QuestionQ64
VerifiedFundamentals of gen AIA 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 However, the training dataset disproportionately contains feedback from a particular demographic and relies on outdated language norms that do not reflect current customer communication styles.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?
Community Discussion