cheering on ugly output only brings ugly allies ? Adversarial prompts and model testing, and deeper understanding of the mechanisms, are productive ways forwar

cheering on ugly output only brings ugly allies ? Adversarial prompts and model testing, and deeper understanding of the mechanisms, are productive ways forward IMHO