AI Safety Thursdays: Self-Other Overlap - Follow Up — LessWrong