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