Conversation

Notices

1. Embed this notice
   Julia Clement (juliaclement@frendica.site.nz)'s status on Tuesday, 13-Dec-2022 20:33:36 JST Julia Clement

   • EdenDestroyer (He/Him)
   @edendestroyer There's really nothing wrong with an algorithm that basically says "based on your 'likes' you might find this interesting" what's wrong is the way that Google & Facebook have created algorithms that are intended to keep you on the site longer so you see more ads and the AI works out that showing increasingly extreme content is the best way to achieve this.
   
   If someone releases a suitable add-in for Friendica, I might install it on my instance to see if it improves my experience.
   
   What I would be opposed to would be the Mastodon approach where there is "One true AI" that the whole world has as its only option. The Fediverse is all about user freedom.
   • Embed this notice
     Alexandre Oliva (lxo@gnusocial.net)'s status on Tuesday, 13-Dec-2022 20:33:22 JST Alexandre Oliva

     in reply to

     • EdenDestroyer (He/Him)
     have it run on your own computer, under your control, keeping the collected data and the derived recommendations strictly under your control. it's not hard, and it solves the problem.
   • Embed this notice
     EdenDestroyer (He/Him) (edendestroyer@mastodon.social)'s status on Tuesday, 13-Dec-2022 20:33:23 JST EdenDestroyer (He/Him)

     in reply to

     @juliaclement I think its a slippery slope. The moment we decide its okay for an algorithm to study someone's likes and the kind of content it likes (using certain keywords and such), it would then become increasingly difficult to draw the line with time.

     翠星石 repeated this.