This research continues many of the directions our team worked on prior to Anthropic, including: GPT-3, Circuit-Based Interpretability, Multimodal Neurons, Scaling Laws, AI & Compute, Concrete Problems in AI Safety, and Learning from Human Preferences. Partner cross-functionally to drive analysis, build financial models, and recommend business decisions across a variety of areas, such as renewals and expansion forecasting, customer retention economics, cost-to-serve, and headcount and capacity planning.