Here's a case where the Google AI ethics researchers should have unleashed their hounds, not on Yann LeCun and large language models (stochastic parrots? lol). Design evaluation datasets and checklists, do research into "bias washing", call them out on biased models in deployment, etc.