About the RoleWe are hiring a senior applied scientist to study how AI affects people and society using OpenAI production data, experiments, and social science methods. This role sits at the intersection of applied social science, AI, and product policy. The scientist will own research from an ambiguous question through study design, data collection and analysis, methodological validation, and recommendations. Their work will inform product and policy decisions and, where appropriate, contribute to external publications. We are looking for a hands-on empirical social scientist with exceptional methodological judgment.
In this role, you will:- Develop and execute a research agenda on AI's social impact, including effects on well-being, learning, social connection, trust, and human decision-making, in partnership with product and policy teams.
- Own empirical studies end to end: define the question and estimand, choose the design, build the dataset, conduct the analysis, and explain what the evidence supports.
- Analyze large-scale production and behavioral data, linking them to experiments, surveys, or external outcomes when appropriate to measure real-world effects.
- Design randomized experiments and credible observational studies; test identification assumptions, quantify uncertainty, and distinguish association from causal effects.
- Develop and validate measures of social outcomes, including human and LLM-assisted annotations; assess measurement error, subgroup performance, and drift.
- Translate findings into concrete product and policy recommendations, with clear limits on causal interpretation, generalizability, and decision confidence.
- Partner with Product Policy, Research, Product, Data Science, Engineering, UXR, Legal, and domain experts; contribute rigorous work to external publications when appropriate.
You might thrive in this role if you have:- A PhD or equivalent research experience in sociology, political science, psychology, communication, computational social science, or a related empirical field, with demonstrated independent research ownership.
- Deep expertise in research design, causal inference, experimentation, statistical analysis, and measurement, demonstrated through work you can personally explain and defend.
- Strong SQL and Python or R skills, with experience independently constructing and analyzing complex behavioral datasets and producing reproducible research.
- A record of research that informs consequential decisions about technology and people, with the judgment to identify both useful findings and unresolved uncertainty.
- Clear writing for technical and policy audiences, and experience producing rigorous research papers or equivalent public technical work.
- The ability to move quickly, collaborate across functions, and maintain scientific independence when findings are sensitive or inconvenient.
