The negative consequences of poorly defined objectives in AI can already be seen in social media and relatively weak machine learning algorithms. The AI safety community's main objective is to properly define objectives to avoid catastrophic consequences.