[ASoT] Reflectivity in Narrow AI — AI Alignment Forum