatting, and sounded more natural)
Add comment