Economics of AI conference from NBER — LessWrong