Question: In a concise analytical response, contrast product or application development with academic reporting and completely unbiased test-set accuracy measurement.

Sample answer: Label-correction bias can be acceptable when the only objective is to develop a product or application. It becomes problematic when the result will be used in an academic research paper or when a completely unbiased measure of test-set accuracy is required. Thus, the judgment depends on whether the intended use can tolerate the bias or explicitly demands an unbiased result.

Key points:

• Acceptable for product development only
• Acceptable for application development only
• Problematic for an academic research paper
• Problematic when completely unbiased test-set accuracy is needed
• Intended use determines the judgment

Rubric: A strong response accurately states the acceptable development-only context, identifies both problematic contexts, and explains that intended use determines acceptability.