A collection of PIBBSS write-ups about ideas from physics—namely renormalization—and its potential to make gains in AI safety and interpretability.