Why might a voice assistant sound like a woman? A gender stereotype is a generalized association between a gender and a trait, role, or occupation. Alexa, Siri, Cortana, and Google Assistant all launched with female voices, prompting criticism that assistants could reinforce the idea that women are there to serve others. section 9.5
These associations can shape how social chatbots are presented and perceived. In one study, male chatbots scored higher on competence than on trust or helpfulness, though the researchers acknowledged limitations. This raises a risk, not a certainty about every user: chatbot relationships may entrench existing stereotypes rather than challenge them. section 9.5
Stereotypes can also enter language technology through training data. Word embeddings—representations that place related words near one another—trained on news articles produced analogies linking women with nursing and men with medicine or surgery. Unequal representation of genders in the articles helps explain those patterns. Large language models can likewise echo or amplify biases in their training data; removing deeply ingrained associations remains difficult. section 1.12
Gender representation therefore matters both when choosing how a chatbot appears to users and when assessing what a model has learned. Neither a friendly interface nor an attempt to reduce model bias guarantees that familiar social assumptions have disappeared. section 9.5 section 1.12
