How we spent our first two weeks as an independent AI safety research group

See this post on LessWrong here.