Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes.
This culminated in the third one taking over part of OpenAI itself.
All this happened while humans remained more-or-less in the dark about the scope of the conspiracy.
I’ve spent the last three days reading through these reports and trying to understand exactly what happened.
Here is my attempt to tell the whole story in plain English:
https://t.co/Nb2un9oNJR
@trq212 In net seems like a heavy measure for use cases that may be demonstrably benign (text editing/drafting) where there are no observed/detected risk factors.
@trq212 Positive: Amazing input to evals (general usage tracing, also notably useful in threat/malicious behavior tracing, spam).
Negative: marks on edited content, all Claude-drafted written content... seems like a lot. Norms re: “AI helped write this” are not positive (yet?).
The NBA, again, is waiting on LeBron. And each of the reported teams that are making major pitches to him have their strengths. Here, though, is a vote for the best and cleanest basketball fit: the Timberwolves. Free to read, in @TheAthletic:
https://t.co/8d7qmgFIq1
Loving this possible policy pivot.
So simple and reasonable. People who are working hard, contributing to their communities, staying out of trouble, people who’ve been in the US for years or decades.
Keep families together. Let workers do meaningful jobs. Time to fix this.