Our goal is to contribute to training Claude to invent, design, and engineer across every discipline and scale, reliably and safely: new manufacturing and experimental techniques, processes, materials, devices, instruments, machines, energy systems, and factories. 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.