QuestionQ64

VerifiedFundamentals of gen AI

A company trains a generative AI model to classify customer feedback as positive, negative, or neutral. However, the training dataset disproportionately contains feedback from a particular demographic and relies on outdated language norms that do not reflect current customer communication styles. After deployment, the model demonstrates strong bias in sentiment analysis of new customer feedback, misclassifying reviews from underrepresented demographics and struggling to understand current slang or phrasing. What type of model limitation does this represent?

Explanation

Data dependency occurs when a model’s performance and fairness are constrained by the quality, representativeness, and currency of its training data. Demographic imbalance and outdated language in the dataset lead to biased sentiment classifications and weak understanding of newer phrasing.

Community Discussion

No comments yet. Be the first to start the discussion!