Accepting Copilot suggested identifiers and comments without any human review is the correct choice because it shows that the team is removing the essential human oversight needed to catch biased or noninclusive language and unrepresentative examples in generated code and documentation.

By accepting suggestions without review the team can unknowingly propagate harmful stereotypes, exclusionary terms, and narrow examples. Responsible use of generative tools relies on human-in-the-loop review, clear coding and documentation standards, and checks for inclusive language to identify and correct biased outputs before they are merged.

Conducting regular reviews of Copilot suggestions for inclusive language and representative examples is a positive mitigation that helps find and fix biased wording and examples, so it does not indicate a failure to address bias.

Scanning pull requests with Cloud DLP to detect sensitive data before merge targets data leakage and privacy risks rather than the presence of biased language or examples, so it does not demonstrate neglect of bias concerns.

Choosing not to use Copilot for drafting logic that involves confidential business rules reduces confidentiality and intellectual property risk, and it is not about bias mitigation, so it is not evidence of failing to address bias.

Full AWS Practitioner Certification Question