Charlie Steiner comments on Investigating Bias Representations in LLMs via Activation Steering