AI Interpretability, Controllability, and Safety Research (Martin Wattenberg, Fernanda Viégas) | Grantmaking.ai