Google

A prevalent form of bias in LLMs is gender bias, where models exhibit a preference for one gender. This often stems from class imbalances in the training corpus, such as the term 'nurse' being more frequently associated with women, which the model learns and reproduces.

Gender Bias in LLMs from Data Imbalance

Based on the provided case study, analyze the most likely reason for the language model's biased outputs. Explain how the characteristics of the training data led to this specific behavior.

AI Recruitment Tool Anomaly

A development team building a large language model identifies that their training data frequently associates the profession 'engineer' with male pronouns and 'teacher' with female pronouns. To correct this, they propose a data augmentation strategy: they will duplicate a portion of the training dataset and, in the duplicated copy, swap all instances of 'he'/'him'/'his' with 'she'/'her'/'hers' and vice-versa. Evaluate the effectiveness of this strategy in mitigating gender bias. In your evaluation, consider both the potential benefits and the potential unintended negative consequences of this approach.

Evaluating a Data Augmentation Strategy for Bias Mitigation

A company develops a large language model to assist with writing professional biographies. They notice that when prompted with the job title 'Surgeon', the model generates biographies using male pronouns and associates the character with stereotypically masculine traits. Conversely, when prompted with 'Administrative Assistant', it consistently uses female pronouns and stereotypically feminine traits. What is the most direct cause of this observed behavior?

Learn Before

Related