Ensure the AI system’s language and behaviors do not reinforce undesirable and unfair stereotypes and biases.
Maintain short-term memory and allow the user to make efficient references to that memory.
Enable the user to provide feedback indicating their preferences during regular interaction with the AI system.