AI safety needs social scientists
OpenAI: AI safety research is missing a critical piece. Social scientists.
Why it matters
OpenAI argues that technical AI alignment work is incomplete without social science expertise on human psychology, bias, and value alignment—and is hiring social scientists to prove it.
The key facts
10 to knowOpenAI publishing research paper on integrating social scientists into AI safety work
Focus on human psychology, rationality, emotion, and bias as critical to alignment
OpenAI committing to full-time social scientist hires
Interdisciplinary collaboration between ML and social science researchers
Published Feb 2019 — early signal on AI safety governance thinking
OpenAI publishing research on integrating social science into AI safety
Paper argues AI alignment algorithms must account for human psychology and biases
OpenAI planning to hire social scientists full-time for this research
Focus on human-AI value alignment as safety challenge
Published February 2019 — foundational positioning on safety governance approach
Go to the source
OpenAI Blogopenai.com
Publisher excerpt: We’ve written a paper arguing that long-term AI safety research needs social scientists to ensure AI alignment algorithms succeed when actual humans are involved. Properly aligning advanced AI systems with human values requires resolving many uncertainties related to the psychology of human…