A recent paper argues that long-term AI safety research requires the involvement of social scientists. The authors claim that current alignment algorithms may not account for human psychology, rationality, emotion, and biases, which are essential to properly aligning advanced AI systems with human values. This collaboration is necessary to resolve uncertainties related to human behavior and ensure AI systems function as intended in real-world scenarios. --- Why it matters: This matters because AI safety relies on understanding how humans interact with and perceive AI systems, and social scientists can provide critical insights into human psychology and behavior. By incorporating social science research into AI development, engineers can create more effective and responsible AI systems that align with human values. Source: https://openai.com/index/ai-safety-needs-social-scientists