AIF-C01 Practice Question: A social media company deploys a content…

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  F1 score

  Why it's wrong here

  F1 score is the harmonic mean of precision and recall, so it penalises false positives and false negatives equally rather than prioritising precision alone. It is tempting because F1 balances both error types on imbalanced data, and it would be correct if the company cared equally about over-censoring and missed harmful content.

• ✗

  Recall

  Why it's wrong here

  Recall measures the proportion of actual harmful content detected, so maximising it reduces false negatives and permits more false positives — the opposite of this goal. It is tempting because recall is the standard priority when missing positives is costly, such as medical screening or fraud detection where every harmful case must be caught.

• ✗

  Accuracy

  Why it's wrong here

  Accuracy aggregates all correct predictions, so with imbalanced moderation data a model can score highly while still misclassifying many legitimate posts. It is tempting because accuracy is the default headline metric, and it suits balanced datasets where false positives and false negatives carry equal cost.

• ✓

  Precision

  Why this is correct

  Precision measures the proportion of predicted harmful posts that are genuinely harmful, so maximising it directly reduces false positives, satisfying the stem's goal of avoiding over-censoring legitimate content. Recall would instead prioritise catching all harmful posts.

Go deeper

Related to this question