Question

A retail company builds a generative AI assistant to recommend job-training courses to employees. After launch, staff notice the tool consistently suggests leadership programs to men and administrative-skills courses to women, even when their profiles are similar. The leadership team asks the most likely root cause of this behavior. What should the AI leader explain?

Accepted Answer

The model learned and reproduced historical patterns of inequality present in its training data

Answer

The model's context window was too small to process the full employee profiles

Answer

The model hallucinated course names that do not exist in the catalog

Answer

The prompts sent to the model contained too many tokens, causing truncation