How does Mastodon prevent itself from becoming email? I'm unfamiliar with the design. Do you simply turtle your trust with other servers?
One of the interesting things I haven't seen with advanced AI is the antispam AI. I would love to have a semi-trustable service that reviewed incoming numbers/messages and evaluated them against expert level manipulation, propaganda, advertising, and con techniques. Of course that then provides access to the agent ...
You'd THINK facebook and google have advanced AI language models for these given ChatGPT's abilities, but I've never seen or heard anything like that. Then again, that would be customer focused, and google and facebook are now more spam/con than pro-consumer utility.
Ha, if ChatGPT is REALLY converging to the point of coding actual systems (1-2 years maybe???), then I would like to weaponize it to counter the api drift defense used by the social media walled gardens.
No one has been able to truly integrate your social media content across the walls. Because no integration stands the constant drift of api interface. BUT, if chatgpt can constantly adapt to the changes, you might be able to do it.
Consider that while the walled gardens can shift their APIs to some degree, they can't shift them REALLY fast, because their "actual" partners need some stability. So I might be able to AI adapt-scrape from my sources and centralize better.
I don't think that is within ChatGPT's current abilities, but I don't think it is that far.
"Turtling" : a defensive, cautious approach to expansion in RTS and 4X games, rather than reckless expansion and "mixing it up". In this context, you carefully whitelist nodes conservatively, but you'll miss a lot of content and people.