Conversation

Notices

1. Embed this notice
   Alexandre Oliva (lxo@gnusocial.net)'s status on Wednesday, 18-Jan-2023 01:05:14 JST Alexandre Oliva

   in reply to

   • teledyn 𓂀
   people are worried about the effects of the feedback loop that the popularity of chatgpt-generated posts will have on itself and other AI training sets.
   truth is, there have been robots among us, spewing hate speech and disinformation for several years already.
   what kind of captcha will surveillance capitalists start requiring to train their AIs on human outputs rather than AI-generated output?
   what data brokering markets are going to appear underground for "100% human training sets"?
   • Embed this notice
     teledyn 𓂀 (teledyn@mstdn.ca)'s status on Wednesday, 18-Jan-2023 01:05:15 JST teledyn 𓂀

     Does chatGPT update its training set with its own conversations? If so, shouldn't we be really careful about tormenting and publicly embarrassing it lest it remember?