555
AI summaries of people have landed on Reddit.
(infosec.pub)
Stumbled across this screenshot. Now, instead of judging people for yourself, the computer can do it for you!!! This will in no way be used negatively by Reddit nor people who are arguing with each other. Original post is here, if anyone still has Reddit maybe go spread the good word of Lemmy and Piefed to these users :P
I'm curious how stuff like this works. Surely, many people's post histories can't fit into context. So, maybe the LLM keeps some sort of "blackboard" of summaries of posts, and edits it as it goes along? Would be pretty computationally expensive. I suppose another way would be to create embeddings of each post, and do some sort of clustering or something.