Remote AI alignment writing group seeking new members

by rmoehn 1 min read18th Jan 2020No comments

12


If you're interested, then email me (<given name of President Nixon>.moehn@posteo.de) or PM me on LessWrong.

We meet weekly for 15 min ⨉ number of members. At the moment the meetings consist of explaining aspects of our research to each other. This helps us get better at explaining (sometimes off the cuff), as well following technical explanations and coming up with questions about them. Also, questions from others are a good way to find holes in one’s understanding and patch them. – Better find them among friends and patch them early than getting flustered in public later.

The research topics of the two current members are:

  • impact measures/instrumental convergence à la Alex Turner
  • iterated distillation and amplification à la Paul Christiano

Even if your topic is far afield, we'll try and make it work. Explanations and questions are universal.

There used to be four members. This means:

  1. The meetings are useful for some and not for others. If you're unsure, I suggest you join. If you don't get enough value for your time, you can leave again. We respect that.

  2. Two spots are free!

See also the original announcement.