Unstructured data without standardized preprocessing (option A) creates the highest data quality risk because AI models depend heavily on the cleanliness, consistency, and structure of input data.
AAIA warns that improperly processed unstructured data leads to:
* Incorrect text extraction
* Lost contextual meaning
* Feature extraction errors
* Misclassification
* Inaccurate audit evidence
Option B is a bias or relevance risk, not data quality.
Option C is a governance/training issue.
Option D is an oversight risk, not a data quality issue.
Therefore, using unstructured data without preprocessing is the most direct threat to data quality.
References:
AAIA Domain 2: Data Preprocessing and Quality
AAIA Domain 3: AI-Assisted Audit Evidence Integrity