
"(My understanding of) What Everyone in Technical Alignment is Doing and Why" by Thomas Larsen & Eli Lifland
LessWrong (Curated & Popular)
00:00
Open a I - The Corps Difficulties of Alignment
The safety team at open a is plan to build an mvp aligned a g i that can help us solve the full alignment problem. They are working on experimenting with this approach by trying to get current da a is to do useful supporting work, such as summarizing books and criticizing itself. The misallined a g i has to help us discover a successor alligned a g i. And this only works when the a g i doesn't recursively self improve to super intelligence. This alignment plan seems complicated and therefore vulnerable to the godzilla problem. I also think it relies on very slow take off speeds.
Play episode from 01:05:00
Transcript


