The primary risk in using generative AI for high-stakes decisions like hiring is its potential to perpetuate and amplify societal biases present in the data it was trained on or the résumés it processes. This can lead to unfair or discriminatory outcomes. Furthermore, LLMs can "hallucinate" or fabricate information, leading to inaccurate summaries that could unjustly influence a hiring decision. Addressing this risk is a core component of Microsoft's "Fairness" principle for Responsible AI.

A: Failing to read file formats is a technical data ingestion issue, not a fundamental risk related to the fairness or accuracy of the AI's decision-influencing output.

C: Requiring additional GPU capacity is an operational and cost consideration (scalability), not a risk related to the system's fairness or its impact on applicants.

D: This statement is generally false. A key mitigation strategy for AI risks, especially in hiring, is to implement a "human-in-the-loop" system where a human reviews and validates the AI-generated content before making a final decision.

1. Microsoft, "Responsible AI Standard v2, Fairness Principle": "AI systems should treat all people fairly... Fairness harms can include... allocation harms, where an AI system withholds opportunities or resources from certain groups, and quality of service harms, where an AI system does not work as well for one group of people as it does for another." This directly applies to hiring scenarios. (Source: Microsoft Official Documentation, Responsible AI Standard).

2. Microsoft Learn, "Identify and mitigate fairness issues": "Fairness is a core principle of Responsible AI... AI models can reflect or even amplify existing societal biases found in their training data. It's critical to assess your models for fairness and mitigate any issues before deploying them." (Source: Microsoft Official Documentation).

3. Microsoft Learn, "Identify and mitigate risks with generative AI": Discusses the risk of "fabrication (or hallucination)" where the model "generates content that is nonsensical or not factually accurate." In a résumé summary, this could lead to incorrect hiring decisions. (Source: Microsoft Official Documentation).